FISHERY BULLETIN: VOL 86, NO. 2 



spatially clustered. Perhaps the most surprising 

 result in this context is that the average catch per 

 set tends to be greater within the defined clusters 

 than outside of them. There is nothing in the clus- 

 tering algorithm that would tend to produce this 

 result. If these larger catches per set reflect large 

 schools of tuna, this argues that these defined 

 clusters are areas where tuna tend to concentrate 

 and not just areas of high densities of sets that 

 could be found in any random distribution. 



Within many of the cruises, the locations of two 

 or more of the defined clusters overlap spatially 

 (Fig. 10), indicating that vessels often return to 



an area after a period of searching elsewhere. In 

 addition, clusters from different cruises overlap 

 spatially and temporally. This overlapping of 

 clusters indicates that the overall searching be- 

 havior is even more nonrandom than the results 

 from this paper suggest. The overlapping both 

 within and among cruises is a dimension that 

 should be considered in future extensions to the 

 present work. 



Orbach (1975), in a nonquantitative, anthropo- 

 logical study of the purse seine fishery, included 

 a general qualitative description of the searching 

 behavior of the fishermen which supports many of 



Figure 10. — Examples of the spatial relation between clusters for the two cruises depicted in Figure 2. Open circles indicate the 

 position of all clusters with at least three members. The radius of each circle is scaled to the estimated radius of a cluster. The 

 associated numbers are the number of sets and chases within a cluster. Solid circles indicate the position of clusters with only two 

 members. Isolated sets are indicated by an x. The value of the clustering parameter equals 100. Arrows indicate the order of 

 movement between clusters. Note A and B are drawn to different scales. Distance between tick msirks equals 300 miles. No geographic 

 coordinates are given and the orientations were rotated at random so as not' to compromise any proprietary fishing information. 



364 



