FISHERY BULLETIN: VOL. 8«. NO. 4 



waters.^ If the high level of Tursiops mortalities ex- 

 perienced in the mid-Atlantic coast during the sum- 

 mer of 1987 was because of an infectious agent, then 

 its spread to conspecifics in more southerly regions 

 may have been caused by contact between in- 

 dividuals from different areas and more extensive 

 migration than has been previously suggested. 



In the present study I used aerial surveys to esti- 

 mate the abundance and examine the distribution 

 of T. truncatus in Virginia coastal waters, including 

 the Chesapeake Bay mouth. I also investigated 

 natality periods by monthly comparison of the aver- 

 age percentage of calves present and residency pat- 

 terns using photographic records of identifiable 

 individuals. 



METHODS 



Aerial surveys were conducted during July- 

 October 1980, and May-June 1981, from a high- 

 winged, single-engine aircraft (U6A DeHavilland 

 Beaver^) at an altitude of 152 m and at an air- 

 speed of 147 km/h. Observers sitting in the two 

 passenger seats searched each side of the transect 

 for bottlenose dolphins. A recorder/navigator sitting 

 forward of the observers and next to the pilot helped 

 to maintain predetermined transect lines and 

 recorded sightings which were communicated via 

 intercom. 



Upon sighting a bottlenose dolphin herd, the per- 

 pendicular distance from the flight path to the herd 

 center was determined from calibrated, taped mark- 

 ings on the wing struts with the aircraft in level 

 flight or a hand-held inclinometer. The transect was 

 then temporarily halted and the herd circled at a 

 lower altitude to count individuals. The herd loca- 

 tion, direction of travel, behavior, and the number 

 of calves were also noted. Transect lengths and the 

 survey area were measured with a digital planimeter 

 from NOS/NOAA navigation charts. 



Depending upon the area surveyed (Fig. 1), two 

 types of survey schemes were used. Systematic, lati- 

 tudinally oriented transects were used in the Chesa- 

 peake Bay mouth (CBM) during 1980. The northern 

 starting point for each survey was randomized, and 

 each transect was located 7.4 km south of the pre- 

 vious transect. Two exceptions to this regime oc- 



■•D. M. Burn, Southeast Fisheries Center, National Marine Fish- 

 eries Service, NOAA, 75 Virginia Beach Drive, Miami, FL 33149, 

 pers. commun. June 1988. 



^Reference to trade names does not imply endorsement by the 

 National Marine Fisheries Service, NOAA. 



curred, but in neither case was the distance between 

 transects less than 3.7 km. Three or four transects 

 were flown during each CBM survey and each 

 survey covered approximately 30% of the total 

 survey area. CBM surveys were not conducted in 

 1981. 



Longshore surveys were flown from north to 

 south in 1980 and 1981, parallel to the coast and 1 

 km offshore from Cape Charles to False Cape (32.3 

 km). Those conducted in 1980 were flown imme- 

 diately upon completing the CBM surveys so that 

 there was no possibility of counting herds in the 

 longshore area that were counted during CBM 

 surveys, except perhaps during transit between 

 Cape Charles and Cape Henry, which was flown 

 over open water on the shortest line between the 

 two points. One additional survey was flown along 

 the northern Virginia coast. 



After subtracting the minimum distance from the 

 transect that could be observed because of limited 

 visibility directly beneath the aircraft, the perpen- 

 dicular sighting distance data were truncated at 1 

 km. Data from all three study sites were then pooled 

 for the calculation ofg{x), the detection function for 

 line transect, and [/'(O)], the probability density func- 

 tion of perpendicular sighting distances evaluated 

 at the transect. In line transect the detection func- 

 tion gix)is the conditional probability of observing 

 an object at perpendicular distance x from the tran- 

 sect line and /(a:;) is ^'(a;) scaled to integrate to one 

 (Burnham et al. 1980). Each survey was treated as 

 a replicate to determine the analytical variance of 

 /(O). Herd density was then calculated separately for 

 each of the survey areas using /(O) estimated from 

 the pooled sightings. 



Several estimates of /(O) and its analytical vari- 

 ance were calculated by fitting parametric and 

 nonparametric models to the distribution of perpen- 

 dicular sighting distances using the Fortran 

 programs TRANSECT (Laake et al. 1979) and 

 HAZARD and HERMITE (Buckland 1985). Maxi- 

 mum likelihood estimates and large-sample vari- 

 ances were found using the procedure of Burnham 

 et al. (1980: 135-136). The Fortran program SIZE- 

 TRAN (Drummer and McDonald 1987) was used to 

 test the hypothesis of independence between herd 

 size and perpendicular sighting distance using a 

 likelihood ratio test and thus determine if the detec- 

 tion function was biased by herd size. 



Herd density was estimated as (Burnham et al. 

 1980, p. 18, eq. 1.3): 



D = nf{0)/2L 



798 



