interfaced with an aircraft LORAN system. A data acquisition program downloaded the 

 time and date, and the latitude, longitude, speed and heading of the aircraft whenever 

 sighting data was entered. The observers rotated positions about every 30 minutes. 



Observers and the flight crew commimicated through headsets via the aircraft 

 intercom system. Observers searched for marine mammals, sea turtles and other marine life 

 at the surface of the water from directly beneath the aircraft out to a perpendicular distance 

 of 629 m. Whenever a sighting was made, the distance of the sighting from the transect hne 

 was measured using calibrated marks delineating 7 perpendicular distance categories on each 

 bubble window (40, 83, 132, 192, 273, 397, 629 m). When necessary, the aircraft was diverted 

 from the transect line to make species identifications and to estimate marine mammal herd 

 sizes. 



Density Estimation 



Bumham et al. (1980) recommended that sighting functions should be based on a 

 minimum of 40 sightings, but stated 60-80 sightings were preferable. However, White et al. 

 (1989) suggest that over 200 sightings may be required. Because only 94 cetacean herds 

 were sighted (91 were bottlenose dolphin herds) within 629 m of the transect line during 

 both surveys, the perpendicular distance sighting data were pooled with 1,523 bottlenose 

 dolphin herd sightings coUected by the same survey team in the northern Gulf of Mexico 

 from the same aircraft. These pooled data were used to construct a sighting histogram. To 

 estimate !(0), the value of the probability density function evaluated at the transect line, a 

 hazard-rate model (Buckland 1985) was fit to the histogram. The hazard-rate model was 

 selected for two reasons: (1) the number of parameters in the model is fixed (there was no 

 subjective decision making regarding the number of parameters), and (2) the model always 

 has a shoulder near the transect line (distance zero). 



Bottlenose dolphin density for each survey block was estimated as the product of a herd 

 density estimate and a estimate of mean herd size. Herd density was estimated separately 

 for each survey block. In order to increase sample sizes and reduce variability of mean herd 

 size estimates, all herds sighted in inshore blocks (152, 153, 154) were pooled as were all 

 herds sighted in offshore blocks (054, B). [Bottlenose dolphin herds in the Gulf of Mexico 

 may increase in size in deeper water.] Of the bottlenose dolphin herds sighted during both 

 surveys, 89% were of 10 dolphins or less. However, three herds were sighted that were 

 greater than 40. Because of relatively small sample sizes, these large herds had a 

 tremendous influence on the means and substantially increased variability. Therefore data 

 were trimmed from each end of the herd size distributions until the means stabilized. This 

 generally occurred after a total of 15% of the data were excluded. The "trimmed" mean 

 herd size for offshore and inshore blocks was estimated as the arithmetic mean. 



105 



