BLAYLOCK: DISTRIBUTION OF THE BOTTLENOSE DOLPHIN 



individuals present during either of those two years, 

 at least 26% were present at some time during both 

 summers. 



14 



o 

 cc 



lU 



X 



z 



CO 



lU 



> 



< 



u 



GC 



UJ 



a. 



12- 



10- 



2- 



16 



13 



20 



MAY 



JUN 



JUL AUG 



MONTH 



SEP 



OCT 



Figure 5.— Percentage of bottlenose dolphin calves in herds by 

 month. Bars represent the standard error of the mean (horizontal 

 line within bars) and vertical lines, 95% confidence intervals. Num- 

 bers above bars denote the number of herds sighted per month. 



DISCUSSION 



The choice of a model for g{x), the probability of 

 detecting an object at a distance x from the tran- 

 sect, is the primary analytical consideration in a line 

 transect estimate of density (Burnham et al. 1980; 

 Seber 1982, 1986). Burnharn et al. (1980) thoroughly 

 review the subject of density estimation from line 

 transect surveys and recommended the Fourier 

 series as a general model for g(x). However, addi- 

 tional models which meet their criteria have since 

 been proposed (Bnckland 1985; Seber 1986). Buck- 

 land (1985) suggested the use of a model where the 

 cosine terms in the Fourier series equation are re- 

 placed by Hermite polynomials. 



Buckland (1985) warned that if the model requires 

 four or more terms to fit the distributional data, one 

 or more of the assumptions of line transect theory 

 may be violated. I suggest that, in aerial surveys 



of cetaceans, the primary assumption that all objects 

 on the transect are observed with a probability of 

 one [giO) = 1] is routinely violated. The diving be- 

 havior of cetaceans during different activities may 

 vary widely, thus the probability of the animals be- 

 ing at the surface when the observers pass may also 

 vary. Also, active dolphins may be more readily 

 detected than resting dolphins. In spite of this, this 

 assumption is somewhat less restrictive than the 

 primary assumption of strip census which assumes 

 that all objects within the strip are detected. If the 

 other assumptions are met, the major consequence 

 of failure to meet the assumption of ^'(0) = 1 is that 

 density will be underestimated. 



A further assumption is that perpendicular dis- 

 tances are measured without error. Even using an 

 inclinometer, vertical motion of the aircraft and 

 inaccuracy of the altimeter introduce error into 

 distance measurements. Grouping distance mea- 

 surements into discrete intervals is a logical way in 

 which to compensate if the model used is robust to 

 grouping. 



The assumption of random location of transects 

 with respect to bottlenose dolphin distribution was 

 met by randomization of the starting point of each 

 survey in the CBM. It is obvious from the cluster 

 of sightings at Cape Henry and Fisherman Island 

 that bottlenose dolphins were not distributed ran- 

 domly in the coastal study area (Fig. 1). This could 

 occur if bottlenose dolphins were counted more than 

 once; however, their movement was slow compared 

 with that of the observers and, because longshore 

 surveys were flown immediately upon completion 

 of CBM surveys, it is unlikely that dolphins were 

 counted more than once. It is more likely that the 

 cluster of sightings was because of an environmen- 

 tal factor, such as the attraction of dolphins to con- 

 centrations of prey in fronts between estuary and 

 ocean waters. 



According to Essapian (1963), mating by the 

 bottlenose dolphin occurs in the spring and birth 

 occurs about one year later (McBride and Kritzler 

 1951; Tavolga and Essapian 1957). Mead (1975), 

 citing True (1891), stated that "Information received 

 from the fishermen at the Hatteras fishery indicated 

 that fetuses were generally small in September, in- 

 creasing in size as the season progressed." This im- 

 plies that natality occurs primarily in the spring. 

 Townsend's (1914) data (also cited in Mead 1975) 

 suggest an additional autumn peak in natality. The 

 June peak in the percentage of calves agrees with 

 those observations suggesting a spring natality 

 peak; however, because of the slight increase in the 



803 



