Extended Data Fig. 10: Similarities and differences in flight behaviour between the lights-on vs. lights-off cue conditions.
a, Aggregated flight paths that occurred across 10 consecutive days of experiments for one example bat, divided into successive lights-on, lights-off and lights-on′ periods (ordered from left to right). Coloured are the four most common flight paths. The red and blue flight paths occurred predominantly in lights-on and lights-on′ sessions. The green and magenta paths occurred predominantly in the lights-off condition. b, Distribution of the occurrence of the four most prevalent flight paths, coloured according to a, during different phases of the experiment across lights-on and lights-off sessions. Flight preference changed dramatically during each period. c, Aggregated flight paths across ten consecutive days for one representative bat divided into chronological thirds of each session where light levels were held constant. Shown are the three most common flight paths in red, blue and green. d, Distribution of the occurrence of the four most prevalent flight paths (colours) during different phases of the experiment across each third of ten sessions. e, Proportion of unique (that is, unstructured) flight paths increase on average in the lights-off cue condition (dark) compared to when the lights are kept on for the duration of the session. The blue line indicates the mean relative proportion of unique flights flown in the lights-on/off experiment. The dark grey line is the mean proportion of unstructured flights for bats where the light is kept on for the duration of the experiment. Shading for both plots represent 95% confidence intervals of the mean. f, The mean number of flights per minute is not significantly different in the light or dark conditions (P > 0.05 two-tailed Wilcoxon rank sum test, n = 80 and 40 binned minutes in the lights-on and lights-off periods respectively). Points are horizontally jittered for ease of visualization. g, Flight duration as function of cue condition. Flight duration is significantly longer in the lights-off condition, (P = 0.0038, two-tailed Wilcoxon rank sum test) although the effect size is small (that is, the distributions are highly overlapping; n = 1,567 and 746 flights in the lights-on and lights-off periods respectively). Points are horizontally jittered for ease of visualization. h, i, Across-day and across-condition behavioural stability for the lights-on versus lights-off experiments. h, Average correlation of repeated paths relative to day one for each bat (n = 5 flight paths). Each colour indicates a different flight path. Bars indicate 95% confidence intervals of the mean correlation on each day. i, Correlation values of shared flights to the mean of their flight path in the first lights-on period. Flights along the same path were not significantly different between the light and dark cue condition (P = 0.21 two-tailed Wilcoxon rank sum test, n = 268, 150, 267 flights for lights-on, lights-off and lights-on conditions, respectively). Points are horizontally jittered for ease of visualization. * P< 0.01.
