ECOLOGY AND BIOLOGY OF THE PACIFIC WALRUS 



247 



ticated statistical methods than in the earlier surveys. Nonetheless, the prob- 

 ability that the population has increased is high, because of the great reduction in 

 magnitude of the harvests. Furthermore, increase has been indicated also by 

 recent reoccupation of abandoned hauling grounds (Gol'tsev 1968; J. J. Burns, 

 F. H. Fay, V. N. Gol'tsev, unpublished data), as well as by expansion of range to 

 pre-exploitation limits (Kosygin and Sobolevskii 1971; Pinigin and Prianishnikov 

 1975). 



The quality of future surveys probably could be improved by sampling most 

 intensively in the areas of greatest concentration of animals. The distributional 

 data now available suggest that the optimal times for such sampling would be in 

 February-March (Figs. 5,6) or in August-September (Figs. 11, 12). In those 

 months, distribution of the animals appears to be most concentrated, most 

 predictable, and most stable. The major areas to be covered in February-March 

 would be the St. Lawrence concentration and the Bristol Bay (including Pribilof) 

 concentration. Each of those probably could be surveyed separately since there 

 seems to be little possibility of interchange between them in those months. 

 Although each concentration covers about 200,000 km^ the animals are likely to 

 be most concentrated in about one-fourth to one-half of that area. With modern 

 high-speed aircraft and capabilities for infrared radiometric sensing (IRS), pre- 

 liminary visual or IRS mapping of distribution could aid greatly in designing the 

 survey for maximally effective coverage (see Ray and Wartzok 1974). 



The concentrations in August-September occupy a much larger area than 

 those in February-March: about 300,000 km^ in western and southern Chukchi 

 Sea (Soviet waters), at least 150,000 km^ in the eastern Chukchi Sea (Alaskan 

 waters), plus several thousand square kilometers in the vicinity of the major 

 hauling grounds in Bering Strait, the Gulf of Anadyr, and Bristol Bay. Hence, 

 the problem of obtaining adequate samples would be greater, and the feasibility 

 of conducting a preliminary visual or IRS reconnaissance would be practically ex- 

 cluded. Also, because the behavior of animals using coastal hauling grounds 

 probably differs from that of animals in the pack ice, the results might require 

 different interpretive methods for each area. Because of restrictive regulations 

 applying to Soviet and American air space, the actual survey would need to be 

 done cooperatively with very close coordination and communication between 

 nations. 



Insofar as possible, the preliminary reconnaissance and the surveys should be 

 conducted within the first 24 h of fair weather following a storm, because the 

 animals probably haul out more often under those than under other conditions 

 (Nikulin 1947; J. Taggart, personal communication). Surveys during very cold 

 or very windy weather should be avoided, for walruses generally stay in the 

 water when the surface "windchill" (Siple and Passel 1945) is equivalent to still 

 air temperatures of -25°C or less (Fig. 137). Animals in the water are more 

 difficult to sight than those on the ice at distances greater than 0.5 km from the 

 flight path of the aircraft (Kenyon 1972; Estes and Gilbert 1978; F. H. Fay, 

 unpublished data). Hence, the in-water sightings will be reliable only within a 

 narrow band along the flight path. Animals on ice generally are easily seen and 

 counted within 1 km of the aircraft. On the basis of surface: subsurface times 

 given in Table 26 and as summarized by Ray and Watkins (1975), the number of 

 animals sighted in the water probably tends to be about one-fourth to one-ninth 



