POLACHECK: DISTRIBUTION OF SEARCHING EFFORT 



the quantitative results in this paper. He stated 

 that fishermen perceive two kinds of areas: The 

 first, in which fishermen refer to their activity as 

 "scratching", are regions where isolated schools of 

 fish are encountered; the second, which fishermen 

 refer to as an "area", are portions of the ocean 

 where schools of tuna are congregated. He also 

 reported that isolated tuna schools tend to be 

 small. Fishermen prefer to concentrate in an 

 "area" and actively search for them. I did not be- 

 come aware of this study by Orbach until the final 

 calculations of this paper were completed. Thus, 

 his description is an independent indication that 

 the results are not an artifact of the clustering 

 algorithm. 



The method used for estimating the location 

 and size of a cluster could be refined. Such refine- 

 ments were beyond the scope of the present work 

 and would not affect any of the main results. 

 However, such refinements might be important if 

 the method used for calculating the adjusted en- 

 counter rates (Table 7) was used to develop rela- 

 tive abundance indices for dolphins in the ETP. 

 As pointed out by one reviewer, the problem of 

 estimating the shape, size, and location of a clus- 



ter is analogous to the problem of determining the 

 home range for an animal from a set of observed 

 positions over time (Sanderson 1966; Cooper 

 1978; Schoener 1981; Swihart and Slade 1985) 

 and is part of the more general problem of how to 

 estimate the limits and size of clusters from any 

 clustering algorithm. The methods used for esti- 

 mating home ranges cannot be directly applied to 

 the tuna boat observer data but might provide a 

 basis for developing a better estimator for these 

 statistics relating to cluster size in any extension 

 to the present work. 



To the extent that the results from this paper 

 indicate that searching is clustered, they suggest 

 that when these data are used for estimating rel- 

 ative densities of dolphin schools, or when catch 

 and effort data from purse seiners are used to 

 assess tuna stocks, rather fine geographic stratifi- 

 cations are needed to avoid biases from the non- 

 random searching within a cruise. The estimates 

 of the cluster radius could be considered as a 

 guide to appropriate levels of stratification. For 

 example, from 31 to 46% of all clusters with at 

 least three members had a radius of less than 60 

 miles (Fig. 11). This suggests that at a minimum, 



50 r 



40- 



5 30 

 Ui 



u 



oc 

 111 

 Q. 20 



10 - 



50 



40 - 



30 - 



20 - 



30 



60 



90 



120 



150 



180 >180 



30 



60 



90 



120 



150 



180 >180 



40 



30 



lU 



O 20 

 c 



10 



C 



N=214 



X 



_L 



X 



X 



_r 



30 60 90 120 150 180 >180 



CLUSTER RADIUS (nautical mllaa) 



40r- 



30- 



20 - 



10 ~ 



30 60 90 120 150 180 >180 



CLUSTER RADIUS (nautical miles) 



Figure 11. — The frequency distribution for estimated radii for all clusters with at least three members. Estimates from 

 all cruises have been pooled. The values of the clustering parameter are A) 50, B) 75, C) 100, and D) 150. 



365 



