740 



Fishery Bulletin 98(4) 



v., = var(^)(r B )'. 



[31 



The sighting curve was fitted to nonindepen- 

 dent individuals instead of to independently 

 sighted groups, so the standard error ofk was 

 estimated by resampling (Hiby and Ham- 

 mond, 1989). The standard jack-knife method 

 was used (Efron and Tibshirani, 1993), and 

 sightings, presumed independent, were taken 

 as observational units. Maximum-likelihood 

 estimates are not necessarily unbiased in 

 small samples but may be subject to sample- 

 size bias. The effective number of sightings,'' 

 57.7, was less than the recommended mini- 

 mum (Buckland et al., 1993); therefore sam- 

 ple-size bias in the estimate of the survey 

 expansion factor was reduced by using the 

 standard jack-knife bias reduction (Efron and 

 Tibshirani, 1993; Buckland et al., 1993; King- 

 sley and Reeves, 1998). 



Sightings made on stratum-boundary tran- 

 sects where transect spacing changed were 

 given half weight in each stratum. 



Few sightings were made on the transect 

 sample survey in Ungava Bay, and none with- 

 in the designed survey strip. The sightings 

 made outside the transect strip were con- 

 verted to a population estimate by assuming 

 an effective strip width based on statistics 

 from other line-transect surveys that had 

 used similar platforms. No standard error 

 was calculated, but an upper confidence limit 

 on the number of groups was calculated by 

 assuming independent binomial sighting probabilities and 

 by answering the question "Given that the sampling frac- 

 tion was f, how many groups could there be for the chance 

 of seeing none or one to be less than p%?" 



Results 



The survey in James Bay was flown quickly in good 

 conditions from 12 August through 14 August. Winds 

 were light and, apart from occasional fog patches, survey 

 conditions were good. The southern part of eastern Hudson 

 Bay was flown on 15-17 August with light winds and 

 good visibility. Aircraft problems imposed a delay from 18 

 through 20 August, and on 21 and 24 August there were 

 strong winds. Richmond Gulf was surveyed on 16 August 

 (Fig. 2). 



The field of view was limited by the flat windows of 

 the aircraft. No observations were recorded closer than a 

 viewing angle of 65° from the horizontal (at a flying height 

 of 1500 feet. 213 m from the line), and few closer than 55° 

 (320 m) (Fig. 4). Sighting distances were grouped, because 

 the observers tended to round sighting angles. The last- 



■* Effective number of sightings is defined here as the number 

 of animals seen divided by the contraharmonic mean of gi-oup 

 size. 



digit frequencies of the recorded angles were analyzed, and 

 it was found that rounding to the nearest 5° was mostly at 

 the expense of the adjacent marks, i.e. those with a remain- 

 der of 1° or 4°. Against an expectation of equal frequency 

 of last digits, rounding caused a mean absolute error of 

 0.42° and was not expected to bias results or to increase 

 uncertainties. Recorded mean group size increased slightly 

 with sighting distance. 



The sighting frequency reached its maximum at 467 

 m (44.2°) and dropped off sharply beyond about 670 m 

 (35°) (Fig. 4). There were few sightings of beluga whales 

 in Hudson Bay; therefore one sighting curve was fitted to 

 the data for all strata. The bias-reduced survey expansion 

 factor was 0.575/km (SE=0.074). 



Beluga whales were widely distributed in James Bay 

 (Fig. 5A). There were 123 sightings in 4520 km of tran- 

 sect (27/1000 km), comprising 295 individuals (65/1000 

 km) (Table 1). Line-transect analysis gave an estimate of 

 3141 detectable beluga whales for James Bay (Table 2). 

 This number is about three times the estimate obtained 

 by Smith and Hammill ( 1986). However, their survey was 

 earlier, when a lot of ice still remained in northwest 

 James Bay and may have affected the distribution of these 

 whales. The highest densities in the present survey were 

 in this area. In eastern Hudson Bay (Fig. 5B) there were 

 63 sightings on 7100 km of transect (9/1000 kml compris- 

 ing 150 beluga whales (21/1000 km) (Table 2). 



