Table 1.
Details of each colony.
Fig 1.
Defining the dance floor for Colony 1, Day 1, 1400 to 1430.
(A) The blue+green region is the convex hull of all dances observed during the 30-minute observation. The yellow+green region is the 2-SD confidence ellipse. The major and minor axes of the ellipse are shown in black. (B) The intersection of the convex hull and the confidence ellipse from (A). The blue lines are the major and minor axes of the confidence ellipse recentred at the centroid of the dance floor. In this midday example, the dance floor captured 96 of the total 105 observed dances (91.4%), Area = 648.4 cm2, Perimeter = 96.3 cm, the dance floor centroid was at X = 20.6 cm and Y = 19.8 cm, Length (major axis) = 41.4 cm, Width (minor axis) = 25.3 cm, and Angle = 40.9° counterclockwise from horizontal. The black arrows indicate the location of the hive entrance. Individual dance locations are indicated by black dots. This example is representative of typical midday conditions, when dance activity was highest; most other observation periods exhibited similar spatial structure, with reduced definition only during early-morning or evening observation periods when dance numbers were low.
Fig 2.
Dance floors from three observation periods from Day 1 of Colonies 2 and 5.
Late morning, midday, and early evening observations are shown for 8-frame (A) and 4-frame (B) observation hives. All dance floor metrics are listed in supplemental S1 Table. Note that the angle of the major axis varies more in the 8-frame hive than in the 4-frame hive.
Fig 3.
Dance floors in Colony 1 across days and observation times.
The 2000-hour observations were omitted to improve readability and because there were only 10, 6, and 6 dances on days 1, 2, and 3, respectively. All eight frames (100 × 100 cm) are shown.
Table 2.
Mean values of the seven dance floor metrics.
Table 3.
Results of the univariate (Area, Length, Width, Angle, X, and Y) and multivariate models.
Fig 4.
Cumulative dance floor regions for all 8-frame and 4-frame hives.
(A) Each polygon represents the ellipse intersection calculated from all waggle dances across all days, times, and colonies for 8-frame hives. (B) Each polygon represents the equivalent cumulative region for 4-frame hives. These footprints illustrate the method’s adaptability to long-term or aggregate spatial patterns but were not used for statistical analysis in the present report. Dance floor metrics are listed in supplemental S1 Table. Note the lack of dances on and near the wooden boundaries of the frames of the beehive (at x = 50 cm and at y = 25 cm). Note also the perceived lack of dances at each of the 5 × 5-cm grid lines due to observers marking dances beside grid lines on the printed maps rather than on the grid lines.
Fig 6.
Summary of the most informative relationships among the four shape parameters.
(A) Area varied significantly by TrialDay, though no consistent pattern was evident across days. (B) Length varied significantly with both TrialDay and hive Size; 4-frame hives generally showed longer dance floors than 8-frame hives. (C) Width exhibited a significant interaction between hive Size and DOY: in 8-frame hives, width peaked mid-season before declining, whereas in 4-frame hives, width was lowest mid-season and increased late in the season. Boxplots represent daily distributions; overlaid curves are quadratic regressions of colony-level means at the median day of each colony. Labels T1–T8 identify the colonies (trials). (D) Dance floor Angle was influenced by a significant Size × Time interaction: in 4-frame hives, orientation remained relatively stable throughout the day, while in 8-frame hives, angle varied more strongly with time. Loess smoothing curves with 95% confidence intervals are shown. Colour key (panel B) applies to panels B–D.
Fig 7.
Summary of spatial dynamics in dance floor location.
(A) X (horizontal location) of the dance floor centroid as a function of DOY and hive Size. Both main effects and their interaction were significant. In 8-frame hives, X remained relatively stable throughout the season, whereas in 4-frame hives, horizontal position varied over time. (B) Y (vertical location) of the centroid modelled using a full fixed-effects model that included Size and its interactions with Time, DOY, and TrialDay. All terms were significant except for Time and its interaction with Size. Y remained stable in 8-frame hives but increased across the season in 4-frame hives. In both panels, quadratic regressions were fit to colony-level means plotted at the median day of each trial. Labels T1–T8 identify the colonies (trials).
Fig 8.
Multivariate effects of key predictors on overall dance floor structure.
Bars show –log₁₀(P) from the MANOVA, with higher values indicating greater significance. Text labels denote the corresponding Pillai’s trace values. The red dashed line indicates the P = 0.05 threshold (–log₁₀(0.05) ≈ 1.3). Hive Size, DOY, and their interaction explained the greatest multivariate variation, followed by a weaker but significant effect of the interaction of hive Size and Time of day. All other factors had minimal effect.
Fig 5.
Dance-density and dance floor-inclusion heatmaps.
Dance densities were calculated from all observation periods over all days from (A) Colonies 1–4 (8-frame) and (B) Colonies 5–8 (4-frame). Note that the lack of dances on the wooden frame boundaries and the perceived lack of dances on the grid lines are still visible in panel A (see Fig 4). Dance floor-inclusions were calculated for each point in the hive as the proportion of observations that are included in the calculated dance floor from (C) Colonies 1–4 (8-frame) and (D) Colonies 5–8 (4-frame).