314 



CHEMICAL SENSES 



The concept was applied to nurse shark data before and during chemical 

 stimulation, acquired by monitoring the locomotor behavior with the square 

 matrix of photocells (Monitor II) (Matis, Kleerekoper, and Childers 1974). 

 By the technique outlined above, two time series (mean velocity and mean 

 step length) of 15-min intervals were obtained for a 17-h control period. 

 Data from the initial, distinct, "exploratory" period were deleted. Gerald et 

 al. (in press), in the analyses referred to above, using an improved model, 

 applied the same concept to 17 time series of locomotor variables, of which 

 Figure 31 illustrates the time series of mean velocity. The data on the left 

 side of the graph, up to the arrow, refer to the control period of almost 20 h 

 and were used to identify and estimate time series models for that period by 

 the Box-Jenkins techniques (Box and Jenkins 1970), an approach necessary 

 because of the distinct time series structure in the locomotor behavior of the 

 nurse shark (Matis, Kleerekoper, and Gruber 1975), as mentioned earlier. 



By the above procedure, discussed in detail in the original reports, a 

 significant response by the nurse shark to a chemical stimulus applied at a 



30 



90% 

 > CONFIDENCE 

 LIMITS 







10 

 TIME (hrs) 



15 



20 



Figure 31 Ginglymostoma cirratum : time series of average veloc- 

 ities (cm/s). On the left side of the arrow is the control period 

 of 16.5 h, after which chemical stimulation was begun through 

 needle 6. The horizontally oriented curves indicate 90% con- 

 fidence limits of the forecast. At the point which stimulation is 

 begun the locomotor variable immediately "escapes from its 

 control"; the values of its time series greatly exceed the lower 

 confidence limit for several hours. From Gerald, Matis, and 

 Kleerekoper (in press). 



