BLAYLOCK: DISTRIBUTION OF THE BOTTLENOSE DOLPHIN 



1.4 



I ,2 



O 



LU 

 C/5 



z 

 < 



tr 



O 



DC 

 U. 



LU 

 O 



CO 

 Q 



10 



0- 



10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 

 HERD SIZE 



Figure 3.— Scatterplot of sighting distance (in kilometers) versus herd 

 size of bottlenose dolphins. (Note that some of the points in the 1-10 size 

 class and the interval 0-0.2 km represent more than one herd.) 



Table 1 .—Models used in calculation of the detection function g{x) 

 for bottlenose dolphin. N is the number of terms used in the model 

 and f(0) is f(x) evaluated at the transect. SE[f(0)] was calculated 

 assuming asymptotic normality with Z = 1.96. 



of variation of the 3-term Hermite polynomial was 

 slightly less than that of the 4-term (0.154 vs. 0.168), 

 but the 4-term Hermite polynomial model provided 

 a better fit to the observed perpendicular sighting 



distance distribution than either the 3-term Hermite 

 polynomial or the 4-term Fourier series models 

 (Table 2). 



The appropriate model for the observed perpen- 

 dicular sighting distances should fit the data most 

 closely near the centerline of the transect (Schweder 

 1977). The 4-term Hermite polynomial model closely 

 approximates the observed sighting distances in the 

 interval 0-200 m and also in the subsequent inter- 

 vals (Fig. 4). 



The nonparametric 4-term Hermite polynomial 

 model yielded an estimate of /(O) = 3.104 (SE = 

 0.522). Dolphin density in the southern Virginia 

 coastal area (3.446 dolphins/km-) was much greater 

 than that in the CBM area (0.159 bottlenose dol- 

 phins/km^); however, the abundance estimates are 

 of similar magnitude (219 vs. 121 bottlenose dol- 

 phins, respectively) due to the greater area sampled 



Table 2.— Observed and expected distribution of bottlenose dolphin herd sightings by distance in- 

 tervals perpendicular to the transect (PSD in meters) with chi-square values. Figures in parentheses 

 are degrees of freedom. Expected values are rounded to one decimal place for clarity of presen- 

 tation. 



801 



