Fig. 3. Potential mechanisms of oscillatory CO₂ release patterns in Drosophila. (A–C) Infra-red video images show the flying fly from below, while recording wing kinematics, flight force production and the release of flight muscle-specific CO₂. Four UV light-activated fluorescent markers on the animal's abdomen (B) allow video-based in-flight tracking of abdominal pumping movements, and light intensity changes within the measurement area (red box) indicate proboscis movements during flight (C). (D–F) Simultaneously recorded flight data of (D) CO₂ release, (E) abdominal length and width changes based on movement of markers in B and (F) occurrence of the proboscis extension reflex (PER), during a 40 s flight sequence. A PER value of zero indicates that the proboscis is fully retracted, whereas a value of 1.0 means full extension. Gray bars indicate examples where CO₂ release decreases (inhalation) as the fly extends the proboscis. No moving visual stimuli were displayed in the surrounding panorama. (G–I) Cross-correlation coefficients r are plotted between (G) the derivative of muscle mass-specific mechanical power output of the flight muscles and the derivative of CO₂ release, (H) the derivative in abdominal length and CO₂ release, and (I) the derivative of proboscis movements and CO₂ release. L = cross-correlation temporal phase shift between data sets (phase lag). Each cross-correlation analysis was performed for six flight sequences over time t, using a sliding data window with 0.5t width. Length of the flight sequences was 141±77 s (mean ± S.D., N=3 flies). In this analysis we limited our data set to flies that showed pronounced and long-lasting gas release oscillations. Mean correlation coefficient r is plotted in black; gray areas indicate S.D.