Methods of band survival analysis applied to studies of the Tundra Swan Cygnus columbianus

Abstract

Data from a study of the Tundra Swan, were used to estimate the lifespan of one neck band design and compare it to the lifespan of the bird. Swans were originally marked with a U.S. Fish and Wildlife Service metal tarsus band, a plastic tarsus band, and a plastic neck band. A sample of 169 recaptured swans were used in order to estimate the median survival time o f the neck band using three methods: a calibration of a weighted least squares fit of independent binomial estimates, the product-limit estimator and, survival analysis with interval censoring. Estimates of the median survival time of the plastic neck bands were 3.44. 3.05 and 2.90 years, respectively. Estimates from the survival analysis techniques are expected to be more precise than those obtained by weighted least squares because they use resighting data to pinpoint when a band was lost. In fact, however, all methods yielded survival estimates within one year o f each other and far below the longevity records of all swan species. In conclusion, estimates from three methods of band survival analysis suggest that neck band loss is a significant factor in this sample of Tundra Swan data, and most likely in the larger study of some 5,000 Tundra Swans. Therefore, estimates of annual and age-specific survival rates based on neck band resightings alone, without accounting for band loss, would be severely biased downward.