Why Petals? Naïve, but Not Experienced Bees, Preferentially Visit Flowers with Larger Visual Signals

The study addresses why many plants have showy flowers and specifically tests whether prominent floral visual signals (e.g., ray petals) primarily function to attract naïve, first-time pollinators rather than experienced visitors. The context is that approximately 85% of flowering plants rely on animal pollination and many have evolved non-green, often UV-patterned petals to contrast with foliage and guide pollinators. Asteraceae, with many species bearing conspicuous ray petals around modest disc florets, provide a model to examine this. The authors hypothesize that showy petals act like a long-distance advertisement to secure initial visits from naïve pollinators, after which foragers use learned information about rewards and location to guide subsequent choices. Demonstrating such a role would clarify the selective pressures shaping floral display evolution and help explain the prevalence of showy petals.

Background work indicates that petals and associated visual cues attract pollinators via high color contrast, often including ultraviolet patterns seen by insects. Prior studies in Asteraceae have shown greater visitation to capitula with ray petals compared to rayless forms, and pollinators display innate preferences for symmetry and larger displays, which can reduce search time due to visual resolution limits. UV nectar guides orient bees at close range and increase visitation. Floral display size and plant density often increase pollinator approaches and visitation rates. However, past work has seldom separated effects on naïve versus experienced visitors, leaving a gap that this study addresses by contrasting first-time (patch-naïve) and experienced bee behavior in manipulated floral displays.

Field experiments were conducted in a private garden in East Sussex, UK (50.880, 0.284), between 10:00–16:00 during July–August 2021 under suitable weather (generally sunny, ≥16°C, light wind). Two exotic Asteraceae cultivars with large, conspicuous ray petals were used to minimize prior local experience: Rudbeckia hirta ('Black Eyed Susan') and Helenium autumnale ('Sahin's Early Flowerer'). Four patches (~4 m² each) were established, each containing 16–20 potted plants (10 L pots, 0.5–1 m tall), closely spaced (<20 cm between inflorescences), averaging 132 ± 33.6 inflorescences. Treatments per capitulum were: (i) zero ray petals (all removed), (ii) four ray petals arranged symmetrically in a cross, and (iii) all petals intact. Symmetrical treatments were used given bees’ innate symmetry preference. Patches were kept balanced across treatments by assigning treatments to newly opening inflorescences and removing old ones. For experienced visitor trials, one patch of each plant with all three treatments was set on 20 July 2021; from 9–11 August, actively foraging visitors were uniquely marked with acrylic paint. From 12–14 August (22 h total observation), marked bees were followed and the sequence of treatment choices across consecutive inflorescences recorded. For naïve visitor trials, additional patches were set (H. autumnale on 19 August 2021; R. hirta on 20 August 2021). Outside data collection, plants were caged with fine mesh to exclude visitors. Observers waited for unmarked arrivals, recorded the sequence of the first ten inflorescences visited, then captured and marked these individuals to prevent re-use. Although true lifetime naiveté could not be guaranteed, individuals were naïve to the experimental patches and treatment context. Inflorescence measurements included counts of ray petals per head and diameters of intact inflorescences and central discs (n=10 per species) to estimate visual display area. Statistical analysis used R (v3.4.3) with beta regression (betareg package) to model the proportion of visits to each treatment as a function of the visit number in the sequence (1–10). Chi-squared tests compared observed counts to equal-expectation across treatments. Bees were analyzed at the genus level (Apis, Bombus). Data visualization was performed with ggplot2.

- Inflorescence morphology: H. autumnale had 14.7 ± 2.06 ray petals; whole inflorescence diameter 64.5 ± 3.92 mm; central disc 21.6 ± 0.70 mm. R. hirta had 13.0 ± 0.67 ray petals; whole diameter 108.5 ± 19.51 mm; central disc 21.6 ± 0.70 mm. Ray petals provided much larger visual display area than discs (approx. 8× in H. autumnale; 24× in R. hirta). - Experienced honey bees: Ten uniquely marked honey bees per plant were followed (avg. 23 consecutive inflorescences). On R. hirta, they visited significantly fewer inflorescences with no petals versus four or all petals (counts: no petals 68, four 99, all 103; χ² = 8.156, df = 2, p = 0.017). On H. autumnale, no significant treatment effect (counts: 58, 65, 60; χ² = 0.426, df = 2, p = 0.808). - Naïve visitors: 42 naïve honey bees (R. hirta: 23; H. autumnale: 19) and 28 bumble bees (Bombus terrestris/lucorum and B. pascuorum; R. hirta: 17; H. autumnale: 11) were tracked across their first ten visits. Initial (first inflorescence) choices strongly favored intact petals: naïve honey bees proportions: R. hirta 0.83, H. autumnale 0.73; naïve bumble bees: R. hirta 0.83, H. autumnale 0.72. Very few initially chose zero-petal inflorescences (honey bees: 0.04, 0.06; bumble bees: 0.00, 0.18). This was significant for honey bees on both plants (R. hirta χ² = 25.391, df = 2, p < 0.001; H. autumnale χ² = 13, df = 2, p = 0.002) and for bumble bees on R. hirta (χ² = 21, df = 2, p < 0.001). By the tenth visit, preferences disappeared and visitation proportions were approximately equal across treatments (honey bees: R. hirta χ² = 1.652, df = 2, p = 0.438; H. autumnale χ² = 0.333, df = 2, p = 0.847; bumble bees on R. hirta χ² = 1.33, df = 2, p = 0.513). H. autumnale bumble bee sample size was insufficient for robust tests, but trends were similar. - Temporal trends (beta regression): For both bee genera and both plants, the proportion of visits to all-petal inflorescences negatively correlated with visit number (2 of 4 analyses significant), whereas the proportion visiting zero-petal inflorescences positively correlated with visit number (all 4 analyses significant). No significant trends were detected for the four-petal treatment. Overall, showy ray petals strongly attract naïve first-time visitors, but the effect attenuates rapidly as bees gain experience within a foraging bout.

The findings support the hypothesis that showy floral signals, specifically ray petals in Asteraceae, function primarily as long-distance advertisements to attract naïve pollinators on their first approach to a patch. Naïve honey bees and bumble bees showed strong initial preference for intact-petal inflorescences, consistent with visual detection and innate biases for larger, more conspicuous, and symmetrical displays. However, as foraging proceeded, bees rapidly updated their choices based on learned information about reward availability and location, eliminating the initial preference by the tenth visit. Experienced bees showed reduced or absent dependence on petals, consistent with learned foraging strategies and differing criteria at long versus short range. These dynamics align with broader literature on bee perceptual constraints, UV guides, and the influence of display size on approach rates. The results imply that the evolutionary maintenance of large, colorful petals may be driven substantially by their role in securing first-time visits in competitive floral environments, after which learned behavior takes over. This learning ability likely exerts selective pressure on floral trait evolution, alongside classical morphological matches between flowers and pollinators.

This exploratory field study demonstrates that showy ray petals in Asteraceae primarily attract naïve, first-time bee visitors, with preferences diminishing rapidly as bees gain experience within a foraging bout. By contrasting naïve and experienced foragers, the work clarifies a key function of petals as initial advertisement signals. Future research should broaden taxonomic scope, manipulate long- versus short-range cues in mixed-signal patches, and test potential confounds of petal removal (e.g., nectar production changes or volatile emissions), to assess the generality and mechanisms underpinning naïve versus experienced pollinator responses.

- Naïve status refers to being naïve to the experimental patches and treatments; true lifetime species-level naiveté could not be confirmed. - Only two plant species (cultivars) and a limited set of bee taxa were studied, limiting generalizability. - Sample sizes constrained some analyses, notably for bumble bees on Helenium autumnale. - Petal removal could potentially alter nectar production or volatile emissions, which were not directly measured and could influence visitation independent of visual cues. - Experiments were conducted in a single garden site and season, which may limit environmental generality. - Ray petal treatments, while symmetric, may still have introduced subtle mechanical or tactile differences not fully controlled.