the tern flocks and in NTEX herring gulls were prominent in some. The SFLA flocks 

 were feeding over schools of baitfish that were rippling the surface, while some of the 

 NTEX flocks gathered around fishing boats. 



On 15 November in the NFLA subunit, a dusky shark (Carcharhinus obscurus) and 

 six bottlenose dolphins were observed attacking a 10- x 20-m school of mullet (Mugil sp.) 

 approximately 25 km offshore. The porpoises were feeding along the trailing edge of the 

 school while the shark cut through at right angles. No interaction was observed between 

 the shark and porpoises. 



Smaller aggregations involving frigatebirds and other bird species were frequently 

 encountered in proximity to dolphins. The significance of such associations is not 

 apparent at present. Additional data are needed on biological associations before 

 assessing their roles in marine systems. 



DENSITY ESTIMATES 



To estimate densities of individual animal taxa, the data were examined using the 

 Linetran statistical package. For purposes of the Pilot Study, which is subject to data 

 limitations detailed previously, density calculations have been limited to the taxa most 

 commonly sighted within a single survey subunit and season. Calculations for other taxa 

 can be made in subsequent studies by combining data from adjacent areas or proximal 

 seasonal samples. 



The Linetran statistical package includes both parametric and nonparametric 

 estimators. For the Pilot Study data, only the nonparametric estimators are appropriate 

 due to the reduced visibility of the area immediately below the survey aircraft (i.e., the 

 transect line). This limitation will not be operant for data collected from an aircraft 

 with maximal forward and downward visibility allowing inspection of transect line 

 (Beechcraft AT-11 or similar aircraft). 



Estimates were calculated using the ungrouped splined method of Gates (1979). 

 This technique attempts to compensate for the truncated data set resulting from the 

 blind area beneath the aircraft, and for the variable distribution of sightings in relation 

 to the transect line. The algorithm was selected after consideration of 14 estimators 

 available in the Linetran package. 



The maximum 'perpendicular distance for sightings in NFLA varied from 234 to 

 1,326 m with a mean for all species of 686 m. This is judged to be an overestimate of this 

 distance due to the numerous sightings of birds above the water's surface, which affects 

 the sighting angle used to compute distance. 



The maximum observable distance for each taxon is important to density 

 calculations. For inconspicuous species the effective observation distance may be 

 limited to 200 m, whereas for others it may exceed 1 km. In line transect computations, 

 as opposed to strip transect methods, the area considered to have been surveyed is 

 relative to the distance at which sightings were made. Consequently, the percent 

 coverage of the subunit varies with the observability of the taxa. For example, the 

 maximum observation distances for bottlenose dolphins, royal terns, and unidentified 

 terns in August-NFLA flights were 279, 872, and 1,326 m, respectively. These distances 



58 



