Summary

Many amphibian tadpoles hatch and swim before their inner ears and sense of spatial orientation differentiate. We describe upward and downward swimming responses in hatchling Xenopus laevis tadpoles from stages 32 to 37/38 in which the body rotates about its longitudinal axis. Tadpoles are heavier than water and, if touched while lying on the substratum, they reliably swim upwards, often in a tight spiral. This response has been observed using stroboscopic photography and high-speed video recordings. The sense of the spiral is not fixed for individual tadpoles. In ‘more horizontal swimming’ (i.e. in directions within +/−30 degrees of the horizontal), the tadpoles usually swim belly-down, but this position is not a prerequisite for subsequent upward spiral swimming. Newly hatched tadpoles spend 99 % of their time hanging tail-down from mucus secreted by a cement gland on the head. When suspended in mid-water by a mucus strand, tadpoles from stage 31 to 37/38 tend to swim spirally down when touched on the head and up when touched on the tail. The three-dimensional swimming paths of stage 33/34 tadpoles were plotted using simultaneous video images recorded from the side and from above. Tadpoles spiralled for 70 % of the swimming time, and the probability of spiralling increased to 1 as swim path angles became more vertical. Tadpoles were neutrally buoyant in Percoll/water mixtures at 1.05 g cm(−)(3), in which anaesthetised tadpoles floated belly-down and head-up at 30 degrees. In water, their centre of mass was ventral to the muscles in the yolk mass. A simple mathematical model suggests that the orientation of tadpoles during swimming is governed by the action of two torques, one of which raises the head (i.e. increases the pitch) and the other rotates (rolls) the body. Consequently, tadpoles (i) swim belly-down when the body is approximately horizontal because the body is ballasted by dense yolk, and (ii) swim spirally at more vertical orientations when the ballasting no longer stabilises orientation. Measurements in tethered tadpoles show that dorsal body flexion, which could produce a dorsal pitch torque, is present during swimming and increases with tailbeat frequency. We discuss how much of the tadpole's behaviour can be explained by our mathematical model and suggest that, at this stage of development, oriented swimming responses may depend on simple touch reflexes, the organisation of the muscles and physical features of the body, rather than on vestibular reflexes.