Files in this Data Supplement:

• Supplemental Figure S1 -

  Fig. S1. Locomotor activity after placement on testing arena. In order to determine how much time is necessary to allow larvae to achieve a steady level of motor initiations after movement onto the testing arena, 400 ms video recordings were made every 30 s for 10 min after placement under the camera. After 3 min, both turn initiations (A) and scoot initiations (B) achieve steady levels, suggesting that this is a sufficient amount of time to allow for 'recovery'.

• Supplemental Figure S2 -

  Fig. S2. Measuring position and curvature of larvae. (A) For each larva, a head position is determined by applying the bandpass and feature-finding algorithms of Crocker and Grier (Crocker and Grier, 1996) to find the sub-pixel location of the centroid of the head. (B) As the centroid position is most influenced by eye density, this method is robust despite individual variation in pigmentation patterns. (C) Centroid location for a single larva during 100 ms of intense movement (a C-bend). Over 100 frames, the position of this centroid was 127±28 μm (mean ± s.d.) caudal to the midpoint of the eyes and 20±25 μm horizontally displaced from the midline. Scale bar, 100 μm. (D) Measurement of body curvature and eye position for a larva engaged in a high angle bend. Colored bars show the three body segments measured and white dots show the eye centroid positions. This procedure for measuring larval curvature remains reliable at high bend angles despite low resolution of individual fish. (E) Comparison of body curvature (blue), head orientation (green) and tail orientation (red) over 150 ms for a scoot (i) and a turn (ii). Head and tail orientations are plotted relative to the initial orientation. Note that for scoots, the initial change in body curvature is entirely derived from tail movements, with the head remaining stable. In comparison, during turns, orientation changes are seen for both head and tail segments.