Mullin and Fulling: Abundance of cetaceans in the southern Atlantic Ocean 



607 



Estimation of mean group-size 



The group-sizes for most species tended to be related to 

 y, because in many cases larger groups are easier to see 

 than small groups with increasingy. In general, the arith- 

 metic mean of group-size may be an overestimate of the 

 true mean group-size and could lead to positively biased 

 abundance estimates. Therefore, a regression of group-size 

 by y was used to estimate an "expected mean group-size" 

 (program DISTANCE). The expected mean group-size was 

 used in the abundance estimate if it was smaller than 

 the arithmetic mean group-size. For estimates based on 

 a small number of sightings, the expected mean group- 

 size was sometimes greater than the arithmetic mean. 

 Because group-size estimates were usually made after 

 the ship approached the group, this was assumed to be an 

 artifact of the small sample size, and the arithmetic mean 

 was used in these cases. Var(S) was the analytical vari- 

 ance for mean group-sizes based on arithmetic means or 

 was estimated as in Buckland et al. ( 1993:79) for expected 

 mean group-sizes. 



Strip-transect estimates 



One requirement for unbiased line-transect estimates of 

 abundance is that the cetacean group should not move in 

 response to the ship before it is sighted (Buckland et al., 

 1993). If cetaceans are not sighted before they respond to 

 the ship, in cases of attraction to the ship, /TO) and abun- 

 dance will be overestimated. In the Gulf of Mexico, five 

 species appear to be consistently attracted to ships to ride 

 the bow waves (i.e. bottlenose, Atlantic spotted, spinner 

 [S. longirostris], Clymene [S. clymene], and pantropical 

 spotted dolphins [S. attenuata]) (Wiirsig et al., 1998). All 

 sightings made with 25x binoculars had radial distances 

 >665 m and were assumed to be made before these species 

 were attracted to the ship. If sightings of these species were 

 made at radial distances <665 m, because of the possibility 

 of attraction, they were not included in the line-transect 

 abundance estimate, and a separate strip-transect abun- 

 dance estimate was made with these sightings. For each 

 species, the width of the strip for strip-transect estimates 

 was set at the line-transect strip width (l/2/(0)) for that 

 species (Tables 1 and 2). This procedure yields the same 

 result as the formula given above with flO) for the spe- 

 cies-group category. However, f\0) for small dolphins and 

 for small whales and large dolphins combined was not 

 positively biased by including sightings of groups that 

 were probably attracted to the transect line. For each spe- 

 cies, the line- and strip-transect estimates were summed 

 for one overall abundance estimate. 



Conditionally independent observer 



The central assumption for estimating abundance with 

 line-transect methods is that cetacean groups on the tran- 

 sect line are detected with certainty (i.e.giO) = 1; Buckland 

 et al., 1993). However, this assumption is usually not met 

 during cetacean surveys because of availability and per- 

 ception bias (i.e.g(O) < 1) (Marsh and Sinclair, 1989). Some 



groups on the transect line are missed because they may 

 not be at the surface during the time the ship is in the area 

 and are not available to be seen, whereas other groups at 

 the surface are missed by observers (i.e. not perceived) 

 because of factors such as observer experience, sea state, 

 and animal behavior, among others. 



An attempt was made to estimate g(0) due to perception 

 bias with a conditionally independent observer (CIO) by 

 using methods based on Barlow ( 1995). The CIO was used 

 when the 4-observer team was on duty and was stationed 

 at 25x binoculars located on a bridge-wing 2.7 m below 

 the primary team. One individual switched teams each 

 day; therefore all seven observers on the ship acted as the 

 CIO at different times. The CIO searched for cetaceans 

 near the transect line (from 30° left to 30° right of the 

 bow) when the primary observers were on-effort. The CIO 

 and the primary team could not see or hear each other. 

 Whenever the primary team made a sighting, the data 

 recorder relayed its bearing and reticle to the CIO. When 

 the CIO made a sighting, the time, bearing and reticle 

 were noted by the CIO, and the sighting was monitored 

 until it was sighted by the primary team or, theoretically, 

 passed abeam, at which time the CIO was to notify the 

 primary team to divert the ship to identify the species and 

 estimate group-size. 



Results 



Abundance estimates were based on 4163 km of effort 

 in Beaufort sea states <,4 and 217 on-effort sightings of 

 cetacean species or other taxonomic categories (Fig. 1 and 

 Table 1). At least 13 cetacean species were sighted. The 

 most commonly sighted species (number of sightings) 

 were bottlenose dolphins (38), sperm whales iPhyseter 

 macrocephalus) (29), Atlantic spotted dolphins (28), and 

 Risso's dolphins (Grampus griseus) (22). Thirty sightings 

 occurred during transit in Beaufort sea states s4 (861 km) 

 and were used to estimate f(Q). Estimates of /iO) ranged 

 from 0.300/km for large whales to 0.561/kin for cryptic 

 whales (Table 2). 



Conditionally independent observer 



The CIO achieved 1775 km of effort (35% of effort, including 

 transit, with Beaufort sea state s4) and sighted 21 cetacean 

 groups. Of these, six groups ranging in size from 1 to 10 ani- 

 mals were missed by the primary team and included three 

 unidentified dolphin groups, two unidentified odontocete 

 groups, and one Mesoplodon sp. Each of these sightings was 

 observed briefly by the CIO but could not be tracked until 

 they passed the beam of the ship; however, in each of the six 

 cases no sightings were made by the primary team during 

 the time frame it would have been possible to sight them. 

 To estimate g^(O) following the analytical methods described 

 by Barlow (1995), a separate estimate of /(O) is therefore 

 required for CIO sightings missed by the primary team. 

 Because there were only six of these, g(0) could not be esti- 

 mated for any /lO) category, andi^(O) = 1 and varlgiO)] = 

 was used in each abundance estimate. 



