We defined our dataset based on temporal and geographic criteria. We 

 included sightings collected during the September-October surveys of 1988, 1989, 

 1990, 1991, 1992, and 1993 within the designated boundaries considered to comprise 

 Tampa Bay (Figure 1). 



Group size estimates were derived from adjustments of field estimates based on 

 photo-analysis (see Appendix 2). Minimum, maximum, and best field estimates were 

 increased if the sum of the number of positively identified individuals plus the 

 number of "other..." dolphins, plus the number of "clean" dolphins exceeded the 

 original field estimates. The resulting revised minimum, revised maximum, and final 

 best estimates were used in all calculations involving group size. 



Several of the abundance and trend estimates and the power analyses were 

 conducted at the Inter- American Tropical Tuna Commission with a VAX 3100/80 

 micro-computer and a 486 IBM-compatible personal computer. Linear regressions 

 were performed using a SAS procedure (SAS, 1990). A FORTRAN program designed for 

 use on IBM-compatible personal computers (TRENDS2; Gerrodette 1993) allowed us to 

 conduct a power analysis to detect trends in abundance (Gerrodette 1987). 



Estimation procedure s: Abundance 



The basic questions considered by this project were: "How many dolphins use 

 the Tampa Bay study area during the September-October survey period, and how does 

 this number vary from year to year?" A closed population was assumed because of 

 the short interval during which the surveys took place. There are a variety of ways 

 to calculate indices of abundance of bottlenose dolphins inhabiting Tampa Bay. 



Method 1 (catalog-size method) simply involves tallying the number of 

 positively identified ("marked") individuals (M) sighted within the study area during 

 the survey period. We derived our overall catalog of marked animals for each survey 

 year by considering all sightings during the survey period regardless of the photo 

 grade. The inclusion of a fin in the catalog was dependent on the recognizability of 

 a dolphin, not the overall quality of coverage of a sighting. The catalog-size method 

 does not account for dolphins that are not distinctively marked. The size of the 

 annual Tampa Bay catalog (M) is an integral part of each of the following three 

 abundance estimation procedures. 



Assuming comparable levels of sighting effort from year to year, the catalog- 

 size approach may provide a reasonable index for detection of trends of abundance. 

 To conduct a power analysis, however, a coefficient of variation (CV = var 1/2 / N) 

 could only be calculated by considering each year (1988-1993) as a replicate sample. 

 A regression analysis of the six annual estimates was conducted to remove the effects 

 of a potential trend; the CV was then calculated from the residuals. 



Method 2 (mark-proportion method) calculated the proportion of positively 

 identified dolphins (m) relative to the total group size (n) in each sighting of "Grade- 

 1 " quality. The accuracy of the population-size estimates depends on the confidence 

 in identifications. Therefore, only Grade- 1 sightings were used to derive the 

 proportion of marked animals. There was no relationship between group size and 

 the proportion of dolphins identified (r 2 = 0.007). 



The proportions of marked dolphins to group size (m/n) for each sighting 

 were averaged for each year. The total number of marked dolphins in the catalog for 



