Flower predation by

Flower predators (florivores) may affect plant reproduction directly through loss of pollen and ovules, or indirectly by deterring pollinators which avoid damaged flowers. Caterpillars of the widespread endemic moth Zelleria maculata feed inside flower buds of the endemic mistletoes Peraxilla tetrapetala and P. colensoi in New Zealand. We measured flower predation rates between 1995 and 2007 at 24 sites throughout New Zealand and assessed Zelleria feeding impact on fruit set. Zelleria predation showed a strong latitudinal gradient, being rare in the North Island but affecting 81% of flowers in 1996 at Waipori (Otago) and averaging > 38% at Waipori and Eglinton (Fiordland). The reasons for this pattern are unknown, but a wasp parasitoid of Zelleria, Campoplex sp. (Ichneumonidae), may be less common further south. Rates of Zelleria attack were higher in mistletoes growing several metres above ground and in less fragmented habitats (i.e. in shaded positions, and away from edges), and higher in P. tetrapetala than P. colensoi. An experiment following 1005 tagged P. tetrapetala flowers at two sites showed that Zelleria attack significantly reduces both pollinator-flower-opening rates (required for effective pollination) and fruit set rates. Zelleria attack increased the chance of a flower not being opened by pollinators from 11 to 37% at Ohau in 1996 (reducing fruit set from 28% to 7.9%) and flower non-opening from 6 to 30% at Craigieburn in 1995 (fruit set: 44.4% to 16.6%), representing reductions in fruit set of 72% and 63% at the two sites. Therefore, Zelleria reduces reproduction in Peraxilla spp. through both pollinator deterrence and direct loss of flowers at many sites. This could affect conservation of these mistletoes, which are seed-limited and in decline. However, Zelleria impact is reduced by its avoidance of edge habitat, as P. tetrapetala is more abundant and flowers more heavily on edges.     

Differential effects of anthropogenic edges and gaps on the reproduction of a forest‐dwelling plant: The role of plant reproductive effort and nectar robbing by bumblebees

Forest fragmentation produces sharp changes in the spatial configuration of remnant forest patches, which include the increasing influence of patch edges on the interior of forest patches (‘edge effects’). Human activities responsible for forest fragmentation tend also to change the internal characteristics of remnant patches, for example, through the creation of gaps by selective logging. While edges and gaps can be expected to cause comparable changes in the micro‐environmental conditions of the forest, their effects on forest‐dwelling species and their interactions are not necessarily comparable. This study compares the effect of forest edges and anthropogenic gaps on the reproductive success of a self‐incompatible epiphytic plant (Mitraria coccinea), mediated by changes in its relationship with mutualists (pollinators) and antagonists (flower larcenists). Mitraria coccinea's flowers are pollinated by hummingbirds and robbed by the bumblebee Bombus dahlbomii. Edges and gaps had comparable positive effects on flower production and fruit set; however, nectar robbing was up to sevenfold higher in patches with numerous gaps and resulted in lower reproductive success (fruit set and total fruit crop). Forest fragmentation studies should therefore avoid treating forest remnants as homogeneous units (i.e. focusing exclusively on their characteristics and connectivity), without taking into account the internal heterogeneity caused by anthropogenic pressures (e.g. gaps and edge effects).     

Alien pollinator promotes invasive mutualism in an insular pollination system

The alien predatory lizard, Anolis carolinensis, has reduced the insect fauna on the two main islands of the Ogasawara archipelago in Japan. As a result of this disturbance, introduced honeybees are now the dominant visitors to flowers instead of endemic bees on these islands. On the other hand, satellite islands not invaded by alien anoles have retained the native flower visitors. The effects of pollinator change on plant reproduction were surveyed on these contrasting island groups. The total visitation rates and the number of interacting visitor groups on main islands were 63% and 30% lower than that on satellite islands, respectively. On the main islands, the honeybees preferred to visit alien flowers, whereas the dominant endemic bees on satellite islands tended to visit native flowers more frequently than alien flowers. These results suggest that alien anoles destroy the endemic pollination system and caused shift to alien mutualism. On the main islands, the natural fruit set of alien plants was significantly higher than that of native plants. In addition, the natural fruit set was positively correlated with the visitation rate of honeybees. Pollen limitation was observed in 53.3% of endemic species but only 16.7% of alien species. These data suggest that reproduction of alien plants was facilitated by the floral preference of introduced honeybees.     

Leaf damage by herbivores affects attractiveness to pollinators in wild radish, Raphanus raphanistrum

 We carried out two experiments to determine the effect of leaf damage on plant attractiveness to pollinators using wild radish, Raphanus raphanistrum (Brassicaceae), a self-incompatible annual herb. Pairs of plants from 36 full-sib families were grown in pots in the greenhouse. One member of each pair was damaged by Pieris rapae larvae that were allowed to remove half of the leaf area of each of the first four rosette leaves. The plants were subsequently taken out for pollinator observations once a week from the beginning of flowering in late June until the end of August. We conducted two experiments to examine how foliar damage affected visitation by pollinators. In the first experiment, numbers of pollinators visiting plants were compared between damaged and control sibling plants. In the second experiment, the number of open flowers during observations was controlled to be the same for both damaged and undamaged sibs. Damage significantly decreased the number and size of flowers during the first observations in late June. Damaged plants received fewer visits by native bees during the first week of observations. Since damage did not affect native bee visits when the number of open flowers was equalized between treatments, flower number was probably the main cue attracting native bees to plants. In the experiment without flower number control, syrphid flies, the other abundant pollinator taxon, spent more time per flower on the undamaged than on the damaged plants. When flower number was controlled, flies probed significantly more flowers during each visit on the undamaged than on the damaged plants and had higher visitation rates to undamaged plants early in the season. Since syrphid flies preferred undamaged plants both with and without flower number control, they apparently used cues apart from flower number for visitation. The difference between undamaged and damaged plants in floral characteristics and pollinator visitation vanished within a few weeks after the start of flowering. This result suggests that early damage may not have a strong fitness effect through reduction in mating success. However, poor weather conditions can cause early mortality of plants in the field, and nutrient depletion and competition decrease fruit set of later flowers. Therefore, conditions exist under which visitation to early flowers may affect plant fitness. Received: 30 July 1996 / Accepted: 10 February 1997     

Effectiveness of short‐tongued bees as pollinators of apparently ornithophilous New Zealand mistletoes

Abstract The flowers of two species of threatened New Zealand mistletoes (Peraxilla tetrapetala and Peraxilla colensoi, Loranthaceae) have explosive buds that do not open unless force is applied by birds or two species of native short‐tongued bees. Opened flowers are visited by a variety of birds and insects. Although both species of Peraxilla conform to a pollination syndrome of ornithophily, bees may be effective alternative pollinators. We investigated the effectiveness of bees and birds as pollinators of P. colensoi at one site and P. tetrapetala at two sites in the South Island. Bees and other insects outnumbered birds as flower visitors at all three sites. By excluding birds with wire cages, we showed that two bee species regularly open flowers of P. tetrapetala, but only rarely open flowers of P. colensoi. Few pollen grains were deposited when either birds or bees opened buds, so opening buds was not by itself sufficient for adequate pollination. Instead, pollen continued to accumulate over the next 6 or 7 days, even inside cages that excluded birds. Both populations of P. tetrapetala were regularly pollen‐limited, but in different ways. At Ohau, opened flowers gained enough pollen to produce seeds, but many buds were not opened and hence failed to set seed. In contrast, at Craigieburn, nearly all buds were opened, but many of these did not receive enough pollen. These results demonstrate that native bees can partially replace birds as pollinators of mistletoes, despite their apparent ornithophilous syndrome. Ongoing reductions in New Zealand forest bird numbers means that the service bees provide may be important for the long‐term future of these plants.     

Orchard pollination in Capitol Reef National Park,Utah, USA. Honey bees or native bees?

Capitol Reef National Park in central Utah, USA surrounds 22 managed fruit orchards started over a century ago by Mormon pioneers. Honey bees are imported for pollination, although the area in which the Park is embedded has over 700 species of native bees, many of which are potential orchard pollinators. We studied the visitation of native bees to apple, pear, apricot, and sweet cherry over 2 years. Thirty species of bees visited the flowers but, except for pear flowers, most were uncommon compared to honey bees. Evidence that honey bees prevented native bees from foraging on orchard crop flowers was equivocal: generally, honey bee and native bee visitation rates to the flowers were not negatively correlated, nor were native bee visitation rates positively correlated with distance of orchards from honey bee hives. Conversely, competition was tentatively suggested by much larger numbers of honey bees than natives on the flowers of apples, apricots and cherry; and by the large increase of native bees on pears, where honey bee numbers were low. At least one-third of the native bee species visiting the flowers are potential pollinators, including cavity-nesting species such as Osmia lignaria propinqua, currently managed for small orchard pollination in the US, plus several fossorial species, including one rosaceous flower specialist (Andrena milwaukiensis). We suggest that gradual withdrawal of honey bees from the Park would help conserve native bee populations without decreasing orchard crop productivity, and would serve as a demonstration of the commercial value of native pollinators.     

The relative importance of birds and bees in the pollination of Metrosideros excelsa (Myrtaceae)

Exclusion experiments were used to assess the effect of different pollinator groups on outcrossing and seed production in Metrosideros excelsa. The main study site was Little Barrier Island, New Zealand where indigenous bird and native solitary bees are the main flower visitors. Our results showed that native birds were more important pollinators of M. excelsa than native bees. Seed production was much higher in open pollination than in two exclusion experiments where either birds were excluded and native bees only had access to flowers, or where all pollinators had been excluded. The number of fertile seeds per capsule was 45% higher after open pollination than in treatments with bee visitation only and 28% higher than in treatments where all flower visitors were excluded. Estimated outcrossing rates were significantly higher (t_m = 0.71) for open pollination in the upper canopy (>4 m above‐ground level) where bird visitation is presumed to be more frequent than for a treatment with native bee access only (t_m = 0.40). Our results also suggest that a large proportion of seeds (66%) arise from autonomous self‐pollination when all pollinators are excluded. In four trees of a modified mainland population with predominantly introduced birds and a mixture of introduced and native bees there was no decrease in seed production for the treatment allowing bee access only, indicating that – in contrast to native bees – honeybees may be more efficient pollinators of M. excelsa. Observation of the foraging behaviour of both groups of bees showed that native bees contact the stigma of flowers less frequently than honeybees. This is likely to be a consequence of their smaller body size relative to honeybees.     

盗蜜对角蒿传粉者行为和生殖成功的影响

一些研究显示盗蜜对自交植物的结实和结籽没有显著影响。然而, 对于既有传粉者为其传粉实现异交又能通过自交实现生殖保障的兼性自交植物来说, 盗蜜对其生殖的影响还知之甚少。由于兼性自交植物可以自交, 盗蜜对其总体结实可能不会有显著影响, 但可能会通过影响传粉者行为而影响传粉者介导的结实。为了验证这一假说, 本研究以兼性自交的一年生角蒿(Invarvillea sinensis var. sinensis)为研究材料, 通过野外调查和控制实验, 探讨了盗蜜对传粉者介导的结实(传粉者行为)和总体结实率的影响。结果表明: 角蒿的盗蜜者和主要传粉者相同, 均为密林熊蜂(Bombus patagiatus)。熊蜂盗蜜频率平均为20.24% (范围为0-51.43%)。盗蜜对角蒿总体结实率、每果结籽数和每果种子重量没有显著影响。然而, 被盗蜜花的柱头闭合比率显著高于未被盗蜜花, 说明盗蜜影响传粉者的访花行为和传粉者介导的结实率。另外, 被盗蜜花的高度显著高于未被盗蜜花, 说明盗蜜者倾向于从较大较高的花上盗蜜。这些结果为全面认识盗蜜对植物生殖的影响提供了新的信息。     

Pollinator effectiveness of native and non‐native flower visitors to an apparently generalist Andean shrub,Duranta mandonii (Verbenaceae)

Not all visitors to flowers are pollinators and pollinating taxa can vary greatly in their effectiveness. Using a combination of observations and experiments we compared the effectiveness of introduced honeybees with that of hummingbirds, native bees and moths on both the male and female components of fitness of the Andean shrub Duranta mandonii (Verbenaceae). Our results demonstrated significant variation among flower visitors in rates of visitation, pollen removal ability and contribution to fruit set. This variation was not always correlated; that is, taxa that regularly visited flowers did not remove the most pollen or contribute to fruit set. Despite the taxonomic diversity of visitors, the main natural pollinators of this shrub are large native bees, such as Bombus spp. Introduced honeybees were found to be as effective as native bees at pollinating this species. Duranta mandonii has high apparent generalization, but low realized generalization and can be considered to be a moderate ecological generalist (a number of species of large bees provide pollination services), but a functional specialist (most pollinators belong to a single functional group). The present study has highlighted the importance of measuring efficiency components when documenting plant–pollinator interactions, and has also demonstrated that visitation rates may give little insight into the relative importance of flower visitors.     

Honey bees and wild pollinators differ in their preference for and use of introduced floral resources

Introduced plants may be important foraging resources for honey bees and wild pollinators, but how often and why pollinators visit introduced plants across an entire plant community is not well understood. Understanding the importance of introduced plants for pollinators could help guide management of these plants and conservation of pollinator habitat. We assessed how floral abundance and pollinator preference influence pollinator visitation rate and diversity on 30 introduced versus 24 native plants in central New York. Honey bees visited introduced and native plants at similar rates regardless of floral abundance. In contrast, as floral abundance increased, wild pollinator visitation rate decreased more strongly for introduced plants than native plants. Introduced plants as a group and native plants as a group did not differ in bee diversity or preference, but honey bees and wild pollinators preferred different plant species. As a case study, we then focused on knapweed (Centaurea spp.), an introduced plant that was the most preferred plant by honey bees, and that beekeepers value as a late‐summer foraging resource. We compared the extent to which honey bees versus wild pollinators visited knapweed relative to coflowering plants, and we quantified knapweed pollen and nectar collection by honey bees across 22 New York apiaries. Honey bees visited knapweed more frequently than coflowering plants and at a similar rate as all wild pollinators combined. All apiaries contained knapweed pollen in nectar, 86% of apiaries contained knapweed pollen in bee bread, and knapweed was sometimes a main pollen or nectar source for honey bees in late summer. Our results suggest that because of diverging responses to floral abundance and preferences for different plants, honey bees and wild pollinators differ in their use of introduced plants. Depending on the plant and its abundance, removing an introduced plant may impact honey bees more than wild pollinators.     

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.     

Breeding system and pollination of a narrowly endemic herb of the Lower Florida Keys: impacts of the urban-wildland interface

We examined the breeding system and pollination of Chamaecrista keyensis Pennell (Fabaceae: Caesalpinioideae) and the effects of urban edge and mosquito control on reproduction of this rare endemic herb of the Lower Florida Keys. Controlled hand-pollination treatments were applied to plants in the field. Although C. keyensis flowers are self-compatible, they are not capable of automatic selfing. Inbreeding depression was observed in both seed set and percentage seed germination. Bees of seven genera were observed visiting C. keyensis flowers during the peak flowering season (June to July). Only Xylocopa micans and Melissodes spp. may be effective pollinators for C. keyensis, as they were the only bees that "buzz pollinate" this species, which has poricidal anther dehiscence. Chamaecrista keyensis received substantially more visits by X. micans, but fewer visits from Melissodes spp. in urban-edge vs. forest sites. Aerial mosquito spraying may exacerbate the existing pollinator limitation suffered by C. keyensis by reducing the number of visits by the buzz-pollinating bees. Individuals of C. keyensis at urban edges produced fewer seeds per fruit than did individuals in a pristine forest mainly because of greater insect seed predation.     

Isolation from natural habitat reduces yield and quality of passion fruit

• Loss and fragmentation of native vegetation negatively affect crop pollinators and productivity of several crop species. The yellow passion fruit is an excellent model to investigate this issue, because its main pollinators are sensitive to deforestation. Seasonality also influences fruit set of crop species that are pollinated by bees. Climate determines plant flowering patterns and consequently affects pollinator activity. Little is known on effects of pollinators on crop quality in general, and particularly of many tropical fruits that have multiple fruiting seasons throughout the year, such as passion fruit. Our hypothesis is that the distance to the native vegetation and climate seasonality affect the pollination and fruit production of cultivated, hermaphrodite, self‐incompatible, protandrous species that depend on large‐sized bees to set fruit.
• We observed flowers of the yellow passion fruit in eight areas at different distances (600–4,000 m) from a dry seasonal forest (Caatinga). We compared the frequency of the main floral visitor (pollinators and robbers) visits and fruit set among areas at the beginning and end of the dry and rainy seasons.
• Xylocopa species were pollinators, and Apis mellifera, Trigona spinipes and a hummingbird were robbers. Xylocopa visits diminished with increasing distance to native remnants, converse to what was observed for the robbers at the beginning of the dry season. Fruit set was higher at the beginning of the dry season. The distance to the forest affected quality of fruits in terms of length, diameter and skin thickness, which all decreased with increase in remnant distance.
• Our results show that insect pollination is higher in the dry season, increasing profitability and reducing labour costs for the producer. Considering the influence of distance to native vegetation on pollination, robbery and fruit quality, our study reinforces the necessity for conservation of native vegetation remnants.

     

Invasive ant (<Emphasis Type="Italic">Anoplolepis gracilipes</Emphasis>) disrupts pollination in pumpkin

Yellow crazy ant (Anoplolepis gracilipes (F. Smith); “YCA”) is known for its aggressive predatory ability and ability to exert exploitation competition on both native and other invasive ants via floral nectar. We argue that YCA invasion can exert both interference and exploitation competition on legitimate pollinators. In pumpkin fields (Cucurbita maxima L.) of south India, YCA infested the flowers, particularly the pistillate flowers, for nectar foraging. Pumpkin is a honey bee-mediated cross-pollinated monoecious plant that produces disproportionately very few pistillate flowers. We hypothesize that YCA presence in the flowers can affect the visitation rate and foraging time of honey bees in the flowers, the fruit set in pumpkins, and can exert predatory pressure on the honey bees if the bees linger in ant-colonized flowers. Both YCA and honey bees preferred to forage on the limited pistillate flowers in the plants. After colonizing the flowers, YCA did not retreat for hours, even upon disturbance by competitors, such as honey bees. Both the visitation frequency and the foraging time of honey bees were drastically reduced in ant-colonized flowers, and none of the ant-colonized flowers developed into fruits, suggesting that the YCA exert both an ecological and evolutionary pressure on pumpkin. The ants preyed upon about 17% of the honey bees that lingered in ant-colonized flowers, and the time the bees spent foraging predicted the fate of the bees. Exploitation competition exerted by the YCA on pumpkin may have far-reaching consequences for the pollination and productivity of this cash crop.     

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.     

Contrasting effects of plant invasion on pollination of two native species with similar morphologies

Invasive plants may decrease native plant density and disrupt interactions between native plants and their pollinators. We hypothesized that invasive Solidago canadensis (Asteraceae) competes for pollination services with two confamilial species, Ixeris chinensis and Sonchus arvensis. Breeding-system studies revealed that both native species are self-incompatible. In plots with all three species we found that Solidago received the highest visitation rates. To test the hypothesis of competition for pollination in the context of reduced native density, we established 3 plots for both native species with three Solidago densities (uninvaded, 50 and 75 % invaded) and corresponding decreases in native density. We investigated the effects of varying densities of Solidago on honeybee visitation rates, number of successive visits within individual ramets, pollen-load composition on bees, and seed set. For both native species, increasing Solidago density and decreasing native density resulted in bees carrying higher ratios of Solidago pollen and in bees visiting fewer capitula prior to departing from a plant. However, for other aspects of pollination, the native species responded very differently to Solidago. With increasing Solidago and decreasing native density, Ixeris received fewer honeybee visits and produced fewer seeds, demonstrating competition for pollination, but Sonchus attracted more honeybee visits and showed a non-significant trend toward setting more seeds, suggesting facilitation. These opposing effects occurred despite similarities in the native species’ floral morphology, suggesting that the effects of invasive plants are difficult to predict. In this case the different effects may relate to Sonchus being a taller plant with larger flowers.     

Role of honey bees (Hymenoptera: Apidae) in the pollination biology of a California native plant, Triteleia laxa (Asparagales: Themidaceae)

A central focus of pollination biology is to document the relative effectiveness of different flower visitors as pollinators. Ongoing research seeks to determine the role that introduced honey bees (Apis mellifera L.) play in the pollination of both invasive and native plants. Here we report on the importance of A. mellifera as pollinators of a California native plant, Triteleia laxa Bentham. In observation plots and transect censuses, A. mellifera overwhelmingly dominated the T. laxa flower visitor assemblage. We believe the proximity to agriculture, where A. mellifera density is higher relative to areas far from agriculture, contributes to the discrepancy between A. mellifera abundance at the two sites. Although A. mellifera were inferior flower visitors qualitatively (visited less flowers per minute), they were the most frequent interactors with flowers. Furthermore, the proportion of visits to flowers on the same plant among flower visitor species did not differ, suggesting a general mechanism by which insects forage at T. laxa flowers and that A. mellifera do not cause more deleterious geitonogamy than do native pollinators. Flower visitation rates as a function of floral display size did not differ between A. mellifera and other flower visitors. The difference in the magnitude of flower visitation (largely by A. mellifera) between sites is consistent with a difference in seed set between sites. These results suggest that non-native A. mellifera bees can play an important role in the pollination of native plant species.     

Native pollinators increase fruit set while honeybees decrease the quality of mandarins in family farms

Family farms can benefit from the presence of a diverse set of native pollinators and associated pollination services. In the present study we assessed the effect of flower visitor richness and visitation rate by honeybees and native insects on mandarin production (Citrus reticulata `Criolla´), in ten citrus family farms located in the Dry Chaco region of northwest Argentina. An exclusion experiment was conducted to explore how pollinators influence the fruit set and quality of `Criolla´ mandarin. The influence of features such as local richness and abundance of flowering plants, farm size, and surrounding natural/semi-natural habitats in the diversity of flower visitors was also evaluated. Fruit set in open pollination branches was three times higher than in bagged branches, where flower visitors were excluded. Moreover, the mandarin fruit set increased with a higher native visitation rate, and mandarin quality (fruit weight and size) decreased with a higher honeybee visitation rate. Flower visitor diversity was higher in farmlands with a greater proportion of surrounding natural and semi-natural habitats. Our results demonstrate the negative effects of excessive honeybee visitation on citrus fruit quality and highlight the importance of native pollinators and natural habitat conservation to increase the fruit set and quality of mandarin in family farms.     

Where have all the blue flowers gone: pollinator responses and selection on flower colour in New Zealand Wahlenbergia albomarginata

Although pollinators are thought to select on flower colour, few studies have experimentally decoupled effects of colour from correlated traits on pollinator visitation and pollen transfer. We combined selection analysis and phenotypic manipulations to measure the effect of petal colour on visitation and pollen export at two spatial scales in Wahlenbergia albomarginata. This species is representative of many New Zealand alpine herbs that have secondarily evolved white or pale flowers. The major pollinators, solitary bees, exerted phenotypic selection on flower size but not colour, quantified by bee vision. When presented with manipulated flowers, bees visited flowers painted blue to resemble a congener over white flowers in large, but not small, experimental arrays. Pollen export was higher for blue flowers in large arrays. Pollinator preference does not explain the pale colouration of W. albomarginata, as commonly hypothesized. Absence of bright blue could be driven instead by indirect selection of correlated characters.