302 CHEMICAL SENSES 



migration, search for food, mate, or chemical quality of the medium, the 

 animal would swim upstream as long as the chemical identity of the rheo- 

 taxis-releasing odor (single substance or mixture) was perceived. 



Because in such an orientation mechanism odor molecule densities are 

 irrelevant as long as the threshold for olfactory stimulation is reached, the 

 theoretical difficulties inherent in the "gradient search" hypotheses, pointed 

 out by several investigators (von Buddenbrock 1952; Precht 1942; Otto 

 1951), are eliminated. In view of the significance such a mechanism may 

 have in the orientation behavior of elasmobranchs, a series of experiments 

 was designed to verify the role of interaction between olfaction and water 

 flow in the ability of these animals to localize an attractant odor source, and 

 to determine some of the quantitative relationships between these stimuli. 



LOCALIZATION OF AN ODOR SOURCE BY 

 GINGLYMOSTOMA AS A FUNCTION OF WATER FLOW 



In this study (Kleerekoper, Matis, and Gruber 1975), the locomotion of 

 single sharks was analyzed, by means of monitor II (square matrix of sen- 

 sors), during 24-h control periods, in both flowing (1.17 cm/s) and stagnant 

 water. Following this, and without interrupting monitoring, the chemical 

 stimulus (dilute, filtered extract of mashed shrimp) was applied through 

 needle no. 1 (Figure 22) embedded in the monitor's floor, at a constant rate 

 (5.0 ml/min), during 2 h. At the end of this period, delivery was stopped 

 while, in the experiments with flowing water, water continued to flow 

 through the tank during at least 3 h; in the experiments with stagnant water, 

 circulation was reinstated to allow filtration and to eliminate shrimp sub- 

 stance from the tank. This procedure was repeated five times with five 

 different needles, as illustrated in Figure 22, which also indicates the direc- 

 tion of water flow, when present, and the quadrants of the tanks. 



For an understanding of the results of these experiments, the following 

 explanation seems pertinent. The monitor system yields, it may be re- 

 membered, a time series record of photocells, in a 44 X 44 matrix, triggered 

 by the shark in the course of its locomotion. Data from each quadrant of the 

 tank were analyzed separately, converted into 15-min intervals, and trans- 

 formed into seven locomotor variables: (1) number of events (photocell 

 triggerings), (2) distance traveled, (3) mean velocity, (4) mean turn size, (5) 

 mean step length, (6) time spent in the quadrant, and (7) frequency of 

 turning. 



These data were then standardized. A distribution of the orientation 

 angles of the steps (or straight pathways between turns) in each quadrant, for 

 each period, was characterized by an angular mean vector with direction 6 

 and strength r, adjusted for activity level. The resulting direction vector 

 index was the eighth time series in the analysis of the response in each 

 quadrant. Polynomial regression equations were calculated for each experi- 

 ment, relating the mean response of each variable to the six nearly equi- 

 distant sources of stimulation to determine those variables, by quadrant, that 



