Visual Edge Orientation Shapes Free-Flight Behavior in Drosophila

3 downloads 31 Views 222KB Size Report
Dec 6, 2007 - Packard Foundation (Michael H. Dickinson). We thank Dr. Michael Tarsitano, Jocelyn. Staunton, Jessica Choe and Will Parke for.
[Fly 1:3, 153-154; May/June 2007]; ©2007 Landes Bioscience

Brief Communication

Visual Edge Orientation Shapes Free-Flight Behavior in Drosophila Mark A. Frye1,* Michael H. Dickinson2

Abstract

2California Institute of Technology; Division of Biology; Pasadena, California USA

*Correspondence to: Mark A. Frye; University of California, Los Angeles; Department of Physiological Science; 621 Charles Young Dr. South; Los Angeles, California 90095-1606 USA; Tel.: 310.825.5360; Fax: 310.206.9184; Email: [email protected]

.

Certain features of fly flight behavior are similar to gaze stabilization in humans in that the eyes are held stationary between rapid reorientations that are thought to minimize the duration of corrupting motion blur.1 During free‑flight, gaze shifts take the form of rapid turns called body saccades, which punctuate straight segments,2 and are triggered by the visual expansion of approaching objects,3 and possibly other external and internal factors.4 Between saccades, free‑flight velocity is regulated by the apparent motion of high‑contrast edges with the visual surroundings.3,5 The direction of motion cues depends upon the animal’s motion as well as the edge’s orientation. For example, horizontal edges generate vertical motion, whereas vertical edges generate horizontal motion and both cues trigger saccades equally well.6 The extent to which different directional motion cues contribute to the control of flight speed and visual depth in flies is unknown. Here, we manipulated the orientation of motion cues, and measured the impact on these behavioral variables. We digitized flight trajectories in three‑dimensions at 60 frames‑per‑second7 within in a flight chamber 1‑meter in diameter and 0.6‑meter high lined with various black‑and‑white patterns. During flight in an arena lined with high‑contrast vertical stripes, flies ­continuously reiterate their basic saccadic flight pattern near the center of the arena (mean position, Fig. 1), and on average turn 40 cm from the approaching wall (Approach distance, Fig. 1). Replacing the vertical stripes with a uniform white panorama results in little change in either the mean position of flight paths or approach distance, although animals fly slightly faster and further between saccades (Velocity, Inter‑saccade distance, Fig. 1). Surprisingly, amid horizontal stripes the flight segments between the saccades, defined by any turn exceeding 300˚ sec‑1, are more curvilinear and distributed very close to the arena walls. In addition, flight velocity increases by 30%, and the distance between saccades increases by 40%. A hybrid arena consisting of half horizontal and half vertical stripes yields intermediate distributions such that flies closely approach the walls only on the horizontal‑striped side, and the velocity and distance distributions are shifted midway between those for the full‑field vertical and horizontal patterns (Fig. 1). The control of flight velocity and inter‑saccade flight distance are independent of the overall saccade rate since the mean inter‑saccade time interval varies across treatments between 1.4 and 1.7 per second, but is not statistically significant (Kurskall‑Wallis ANOVA p > 0.05). The finding that mean flight velocity is higher in the uniform arena than in the vertical striped arena confirms that optic flow is used to regulate flight speed in flies.3,5 The conspicuously curved trajectories in the horizontal stripe treatment are most likely due to the excessive flight velocity, because the trajectories are both less curved and slower in the hybrid arena. It is intriguing that the horizontal stripes should cause flies to fly farther, faster and closer to the walls between saccades. Why don’t they do so within the uniform arena? Despite our efforts, the uniform background actually contains faint shadows due to imperfect lighting conditions. Contrast gain adaptation is a physiological ­phenomenon by which visual responses in interneurons of the lobula plate are attenuated under high contrast conditions, such that relative response amplitude is larger for low

optic flow, flight control, lobula plate, vision, motor control Acknowledgements

©

20

07

LA

ND

ES

BIO

SC

IEN

This work was supported by the Whitehall Foundation (2006-12-10) and the Sloan Foundation (Mark A. Frye), and by the NSF (FD97-23424), the Office of Naval Research (FDN00014-99-0892) and the Packard Foundation (Michael H. Dickinson). We thank Dr. Michael Tarsitano, Jocelyn Staunton, Jessica Choe and Will Parke for data collection.

.D

Keywords

CE

Previously published online as a Fly E-publication: http://www.landesbioscience.com/journals/fly/article/4563

ON

OT D

Original manuscript submitted: 06/12/07 Manuscript accepted: 06/13/07

UT E

of California, Los Angeles; Department of Physiological Science; Los Angeles, California USA

Insects rely on visual cues to estimate and control their distance to approaching objects and their flight speed. Here we show that in free‑flight, the motion cues generated by high‑contrast vertical edges are crucial for these estimates. Within a visual environment dominated by high‑contrast horizontal edges, flies fly unusually fast and barely avoid colliding with the walls of the enclosure. The disruption of flight behavior by horizontal edges provides insight into the structure of visually‑mediated control algorithms.

IST RIB

1University

www.landesbioscience.com

Fly

153

Visual Flight Control

Figure 1. Individual flies were released within a 1‑meter circular chamber and tracked with a three‑dimensional infrared video system.7 The arena was i­lluminated from the outside with a circular array of high intensity halogen lamps, and striped patterns were affixed to the inner walls. Other analyses as described previously.3,11 Results from different visual treatments are plotted on each row as indicated. (Column 2) Mean arena position is represented by pseudo‑colored surface histograms, which have been normalized to the number of data points such that area‑under‑the‑curve equals 1. (Column 3) For each saccade, the approach distance to the wall along the fly’s heading was plotted against the angular location of the fly’s heading (0–360°, see raw trace for horizontal/vertical stripes). For clarity, data from all flies were pooled, scrambled, and decimated such that only every 10th datum was plotted. Mean ­collision ­ distance binned in 10‑degree increments of arena heading is plotted with red lines. For comparison, shaded red bars represent the ­ envelope of mean ­ collision distance for the vertical stripe condition and are overlaid for subsequent treatments. (Columns 4 and 5) For each saccade interval, mean ­three‑dimensional flight velocity and total flight distance was binned into histograms, which are normalized to the number of data points such that ­area‑under‑the‑curve equals 1. For comparison, gray histograms represent distributions for the vertical stripe condition and are overlaid for subsequent ­treatments. Mean velocity and inter‑saccade distance is indicated numerically. Kruskal‑Wallis Nonparametric one‑way ANOVA p