An examination of methods to estimate population size in wintering geese

We estimated the size of the Svalbard population of Pink-footed Geese Anser brachyrhynchus in 1991–98 using three different methods: (a) counts; (b) mark–resight estimates; (c) annual productivity and survival. Count data showed a slight increase of population size, from 33 000 in 1991 to 38 500 in 1998. Mark–resight estimates showed a larger fluctuation, but were almost always greater than counts. By contrast, estimates of survival and productivity suggested stability or at least a less pronounced increase in the population size, the discrepancy in the number estimated when compared to the other methods being especially large in the last two years of the study. A detailed examination of the assumptions underlying each of the methods reveals possible explanations for some, but not all, of the discrepancies. We conclude that goose population estimates derived from total population counts may be less reliable than commonly assumed, and moderate year-to-year trends should not be over-interpreted. Similarly, assessment of annual productivity and survival may be subject to undetected biases, and these uncertainties should be considered when interpreting results and trends in these parameters. Repeated cross-validation of parameter estimation methods in this and other populations is highly desirable.