Page 1
Supplementary material:
Effect of flower perceptibility on spatial-reward associative learning by bumble bees
Behavioral Ecology and Sociobiology
Shohei G. Tsujimoto* & Hiroshi S. Ishii
Graduate School of Science and Engineering, University of Toyama, 3190 Gofuku, Toyama 930-8555, Japan
*Corresponding author
E-mail:
Telephone: +81-90-1173-6411
Table S1 Generalized linear mixed model (a: with binomial errors and logit link function) and linear mixed models (b and c) of the effects of flower size and foraging bout number on bumble bee behavior. All models considered bee identity as a random effect. In (b) and (c), non-significant interactions were removed from the models.
Fixed factor / Estimate / Standard error / Statistics value / P value(a) Probability of visits to high-rewarding flowers
Intercept / 3.78 × 10-1 / 0.72 × 10-1 / Z = 5.24 / <0.001
Foraging bout number / 3.38 × 10-2 / 0.93 × 10-2 / Z =3.63 / <0.001
Flower size: 6 cm (vs. 2 cm) / 5.16 × 10-2 / 10.33 × 10-2 / Z = 0.50 / 0.617
Bout number × Flower size / -2.93 × 10-2 / 1.39 × 10-2 / Z = -2.11 / 0.0350
(b) Route repeatability index
Intercept / -4.85 × 10-1 / 0.16 × 10-1 / t113 = -29.49 / <0.001
ln(bout number) / 5.41 × 10-2 / 1.05 × 10-2 / t113 = 5.15 / <0.001
Flower size: 6 cm (vs. 2 cm) / -0.95 × 10-2 / 1.83 × 10-2 / t113 = -0.52 / 0.609
(c) Return interval
Intercept / 25.17 / 0.34 / t 45088 = 75.40 / <0.001
Bout number / 4.76 × 10-1 / 0.91 × 10-1 / t 45088 = 5.22 / <0.001
Flower size: 6 cm (vs. 2 cm) / 0.65 × 10-2 / 25.50 × 10-2 / t 45088 = 0.03 / 0.980
Fig. S1 (a) Relative spectral-power distribution of the UVA-emitting fluorescent light used in experiments (Vita-Lite, 3826EX-LS, Light Sources Inc., CT, USA) and D65 (Wyszecki and Stiles 1982), (b) spectral reflectance of the blue artificial flowers and green background, and (c) color loci of the blue flowers in the bee color hexagon (Chittka et al. 1992) under the UVA-emitting fluorescent light (filled circle) and D65 (open circle), and of the green background (grey circle). Spectral reflectance measurements were obtained using a charge-coupled-device spectrometer (BRC112; BWTEK Inc., DE, USA), deuterium–tungsten light source (BDS130; BWTEK), white standard (SRR; BWTEK), and Y-shaped fiber optic cable (FRP-400-0.22-1.5-UV; BWTEK) at 1-nm wavelength intervals from 300 to 800 nm. The probe was placed at a constant angle of 45° relative to the object surface to measure diffuse spectral reflectance (Chittka and Kevan 2005). Three replicates were measured and their mean value was used.
Fig. S2 Frequency distribution of the return and estimated intervals between revisiting flowers. Return interval was defined as the number of flowers visited before revisiting the same flower (the arrow represents the median). Mean inter-flower flight time between the nearest two flowers and handling time of flowers was 0.795 ± 0.040 (mean ± SE, N = 24) and 2.471 ± 0.572 (mean ± SE, N = 24), respectively. Thus, we estimated the interval time for the return to the same flower as 2.471X + 0.795(X + 1), where X is the return interval.
Fig. S3 Design of the ten flower arrangements (1–10) used in the 12 replicates performed for each of the experiments using large and small flowers, and the location identity of the flowers. Arrangements 1 and 2 were used twice and arrangements 3–10 were used once for each flower size; two of the arrangements were used twice in each experiment because we prepared only 10 different arrangements for 12 replicates. Filled and open circles represent high and low rewarding flowers, respectively. The arrow indicates the direction in which the beehive was located.
Fig. S4 Changes in (a) probability of visits to high-rewarding flowers, (b) route repeatability index and (c) return interval with foraging bout number. In (a) and (c), each point represents the mean (± SE) of the individual means of the dependent variables at each foraging bout. In (b) each point represents the mean (± SE) at each foraging bouts. Lines represent regressions based on GLMM (a) and linear mixed models (b, c).
Appendix S1 The relevant raw data [separate XLSX file].