BARHAM KT AI. AERIAL CENSUS OF THE BOTTLENOSE DOLPHIN 



Var Nj = 5^(nj^ Var h^ + /j/ rij - nj Var hj) . (9) 



Before proceeding, a one-way analysis of var- 

 iance with unequal sample sizes was performed on 

 herd sizes with a logm transformation for counts. 

 No significant differences i« = 0.05) between 

 replicate herd sizes were found, thereby allowing 

 the pooling of the four variances as: 



Var dj = — (n,'^ Var h, + hj^ nj - iij Var h,) (12) 



VarN = Var 



Var (Nj) .(10) 



These computations produced an estimated 

 mean T. truncatus population size of 1,319 with a 

 standard error (SE) of 189 (Table 2). 



The susceptibility of the above analysis to possi- 

 ble nonindependence of the mean herd size and 

 herd density parameters was recognized by 

 Leatherwood et al. ( 1978), and they suggested that 

 mean herd size be established in preliminary 

 flights before the herd counting phase of the sur- 

 vey is initiated. In the case of our work, however, 

 because of inclement weather and limited re- 

 sources we decided to make as many replicate sur- 

 veys as possible rather than dividing the flight 

 functions. 



Despite the assurance of ranking tests, if inde- 

 pendence between /?, and n, does not hold, use of 

 Equation (9i will probably underestimate the var- 

 iance of iV, . An alternative more robust approach 

 suggested by one reviewer was to compute the SE 

 of the replicate estimates of numbers of dolphin on 

 the four surveys (Table 2). This procedure pro- 

 duces a SE of 130.0 which is reasonably close to the 

 theoretical value of 189 obtained from Equation 

 (9) and tempers to some extent doubts of the valid- 

 ity of this approach. 



Estimated Dolphin Density 



For comparative purposes we also estimated the 

 density of dolphins in the study area from: 



dj = Djhj = — hi . 

 ' ' ' a 



(11) 



The same rationale and procedures for calculat- 

 ing the replicate and overall variances of popula- 

 tion estimates were used to calculate the var- 

 iances for dolphin density. Thus: 



and 



Varrf =(j) V Yard,. (13) 



This treatment gave an estimate of 0.752 

 dolphins/km" with an SE of 0.074. The SE calcu- 

 lated from the variance of the mean of the repli- 

 cates was 0.108 (Table 2). 



Comparisons with Other Population Studies 



We can roughly compare our counts from the 

 Aransas Pass area with those of Shane's (1977) 

 who counted T. truncatus in the same area from a 

 skiff run on a meandering course through the ship 

 channels and cuts almost on a daily basis over a 

 1-yr period. For March and April 1977, her mean 

 was 95 dolphins. The mean of our scores for tran- 

 sects 1 and 2 that covered part of her study area 

 was 53. Considering the differences in methods 

 and area covered, the results do not seem unrea- 

 sonably diverse. 



Our mean density estimate for all transects is 

 compared with the results of recent aerial surveys 

 of T. truncatus populations in waters adjacent to 

 Florida, Mississippi, and Louisiana in Table 3. 

 While it is clearly tenuous to contrast densities 

 from different environments, it is worth noting 

 that the two semienclosed areas, Indian River, 

 Fla., and the Texas bays, appear to support similar 

 densities, 0.52 dolphins/km^ and 0.75 dolphins/ 

 km^, respectively. The mean percent of the calves 



Table 3. — Density estimates of bottlenose dolphin populations 

 in southeastern US, coastal waters, based on recent aerial sur- 

 veys. There are considerable differences in the nature and extent 

 of the areas covered in these studies, thus the results are not 

 strictly comparable. 



^Derived from their table 10 by computing the product of mean herd size 

 (5 43) and mean herd density (0 0497) 



591 



