FISHERY BULLETIN: VOL. 86, NO. 2 



Whenever it seemed important to display the re- 

 sults by cruise, a value of 100 miles for the clus- 

 tering parameter was used. It should be empha- 

 sized that the general patterns appear to be 

 independent of the exact value of the clustering 

 parameter. 



Is Searching DiflFerent in Clusters? 



To determine whether vessels tended to in- 

 crease their searching effort in the vicinity of 

 sets, the ratio of the total distance travelled (i.e., 

 the distance searched plus the distance run) to 

 the actual physical distances between sets was 

 compared between and within clusters. This ratio 

 is about 1.7 times greater in clusters than be- 

 tween clusters (Table 3). Also for no more than 7 



cruises is the ratio within clusters less than the 

 ratio between clusters (Table 3). By a sign test 

 (Snedecor and Cochran 1967), these results imply 

 that this ratio is significantly greater than 1.00 

 (P < 0.005). 



In addition, the proportion of the total distance 

 travelled that is devoted to searching is much 

 greater when a vessel is between than when it is 

 within clusters (Table 4). The fact that vessels 

 run, and are not actively looking for tuna, propor- 

 tionately more between than within clusters is an 

 indication that clusters are areas in which a ves- 

 sel has decided to remain. Vessels tend only to 

 run at night. Usually, as long as the vessel is 

 moving during daylight, the crew will be search- 

 ing. Thus, large amounts of running tend to occur 

 when vessels are actively moving to new areas. 



Table 3. — Comparison between and within clusters of the ratio of the total distance 

 travelled (i.e., distance searched plus distance run) to the actual physical distance 

 between sets and chases. Means are the average values of the ratio for the 35 cruises 

 being considered. 



Value of the cluster parameter 



ip < 0.05 of observing this many ratio greater within clusters than between if they were in fact 

 equal based on a sign test (Snedecor and Cochran 1967). 



Table 4. — Comparison between and within clusters of the ratio of the distance run without 

 searching to the distance searched. Means are the average values of this proportion for the 

 35 cruises being considered. 



ip < 001 of observing the portion being greater between clusters than within 34 out of 35 cruises, if 

 in fact the proportion was equal based on a sign test (Snedecor and Cochran 1 967). 



360 



