306 CHEMICAL SENSES 



quadrants 1 and 2. At the first stimulation through needle 1, quadrant 4, the 

 shark responded immediately by restricting its movements almost entirely to 

 that quadrant. During the following control phase (no stimulation), the pre- 

 vious upstream activity was gradually resumed. Throughout the remainder of 

 the experiments, the alternation between periods of stimulation and control 

 produced statistically highly significant (a = 0.001) responses, which, though 

 similar, were quantitatively different as a function of the location of the 

 stimulus. 



The relationship between the percent of distance traveled and the six 

 sources of stimulation was determined by regression analysis and modeled by 

 a polynomial regression of up to second order. All the resulting regressions 

 (Figure 25) were significant (a = 0.05) and quadratic except for the linear 

 equation of quadrant 2. It is important to note that stimulation of site 1 (the 

 most downstream location), in quadrant 4, resulted in nearly 90% of the 

 distance covered by the shark to be in that quadrant. As the site of stimula- 

 tion moved upstream, the distance covered in quadrant 4 decreased rapidly 

 and that in upstream quadrants 1 and 2 (and to a much smaller degree in 

 quadrant 3) increased. 



The relationship between the response in direction vector and the site of 

 the stimulation is of particular interest. As illustrated in Figure 26, release at 

 site 1 (the most downstream location) resulted in a low directionality meas- 

 ure in each of the quadrants. As the stimulation site moved upstream, the 

 statistic increased in the two upstream quadrants, with the response in quad- 

 rant 2 exceeding that in quadrant 1 for all sites. No such increase occurred in 

 the downstream quadrants. By applying the above analytical procedures to 

 the- data recorded from the experiments in stagnant water, the regressions 

 illustrated in Figures 27 and 28 were obtained. In interpreting the above 

 results, it should be remembered that the analyses refer to localization by 

 the shark of the quadrants in which the release sites are situated, and not to 

 the exact positions of those sites. The results, therefore, demonstrate the 

 general localization of the attractant stimulus rather than the exact localiza- 

 tion of its source, which will be discussed later. 



The analyses showed that of all the variables studied, distance traveled, 

 direction vector, and turning frequency are the most valuable in the analysis 

 of the locomotor behavior involved in general localization. In flowing water, 

 the regressions of percent of distance traveled demonstrate a fine distinction 

 between adjacent release sites in downstream quadrant 4 (sites 1, 2, and 3), 

 although some clear discrimination occurred also in upstream quadrant 1, 

 which contained sites 4, 5, and 6. 



The behavior of the direction vector further demonstrates the precision of 

 the localization. A tight swimming pattern in the quadrant of stimulation 

 (site 1, quadrant 4) is characterized by low vector values accompanied by 

 high values for distance traveled (Figures 25 and 26). As the sites of stimula- 

 tion move upstream, the locomotor pattern loosens, until at site 6 the pat- 

 tern spreads beyond quadrant 1. The fact that the regression curves of quad- 

 rants 1 and 4 are not mirror images demonstrates that precise localization of 

 the stimulus source is not exclusively dependent on the strength, rate of 



