FISHERY BULLETIN: VOL. 78. NO. 3 



Table 2. 



-Numbers of Halobates spp. caught in the EASTROPAC area. Roman num- 

 erals = nymphal instar number. 



23.3'^, respectively), we will confine further dis- 

 cussions of abundance to these two species. 



In an attempt to determine if H. micans and H. 

 sobrinus are randomly distributed on the ocean 

 surface, curves of "number per tow" vs. "number of 

 tows with that number of insects" were compared 

 with Poisson probability density functions. Such 

 tests are appropriate for these data (Sokal and 

 Rohlf 1969) but difficult to perform because of un- 

 certainty as to how many "zero" catches should be 

 included in the divisor when calculating a mean 

 catch-per-tow Maximum likelihood estimates of 

 the means for truncated Poisson distributions ( i.e., 

 lacking a zero class) were therefore calculated for 

 both species (Cohen 1960), and the frequency dis- 

 tributions of Figure 8 were compared with the 

 calculated (expected) Poisson distribution. Chi- 

 square tests of expected vs. observed were very 

 highly significant for both species (P<<0.001), 

 leading to the rejection of the null hypothesis that 

 observed distribution cannot be told from a Pois- 

 son. We therefore conclude that both H. micans 

 and H. sobrinus are nonrandomly distributed 



across the ocean surface in the regions where they 

 occur. 



The coefficient of disperson is^/x, Sokal and 

 Rohlf 1969) for H. sobrinus is —46.0 and for H. 

 micans is =5.8. We thus conclude that both species 

 are very strongly clumped ("patchy"),//, sobrinus 

 more so than H. micans. The numbers per sam- 

 ple also vary widely, e.g., from to 179 for H. 

 sobrinus. It is not known what environmental fac- 

 tors cause one location to provide higher catches 

 than another. Assuming optimum sampling condi- 

 tions (perfect net performance, etc.), the highest 

 population densities calculatable for each of the 

 four species are presented in Table 3. These are 

 lower limits because of probable net avoidance 



Table 3. — Highest estimates of population density for Halo- 

 bates spp. in the EASTROPAC area. 



Figure 8. — Frequency distribution of 

 Halobates micans and H. sobrinus in 

 positive samples. The number of insects 

 caught per tow is plotted against the 

 number of tows containing that many 

 insects. "Zero catches" are excluded (see 

 text). 



H micans 



8 12 16 20 24 28 32 36 40 

 INDIVIDUALS /SAMPLE 



H sobrinus 



r«-^V-f 



20 24 



28 32 36 40 44 48 52 

 INDIVIDUALS /SAMPLE 



56 60 70 80 100 140 ISO 



588 



