Despite the smalliiess of the herds, it was not 

 easy to accurately count animals that were some- 

 times spread over a relatively large area, and in 

 subgroups that only showed for brief periods at the 

 surface. Obviously, accuracy of such counts will 

 also improve with experience. However, by scor- 

 ing a consensus opinion the judgment and bias of 

 the most experienced observer probably carried 

 more weight, and as a result we feel that in all 

 cases the counts were conservative. Because of the 

 experience factor we also think that, other 

 influences being equal, the latter surveys were 

 probably the more accurate. 



Last, one factor, the "gerrymandered" lines of 

 transects 1 and 2, clearly tended to influence the 

 counts upward. Our rationale for altering the line 

 of these transects was based on the desirability of 

 obtaining data in an area for which baseline in- 

 formation already was available (Shane 1977). 

 Unfortunately, the terrain was not ideal for tran- 

 sect sampling, and flying an east-west line over 

 the ship channels would have resulted in gross 

 underestimation of an area known to hold a rela- 

 tively large number of dolphins. 



Clearly, the results for transects 1 and 2 (23"^"^ of 

 the animals sighted in only 6.6'J of the total area) 

 were strikingly different from those data for the 

 rest of the tran.sects. Estimated dolphin density for 

 the ship channels was 2.6.33/km^, some 4.25 times 

 greater than the 0.619/km''^ estimated for tran- 

 sects 3-21 (Table 4). Based on these densities the 

 total population estimate could be partitioned into 

 304 dolphins for the ship channels and 1,015 ani- 



FISHERY BULLETIN: VOL. 77. NO 3 



mals in the rest of the area. Shane's (1977) 

 maximum estimate for the ship channel area for 

 any month of the year was about 280, thus the two 

 estimates are in reasonable agi'eement. We still 

 feel, however, that there were some unresolvable 

 problems with our survey methodology as it 

 applied to the Aransas Pass ship channels, and 

 that the soundest procedure was to lump the re- 

 sults from the minority area with those from the 

 major region, as we have done. 



Alternative Density Estimate 



As previously discussed, the decrease in the 

 number of dolphin sightings at increasing ranges 

 of the herds from the flight path (Figure 5) indi- 

 cated violation of strip transect theory assumption 

 that all herds within the delineated area were 

 sighted. Line transect theory (Seber 1973) pro- 

 vided an alternative method of analyzing the re- 

 sults. Because there were few observations in the 

 0-50 m increment, creating a marked gap in the 

 frequency distribution, and the "tail" of the fre- 

 quency distribution was truncated, in part be- 

 cause we limited observations to about 400 m 

 range, our data were not strictly applicable to line 

 transect theory, either. Despite these discrepan- 

 cies, however, we obtained for comparative pur- 

 poses a rough approximation of the level of bias by 

 applying a simple modification of the so-called ex- 

 ponential estimator (Gates et al. 1968) which cor- 

 rects for the gap in the 0-50 m frequency distribu- 

 tion interval as follows: 



Table 4.— The basic terms and figures for comparing the estimated bottlenose dolphin density i n two parts of the 

 survey, the Port Aransas ship channels (transects 1 and 2) and rest of the area (transects 3 to 21). 



'From Equation (13) 



594 



