FISHERY BULLETIN: VOL. 84, NO. 2 



7000 



6000 



5000 



4000 



3000 



2000 



1000 



1 = Winter (Jan., Feb., March) 



2 = Spring (April, May, June) 



3 - Summer (July, Aug., Sept.) 



4 = Autumn (Oct., Nov., Dec.) 



Figure 2.— Comparison of total counts of common dolphins on aerial surveys of the 

 Southern California Bight by season, 1975-78. 



ment of the subject, in print, is found in two papers 

 by Leatherwood et al. (1982, 1983). 



Because sample size was small in each grid-cell 

 and in each season, data were combined to calculate 

 a single value of f(0). The pooling of data was based 

 on the assumption that the sightability of common 

 dolphin groups did not vary between seasons or be- 

 tween regions of the surveyed area. Violation of this 

 assumption would lead to biases in the estimates of 

 relative densities between seasons or regions, al- 

 though it would not necessarily effect mean popula- 

 tion size estimates. The assumption of seasonal 

 homogeneity was tested using a single classification 

 ANOVA (two groups, unequal samples; Sokal and 

 Rohlf 1969, p. 208). No significant difference be- 

 tween the distribution of perpendicular sighting 

 distances collected in summer-autumn and winter- 

 spring seasons was found (F 1(111 = 2.01, P = 0.18). 

 The same test was used to compare frequency 

 distributions with distance of sightings collected in 



calmer inshore waters, with sightings collected in 

 rougher offshore waters, since this seemed to be the 

 most likely source of bias in sightability. No signifi- 

 cant difference was found between the distribution 

 of perpendicular sighting distances in the two sub- 

 regions (F 1>108 = 1.78, P = 0.20). 



The rescaled frequency distribution of perpendi- 

 cular sighting distance is shown in Figure 3. The 

 probability density function, f(x), is from a three- 

 term, Fourier-series model, which provides the best 

 fit to these data (x 2 = 6.026, df = 3, P = 0.11). 

 Data were truncated at 6,600 ft in order to remove 

 two extreme values. Intervals were specified, by in- 

 spection of the data, in order to smooth the func- 

 tion and minimize the effects of "heaping" in per- 

 pendicular distance measurements (Burnham et al. 

 1980, p. 47). 



For estimation of common dolphin density (ani- 

 mals/km 2 ) in a given grid-cell for a given season, 

 we multiplied the density of groups in a given cell 



336 



