Fig 1.
Experimental schematic showing tracking, lifespans, and behavioral segmentations across timescales. A Image showing the experimental arena as viewed from below. The behavior of 4 D. melanogaster is captured simultaneously while giving each fly enough room to freely carry out all behaviors except flight. B Magnified view of a single individual showing tracks for each node of the SLEAP skeleton. Each color denotes a node and circle sizes increase with time. C Survival curve of the 47 flies included in the experiment. Death occurs on average after ~ 119 h, or almost 5 full days into the experiment. D Ethogram and egocentrized traces for each tarsi and a raster denoting proboscis visibility. E Barplot showing the geometric means of stereotyped behavior components across all flies and all complete 24 h periods. F Barplot showing the geometric means of stereotyped behavior components across all flies and all hours grouped by experimental day.
Fig 2.
Principal component analysis of stereotyped behavioral components of all flies across all experimental hours. A Biplot showing the projections of individual fly hours and loadings of the centered log-ratio of each stereotyped behavioral component. Dark gray dots show time points from when lights are off and light gray dots show time points when lights are on. B Mean projection of PC1 against time of day for all complete 24h periods of all flies. The shaded region is the standard error. C Mean Day/Night difference vs day of experiment as measured by the absolute value the average projection onto PC1 during night hours minus the projection onto PC1 during day hours. The shaded region is the standard error. D Mean projection of PC2 against time of day for all complete 24h periods of all flies. The shaded region is the standard error.
Fig 3.
Diurnal patterns of behavior on experimental Day 1. A Barplot showing the geometric means of stereotyped behavioral components of the first experimental day across all flies. B Ternary plot showing the geometric means of condensed behavioral components across all flies for each hour of Day 1. Directions along PC1 (dashed) and PC2 (solid) as calculated by perturbing the geometric mean of the displayed data points [32]. The ternary plot was generated using the Ternary Plots package in MATLAB [33]. C Grooming enrichment with respect to the geometric mean of the condensed grooming behavioral component of the first experimental day for each fly with bootstrapped confidence intervals. D Mean proboscis bout length by hour of the first experimental day. The shaded region is the standard error. E Mean locomotion speed (mm/s) during stereotyped locomotion state by hour of the first experimental day. The shaded region is the standard error.
Fig 4.
Day wise behavioral changes throughout the experiment. A Ternary plot showing the geometric mean of the condensed stereotyped behavioral components of the first hour after lights on across all surviving flies for each complete 24h period. Directions of PC1 (dashed) and PC2 (solid) are also shown, as calculated based on perturbation of the geometric means of all hours from Day 1. B Mean proboscis bout length by day of experiment. The shaded region is the standard error. C Mean locomotion speed (mm/s) during the stereotyped locomotion state by day of experiment. The shaded region is the standard error.