Abstract

In an approach to quantify the locomotor response to environmental stimuli in fishes and its central control mechanisms, initially stochastic models of spontaneous locomotor behavior are being formulated. In the present paper, the locomotor patterns of three active nurse shark,Ginglymostoma cirratum, in six experiments are converted into 17 locomotor variables and found to have definite time series structure. Sixty-seven of the 102 first order serial correlation coefficients are statistically significant, the incidence rate of which differs between experiments and between locomotor variables. The percent residual variation about an autoregressive equation is transformed by a Fisher's z statistic and the means are found to differ significantly among experiments and among variables. The experiment by variable interaction is very low implying a consistency in the ranks of the variables despite the wide differences in their absolute levels over the experiments. This suggests the existence of a relatively rigid albeit occasionally latent locomotor control mechanism in the central nervous system