Review of previous Pacific walrus surveys to develop improved survey designs

The Pacific walrus (Odobenus rosmarus divergens) occupies the Bering and Ch ukchi seas between the United States and Russia, and has fluctuated over th e past 200 years as a result of exploitation. Between 1880 and 1980, annual harvests were as high as 9000 in some years. High harvests in much of that time interval were not sustainable. While the present harvest is below his torical highs, its impact on the population is not known. Efforts to monito r the population size and trend are necessary to conserve walrus population s. Surveys of Pacific walruses have been hampered by a variety of problems. Early efforts were not coordinated between American and Russian researcher s. The chronology of efforts was coordinated in 1975, 1980, and 1985, but s urvey designs differed. In 1990, scientists were exchanged and sampling des igns were shared. However, ice distribution confounded interpretation of re sults. Walrus numbers often were estimated without estimates of precision. Variances were calculated only for portions of sea-ice surveys, and land ha ulouts usually were counted only once. When they were calculated, confidenc e limits approached the size of the estimate. On sea ice, walrus numbers we re estimated from counts on one or two days when very large numbers were ob served. It was assumed that on these days a peak fraction of the population was hauled out. Problems with past surveys include: walruses could move be tween land haulouts acid nearby sea ice in the Western Chukchi Sea during t he course of a survey, walrus groups on sea ice, numbering up to several th ousands, were difficult to count with accuracy, weather and lack of aircraf t restricted timing and location of sampling efforts. After participating i n and reviewing results of past surveys, I recommend consideration of a sur vey in the spring. This survey should incorporate deployment and monitoring of satellite transmitters, and replicated counts on both the sea ice and a t land haulouts. Data for all surveys should be averaged then corrected for the average fraction of observations of walruses available to be counted.