Estimation of Confidence Intervals for Federal Waterfowl Harvest Surveys

Abstract

I developed methods of estimating confidence intervals for the federal waterfowl harvest surveys conducted by the U.S. Fish and Wildlife Service (USFWS). I estimated flyway harvest confidence intervals for mallards (Anas platyrhynchos) (95% CI are ?8% of the estimate), Canada geese (Branta canadensis) (? 11%), black ducks (Anas rubripes) ( 16%), canvasbacks (Aythya valisineria) (?32%), snow geese (Chen caerulescens) (?43%), and brant (Branta bernicla) (?46%). Differences between annual estimates of 10, 13, 22, 42, 43, and 58% could be detected with mallards, Canada geese, black ducks, canvasbacks, snow geese, and brant, respectively. Estimated confidence intervals for state harvests tended to be much larger than those for the flyway estimates. J. WILDL. MANAGE. 54(2):201-205 The federal waterfowl harvest surveys provide estimates of (1) waterfowl harvest by species for flyways and states, (2) harvests of most other migratory game bird species by waterfowl hunters, (3) crippling losses for ducks, geese, and American coots (Fulica americana), (4) days hunted, and (5) bag per hunter (Voelzer et al. 1982). Information on waterfowl harvests in the United States comes from the Postal Service's report of duck stamp sales, the waterfowl hunter survey, and the waterfowl parts collection survey (Couling et al. 1982, Voelzer et al. 1982). Waterfowl managers should consider the reliability of harvest estimates when making management decisions. Differences among annual harvest estimates may result from sampling variability and may not represent real changes in harvest. Confidence intervals help managers distinguish between sampling variability and changes that may require action. Consequently, I documented methods used to estimate waterfowl harvest and developed methods of estimating confidence intervals. Confidence intervals are not usually reported with harvest estimates. The intervals I computed suggest the magnitude of annual fluctuations in harvest estimates that can be expected due to chance alone. Confidence intervals are presented for flyway harvests of mallards, black ducks, canvasbacks, Canada geese, snow geese, and brant, and for state harvests of mallards. E. M. Martin provided advice on the procedures for conducting and analyzing the current waterfowl harvest survey. C. M. Bunck, J. E. Hines, R. E. Trost, and B. K. Williams made helpful suggestions, and C. B. Thrash and E. Brown, Jr. provided computing assistance. Waterfowl Survey Procedures In spring, the Migratory Bird Management Office (USFWS) randomly selects a sample of 3,000 out of 16,000 post offices that sell duck stamps. The survey is stratified by geographic zones within states and by 3 post office size categories based on the number of duck stamps sold ( 999). Numbers of sample post offices are allocated to strata so that the same proportion is sampled in every stratum. Each post office is a cluster sample of duck stamp buyers. Contact cards for this hunter survey are sent to sample post offices before duck stamp sales begin on 1 July. Two reminders to hand out contact cards are sent to sample post offices, one shortly before the season opens and another about mid-season. When a person buys a duck stamp a a selected post office, the clerk gives the buyer a contact card. The buyer is asked to return the card with his/her name and address and an indication of whether or not the stamp was purchased for hunting or collecting. The card informs buyers that they will receive a questionnaire about duck harvests, and it provides a brief diary in which to record hunting activity. Approximately 19% of the stamp purchasers at sample post offices return contact cards by deadline just before the close of the hunting season in their states. Those who plan to hunt are included in the hunter survey. About 70,000 questionnaires are mailed shortly after the close of the season, and after 3 or 4 weeks a follow up ' Present address: U.S. Fish and Wildlife Service, Migratory Bird Management Office, Laurel, MD 20708.