QUICK SEARCH:   [advanced] Author: Keyword(s):
Home     Help     Feedback     Subscriptions     Archive     Search     Table of Contents

First published online January 8, 2007
Journal of Experimental Biology 210, 311-324 (2007)
Published by The Company of Biologists 2007
doi: 10.1242/jeb.02646

This Article Summary Full Text Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Wöhl, S. Articles by Schuster, S. Search for Related Content PubMed PubMed Citation Articles by Wöhl, S. Articles by Schuster, S.

The predictive start of hunting archer fish: a flexible and precise motor pattern performed with the kinematics of an escape C-start

Saskia Wöhl and Stefan Schuster^*

Universität Erlangen-Nürnberg, Institut für Zoologie II, Staudtstrasse 5, D-91058 Erlangen, Germany

View larger version (7K): [in this window] [in a new window]   Fig. 1. Diagram based on Domenici and Blake (Domenici and Blake, 1997) and Schriefer and Hale (Schriefer and Hale, 2004) summarizing present knowledge of fast-start motor patterns in fish. The present study provides the first demonstration that fast C-starts, on which most neurobiological work has focussed, can be used in precisely aimed feeding strikes (red arrow). Our findings imply that C-start circuitry can be used to drive rather complex behavior at high precision and top speed.

View larger version (14K): [in this window] [in a new window]   Fig. 2. The predictive start of archer fish. (A) After the shot of a group member has just dislodged an aerial insect, a responding fish (lower right) turns from its initial orientation (dotted line) and takes off directly towards the later point of impact, with a speed matched to distance. Because the initial values of prey motion (initial height, speed and take-off angle) vary independently over broad ranges, predicted points P of impact can be anywhere within a large area. They must be inferred from a quick judgement of the prey's initial motion, based on which the fish must rapidly select an appropriate turn (angle ) and take-off (speed v). The match that the responding fish must make is illustrated by showing three ways (in different colors) in which a given shot might dislodge the target insect. The correspondingly different impact points P1-P3 (shown in the respective colors) would require speed levels v1-v3 and turning angles 1-3. Responses aimed at P1 and P2 would require different turns but the same take-off speed, whereas responses to P2 and P3 would require the same turn, but different take-off speed. (B) The actual turning range and accuracy in the set of predictive starts (N=76) whose kinematics are analyzed in this study using digital high-speed video. Turns were accurately set to the later point of impact and accuracy did not decrease when larger turns were required. The regression line (r2=0.9016, P<0.0001) is not significantly different from that expected if actual turn size equalled the expected turn size.

View larger version (25K): [in this window] [in a new window]   Fig. 3. An archer fish predictive start imaged at 500 frames s-1. Within 22 ms, the fish rotated its anterior body by 43.6° towards the later point of impact of a dislodged prey insect and then accelerated to a 1.225 m s-1 take-off speed. The silhouette of the fish is shown every 2 ms. The background color indicates the two distinct kinematic stages of the response, an initial bending phase (stage 1, dark blue) and a straightening phase (stage 2, light blue). Subsequent frames show the actual take-off. The location of the stretched-straight center of mass (CM, red asterisk) is indicated in each frame and the yellow dots show the points digitized in each frame for later analysis. The first two dots were used to define the turning angle and the position of the relatively stiff anterior body part to which the CM is adjacent.

View larger version (26K): [in this window] [in a new window]   Fig. 4. Kinematics of archer fish predictive starts. Comparison of two responses (imaged at 500 frames s-1) in which the fish had to execute a small turn (filled circles; response of Fig. 3) and a larger turn (open circles; frames shown in Fig. 11A). Color coding of the two kinematic stages is as in Fig. 3 with dark blue highlighting the initial bending phase and light blue indicating subsequent straightening. Time is set to zero at the start of stage 2. (A) Time course of chord length (CL) shows initial bending and later straightening. Inset illustrates how CL is defined as the minimal distance between mouth and caudal peduncle, normalized to body length. The amount of bending (i.e. the minimum CL) was larger for the larger turn. The courses shown are smoothed using a 5-point moving average. (B) The course of turning (accumulated angle ). The inset illustrates how the angular increment was derived by lines (dotted) drawn through the anterior body portion of the fish in successive frames. The lesser turn (filled circles) already aligned the fish to the point of impact by the end of stage 1 and the angle was then kept to. In the large turn (open circles) turning continued and alignment was achieved during stage 2. (C) The accumulated displacement (s). The inset illustrates how the increment s between successive frames was derived. The distances (a, b) between the mouth and the point of intersection (red) of the stippled lines through the anterior body part were taken and s taken as their difference. For each frame, the mouth position was independently digitized 3 times and the average position was taken. In the two responses shown, speed values acquired in the first 20 ms after the end of stage 2 were 1.225 m s-1 for the lesser turn (filled circles) and 1.015 m s-1 for the wide-angle turn (open circles).

View larger version (18K): [in this window] [in a new window]   Fig. 5. Profiles of angular speed and acceleration of archer fish predictive starts. Same starts as those analyzed in Fig. 4, silhouettes of fish shown in Fig. 3 and Fig. 11A. Open circles: response with large turn (82.1°). Filled circles: response with lesser turn (43.6°). Course of angular speed d/dt (A) and of angular acceleration d2/dt2 (B). Time zero is at start of stage 2. Background color: dark blue (stage 1), light blue (stage 2). In the lesser turn, braking at end of stage 1 was large and stage 2 angular speed approximately zero. For the larger turn angular speed was nonzero during the initial part of stage 2.

View larger version (17K): [in this window] [in a new window]   Fig. 6. Linear speed (A) and acceleration (B) profiles of archer fish predictive starts. Filled circles relate to the start shown in Fig. 3 (take-off speed 1.225 m s-1, turning angle 43.6°). The half-open circles relate to a response with lesser take-off speed (0.770 m s-1) but similar turning angle (39.5°). Speed and acceleration data end three frames before actual end of stage 2. This is because after that, the 5-point moving regression method used to take the derivatives causes ambiguities.

View larger version (30K): [in this window] [in a new window]   Fig. 7. The spectrum of kinematic variations among archer fish predictive fast-starts. The predictive starts required very different turns and take-off speed levels. This desired variability in the motor output should be reflected in corresponding variations in kinematics. To test this, a large number of responses (N=76) with known angles of turning and take-off speed were analyzed in detail using digital high-speed video (500 frames s-1). The substantial variability among these responses is illustrated by showing (A) the course of chord length changes (CL), (B) the course of accumulated angle () and (C) the course of accumulated displacement (s) for all starts analyzed. Variables and how they deviate for different turning angles and take-off speed are introduced in Figs 4, 5, 6, respectively. To aid comparison, time zero is set at onset of stage 2. Background color: Stage 1 dark blue, stage 2 light blue.

View larger version (16K): [in this window] [in a new window]   Fig. 8. Relation among kinematic stages 1 and 2 of archer fish fast-starts. Red circles relate to predictive starts, blue circles relate to escape C-starts elicited in the same group of fish. (A) Duration of stage 2 versus duration of stage 1. Slope and intercept of regression lines (predictive starts, red: r2=0.319, P<0.0001, N=76; escapes, blue: r2=0.120, P<0.05, N=34) were not different between escapes and predictive starts (P>0.05). (B) Relation of straightening rate in stage 2 (maximum of dCL/dt; CL=dimensionless chord length) to rate of bending in stage 1 (maximum of -dCL/dt). Derivatives are taken from a linear regression analysis around the point of maximum changes, using at least 5 neighboring points. Slope and intercept of regression lines (predictive starts, red: r2=0.288, P<0.0001, N=76; escapes, blue: r2=0.418, P<0.0001, N=34) were not different between escapes and predictive starts (P>0.05).

View larger version (7K): [in this window] [in a new window]   Fig. 9. Matching the fast-start to the required angle of turning. The best predictor of turn size in predictive starts (red circles) as well as in escapes (blue circles) was how much the fish's body was bent by the end of stage 1. This is quantified here as the minimum chord length (CL) achieved (see Fig. 4 for introduction of CL). Slope and intercept of the regression lines (predictive starts, red: r2=0.749, P<0.0001, N=76; escapes, blue: r2=0.473, P<0.0001, N=34) were not different between escapes and predictive starts (P>0.05).

View larger version (8K): [in this window] [in a new window]   Fig. 10. Matching the fast-start to the desired take-off speed. For both predictive starts (red) and escapes (blue) take-off speed (defined as average speed within the first 20 ms subsequent to end of stage 2) correlated best with the rate of straightening during stage 2. This rate is taken as the maximum slope dCL/dt in stage 2. Slope and intercept of the regression lines (predictive starts: r2=0.216, P<0.0001, N=76; escapes: r2=0.484, P<0.0001, N=34) did not differ significantly between escapes and predictive starts (P>0.05).

View larger version (21K): [in this window] [in a new window]   Fig. 11. The spectrum of kinematic variations among archer fish C-type escape starts. A large set of responses (N=34, black lines) from the same group used in characterizing the predictive starts was analyzed in detail using digital high-speed video (500 frames s-1). The substantial variability is illustrated by showing all courses of chord length changes (CL; A), accumulated angular changes (; B) and of accumulated displacement (s; C). Time zero (dotted vertical line) is set at onset of stage 2. Variables and how they deviate for wide and narrow turns as well as for different take-off speed are introduced in Figs 4, 5, 6. The courses of the predictive starts recorded in the same group (Fig. 7) are shown for comparison (gray lines).

View larger version (11K): [in this window] [in a new window]   Fig. 12. Comparison of (A) a predictive start with (B) an escape C-start. Both starts involved similar turns and were imaged at 500 frames s-1; every second frame is shown. Color of background behind fish silhouettes highlights initial bending phase (stage 1; dark blue) and propulsive phase (stage 2; light blue). An analysis of the predictive start (A) is reported in Figs 4 and 5. Take-off speed was 1.015 m s-1 (A) and 0.595 m s-1 (B) and turn size was 82.1° (A) and 99.7° (B).

View larger version (31K): [in this window] [in a new window]   Fig. 13. Archer fish C-type escapes and predictive starts follow an identical temporal pattern. The histograms show how the timing of the maximum speed and acceleration attained was distributed among the predictive starts (red columns) and the escapes (blue columns). Time zero is at onset of stage 2. Distribution of timing of (A) maximum linear speed, (B) maximum linear acceleration, (C) maximum angular speed, and (D) maximum angular acceleration. A MANOVA detects no significant difference among these timing parameters (P=0.393). Respective bin sizes are: 10 ms (A,B), 2 ms (C), 5 ms (D). Total counts in the respective histograms were for the predictive starts N=69 (A), N=59 (B), N=72 (C), N=68 (D), and for the escapes N=33 (A), N=30 (B), N=31 (C), N=31 (D).