Population ecologists have devoted disproportionate attention to the estimation and study of birth and death rates and far less effort to rates of movement. Movement and fidelity to wintering areas have important ecological and evolutionary implications for avian populations. Previous inferences about movement among and fidelity to wintering areas have been restricted by limitations of data and methodology. We use multiple observation data from a large-scale capture-resighting study of Canada Geese in the Atlantic flyway to estimate probabilities of returning to previous wintering locations and moving to new locations. Mark-resight data from 28 849 Canada Geese (Branta canadensis) banded with individually coded neck bands in the mid-Atlantic (New York, Pennsylvania, New Jersey), Chesapeake (Delaware, Maryland, Virginia), and Carolinas (North and South Carolina) were used to estimate movement and site fidelity. Two three-sample mark-resight models were developed and programmed using SURVIV to estimate the probability of moving among or remaining within these three wintering regions. The model (MV2) that incorporated "tradition" or memory of previous wintering regions fit the data better than the model (MV1) that assumes that a first-order Markov chain described movement among regions. Considerable levels of movement occurred among regions of the Atlantic flyway. The annual probability of remaining in the same region for two successive winters, used as a measure of site fidelity, was 0.710 +/- 0.016 (estimated mean +/- SE [SE], 0.889 +/- 0.006, and 0.562 +/- 0.025, for the mid-Atlantic, Chesapeake, and Carolinas, respectively. The estimated probability of moving to the Chesapeake from the mid-Atlantic or from the Carolinas was 3 x and 25 x as high, respectively, as the probability of moving in the opposite directions. Changes in estimated probabilities of moving between years corresponded to changes in winter harshness. In warm years, geese moved north and in cold years, they moved south. Geese had a high probability of moving to and remaining in the Chesapeake. Annual changes in the movement probabilities did not correspond to annual changes in the United States Fish and Wildlife Service midwinter survey. Considerable numbers of geese from the Carolinas appeared to be wintering in more northerly locations ("short-stopped") in subsequent winters.
