Subjects and pre-training
We used 30 B. impatiens workers as subjects, selected from a colony obtained from Koppert Biological Systems (Romulus, MI, USA). The colony was provided with pollen ad libitum and housed in a plastic box (L × W × H: 22·0 × 24·0 × 12·0 cm). Experiments occurred in a room-sized experimental chamber (L × W × H: 3·05 × 1·92 × 1·55 m), fitted with a screen door to permit observation. The experimental chamber was connected to the colony via a gated buffer box (L × W × H: 35·0 × 22·0 × 15·0 cm) that allowed us to release individual foragers, fitted on the thorax with numbered tags (E.H. Thorne Ltd., Wragby, Lincolnshire, UK) for identification. The chamber was illuminated by fluorescent lighting (see Fig. S1a, Supporting Information; Sylvania Cool White 34 Watt bulbs, # F40CW1SS, Osram Sylvania, Danvers, MA, USA 560 lux measured at centre of array).
In order to train bees to visit the experimental chamber and forage at the floral array, we allowed the colony free overnight access to a pre-training array. The pre-training array offered nine feeders, each providing 10 mL of 30% w/w sucrose through a cotton wick. The pre-training array was similar to the experimental array, and it consisted of a green horizontal board (60 × 45 cm, DecoArt acrylic paint, ‘Avocado’ #DA052) with feeders spaced 10 cm apart. Each white cylindrical feeder base (height: 5·5 cm; diameter: 1·7 cm) was topped with a light grey artificial flower, the same size (5·0 cm diameter) and shape (circle, N = 4 or petaloid, N = 5) as the flowers used in foraging experiments. No patterns were present on pre-training flowers. We varied the wick’s position relative to the centre or edges on each of the nine flower surfaces to prevent bees from learning to feed in a particular flower region. The floral arrays used in pre-training and experiments were positioned on a stool in the centre of the experimental chamber, at a height of 50 cm above the ground.
The floral array held 12 artificial flowers (Fig. 1e,f), arranged at 10cm intervals in a 3 × 4 grid. Flowers consisted of a white cylindrical base (height: 5·5 cm; diameter: 1·7 cm) connected to a flower top (5·0 cm diameter). Flower tops were light blue, printed on water-resistant paper (Adventure Paper, National Geographic, Margate, FL, USA), using a Canon Pixma MX860 inkjet printer, and laminated (Xyron matte laminate, Scottsdale, AZ, USA). Figure 1g shows the reflectance spectra for colours present on the floral array. Because humans and bees are sensitive to different wavelengths of light, these colours can be represented in bee colour space (Chittka 1992) using the reflectance data, information about the lighting conditions, as well as the sensitivity of B. impatiens’ three photoreceptors (Skorupski & Chittka 2010) (Fig. S1, Supporting Information). Using this model of bee colour space, we determined that our flower colours offered substantial green contrast against the background, as well as green contrast between corolla and guide (Fig. S1c, Supporting Information). Green contrast is critical for bees’ long-range detection of flowers (Giurfa et al. 1996; Kevan & Backhaus 1998; Spaethe, Tautz & Chittka 2001). The flower top had a small central hole (d: 1·5 mm) through which bees accessed a well hidden in the white plastic base. Depending on treatment, this well contained either 10 μL of 50% w/w sucrose (rewarding flowers) or 10 μL of deionized water (unrewarding flowers). Rewarding and unrewarding flowers were distributed haphazardly across the grid, and their position was changed between foraging trips.
Flowers were either circular (N = 16 bees) or petaloid (N = 14 bees). The petaloid flower had 19·8% less surface area, but 51·6% more perimeter than the circle (surface area of circle: 19·63 cm2; petaloid: 15·75 cm2; perimeter of circle: 15·70 cm; petaloid: 23·80 cm; measured with Adobe Photoshop CS3, Adobe Systems, San Jose CA, USA). Shape differences may have altered the detectability of flower targets, as honey bees detect circular targets at a longer distance than petaloid targets (Ne’eman & Kevan 2001). On any given foraging trip, six of these flowers were plain, and six had a light (human white; bee UV-blue), radially symmetrical, nectar guide pattern. The pattern was identical for both circular and petaloid flowers. Although our artificial flowers were not modelled on a specific species, many bee-pollinated flowers present a blue corolla with light, UV-reflective radiating lines (e.g. Iris, Salvia, Ipomoea). Generally, chromatic contrast between pattern and corolla is important in guiding bumblebees’ orientation towards flowers (e.g. Lunau, Wacht & Chittka 1996; Lunau et al. 2006). Likewise, both circular- and star-shaped flowers commonly display star-shaped guides (Dafni & Kevan 1996).
Each bee was videotaped (30 frames/s; Sony DVM-60PR Mini DV cassettes) during two foraging trips, occurring 2 days apart. During each trip, the bee encountered both plain and patterned flowers; one type was rewarding and one type was unrewarding, but this relationship was switched for each individual bee across foraging trips. Repeatedly assaying the same forager allowed us to determine whether handling times were shorter when rewarding flowers had patterns, even if individual bees varied in their foraging speed. On Day 1, 14 bees had plain flowers rewarding/patterned flowers unrewarding, followed by patterned flowers rewarding/plain flowers unrewarding on Day 3. For seven of these bees, all flowers were circular, and for seven of these bees, all flowers were petaloid. Conversely, 16 bees on Day 1 had patterned flowers rewarding/plain flowers unrewarding, followed by plain flowers rewarding/patterned flowers unrewarding on Day 3. For nine of these bees, all flowers were circular, and for seven of these bees, all flowers were petaloid. Between foraging trips (Day 2), bees were provided access to the pre-training array described above. During a foraging trip, bees were allowed to visit flowers until they had drained all six of the rewarding flowers, or until they did not visit the array for 3 min, at which point they were collected and returned to the colony. After each foraging trip, flowers were cleaned with 30% ethanol to remove any scent marks deposited by foragers.
We used iMovie 8·0·6 (Apple Computer Inc., California, USA) to record the sequence of landings and measure the frame-by-frame details of initial landings on up to 12 flowers (six rewarding, six unrewarding) within a foraging trip. If a bee departed its first visit to a rewarding flower without having located the reward, we scored the visit as a ‘miss’ and used data from its first successful revisit to that flower. During a typical visit to a rewarding flower, a bee landed, searched on the surface of the flower, located the reward, consumed all the reward and then searched again on the surface of the flower before leaving. We measured the time spent searching for the reward after landing, as well as the time spent searching on the flower after feeding. Our sample sizes for certain comparisons were <30 because some bees did not feed on their Day 3 rewarding flower type (N = 2). Sample sizes for comparisons of post-feeding search time on the final flower visited were similarly reduced because a few bees (N = 3) remained motionless for longer than 30 s and were therefore excluded from this analysis.