Odor localization requires visual feedback during free flight in Drosophila melanogaster
Mark A. Frye1,*,
Michael Tarsitano1 and
Michael H. Dickinson2
1 Department of Integrative Biology, University of California —
Berkeley, Berkeley, CA 94720, USA
2 Bioengineering, Mail Code 138-78, California Institute of Technology, 1200
E. California Blvd, Pasadena, CA 91125, USA
View larger version (26K):
[in a new window]
|
Fig. 1. Geometry of free-flight arena and experimental visual patterns. (A) Flies
explored an arena 1 m in diameter, 0.6 m high. A pair of video cameras tracked
Drosophila flight trajectories within a conical region of the arena
(broken lines). (B—F) Different black-and-white patterns, printed on
white paper, formed experimental visual landscapes. In all cases, the ceiling
and floor of the arena were left black, forming two contrasting vertical
edges. Circular icons, representing the top view of each visual treatment, are
used in subsequent figures. The asterisk indicates a region of pattern that
may have biased trajectory distributions (see Materials and methods).
|
|
View larger version (49K):
[in a new window]
|
Fig. 2. Flies fail to localize the horizontal position of an attractive odor source
in the absence of textured visual surroundings. Sample flight trajectories,
viewed from above, of flies flying within random checkerboard (A) and uniform
white (B) visual surroundings. Embedding a vial of apple cider vinegar in the
floor of the arena (location indicated by white circles) resulted in biased
flight trajectories for animals flying within the random (C), but not the
uniform (D), treatment. Average position indicated with 2-D histograms plotted
in pseudocolor. Position bins are 50 mmx50 mm. Number of flies
(N) and total flight time in min (t) are indicated for each
plot.
|
|
View larger version (26K):
[in a new window]
|
Fig. 3. Flies respond to odor by flying at lower arena altitudes. Sample flight
trajectories for visual and olfactory treatments viewed from the side. (A)
random checkerboard, (B) uniform, (C) checkerboard plus odor and (D) uniform
plus odor. Probability histograms show distributions for altitude in (E)
random checkerboard, (F) uniform, (G) checkerboard plus odor and (H) uniform
plus odor. Location of odor source indicated by white circles. Numbers of
flies and total flight times are the same as in
Fig. 2.
|
|
View larger version (15K):
[in a new window]
|
Fig. 4. Measured parameters from fly flight trajectories. (A) Flies exhibit
segments of straight flight punctuated by turns of approximately 90°. (B)
Saccades are characterized by rapid increases in angular velocity, therefore
easily distinguished in time and space. (C) For each saccade, we measured
several parameters of flight control, including saccade angle, collision
distance, arena heading, segment length and odor distance.
|
|
View larger version (22K):
[in a new window]
|
Fig. 5. For a sample flight trajectory within the random checkerboard arena,
saccade angle (A) does not fluctuate, whereas velocity (B), segment length
(C), and collision distance (D) vary as a fly approaches the odor source (E).
The gray bar highlights low odor distance.
|
|
View larger version (32K):
[in a new window]
|
Fig. 6. Flight speed between saccades varies within an envelope across segment
length. Each point represents the mean velocity between two saccades. Data
plotted from the random checkerboard treatment. Numbers of flies and total
flight time are the same as in Fig.
2.
|
|
View larger version (28K):
[in a new window]
|
Fig. 7. Segment length between saccades is reduced near the odor source. Segment
length plotted against odor distance for each saccade exhibited by flies
flying in (A) random, (B) uniform, (C) random checkerboard plus odor and (D)
uniform background plus odor. A greater proportion of saccades exhibited in
odor treatments occurs near the source, and saccades exhibited near the source
follow short segments (gray bars highlight data points within a 75 mm radius
of the odor source). Probability distributions for each saccade were generated
by binning segment length according to odor distance: (E) random checkerboard
background and (F) uniform white background plus odor (red) or minus odor
(blue). Median segment length was calculated for each 30 mm bin of odor
distance: (G) random checkerboard background and (H) uniform white background
plus odor (red) or minus odor (blue). Numbers of flies and total flight time
are the same as in Fig. 2.
|
|
View larger version (50K):
[in a new window]
|
Fig. 8. Flies orient towards an odor source located near the wall at 270°
(indicated by broken lines) when flying within visually textured surroundings.
For each saccade, collision distance is plotted against arena heading, with
median collision distance for bins in arena heading indicated in red and
optimized sine fit indicated in blue. The amplitude of the sinusoidal fits
indicates the distance between the peak in the average transit distribution
and the center of the arena (for details, see Appendix). Random checkerboard
(A), uniform (B), random plus odor (C) and uniform plus odor (D) treatments.
Numbers of flies and total flight time are the same as in
Fig. 2.
|
|
View larger version (48K):
[in a new window]
|
Fig. 9. Flies localize odor within backgrounds of uniformly spaced vertical
stripes. Sample flight trajectories from flies flying within arenas of (A)
three thick vertical stripes and (B) evenly spaced vertical stripes. Embedding
a vial of apple cider vinegar in the floor of the arena (location indicated by
white circles) resulted in biased flight trajectories for animals flying
within both three-stripe (G) and multiple vertical stripe (H) treatments.
Average position indicated with 2-D histograms plotted in pseudocolor.
Position bins are 50 mmx50 mm. Number of flies (N) and total
flight time in min (t) are indicated for each plot.
|
|
View larger version (21K):
[in a new window]
|
Fig. 10. Flies respond to odor by flying lower in the arena within vertical striped
visual conditions. (A—D) Sample flight trajectories for visual
treatments and odor viewed from the side (odor location indicated by white
circles). (E—H) Average probability distributions for altitude (number
of flies and total flight time are the same as in
Fig. 9).
|
|
View larger version (23K):
[in a new window]
|
Fig. 11. Segment length between saccades is reduced near the odor source in
vertically striped treatments. (A,B) Probability distributions for each
saccade were generated by binning segment length according to odor distance
(raw data not shown). (C,D) Median segment length was calculated for each 30
mm bin of odor distance. Odor treatments are indicated in red, whereas
non-odor treatments are indicated in blue. Numbers of flies and total flight
time are the same as in Fig.
9.
|
|
View larger version (51K):
[in a new window]
|
Fig. 12. Flies orient towards an odor source when flying within visual surroundings
containing vertically oriented edges (odor location, 270°, indicated by
broken lines). For each saccade, collision distance is plotted against arena
heading, with median values of collision distance for bins in arena heading
indicated in red and optimized sine fit indicated in blue (see Appendix).
(A,B) Non-odor treatments, (C,D) odor treatment. Numbers of flies and total
flight time are the same as in Fig.
9.
|
|
View larger version (28K):
[in a new window]
|
Fig. 13. Within a background of horizontal stripes, flies show qualitatively
different flight trajectories compared with other visual conditions and fail
to localize an odor source. (A,B) Sample flight trajectories and (C,D) mean
transit distributions indicate that flies fly curved paths close to the wall.
(E—H) Flies respond to odor by reducing altitude, as in
Fig. 3. Probability
distributions (I) and median segment length (J) shift for animals in odor as
in other visual conditions. (K,L) Modulations in collision distance are weak
for animals tested in odor. Median segment length and sine fits are determined
as in Figs 8,
12.
|
|
View larger version (49K):
[in a new window]
|
Fig. 14. Flight simulations based on trajectory statistics exhibited by real flies
show that modulation of segment length and collision distance is sufficient to
account for odor localization. Individual trajectories are composed of 200
saccades, and average spatial distributions resulting from 100 simulations are
plotted in pseudocolor as in Fig.
2. (A) No odor modulation of saccade statistics. (B) For saccades
that occur by chance near the odor source, segment length is reduced. (C)
Collision distance scaled as a sine function of approach angle, as exhibited
by real flies (see Appendix). (D) Both segment length and collision distance
modulated by odor.
|
|
© The Company of Biologists Ltd 2003
