| Summary: | The hunting of geese and other waterfowl is an activity that circulates millions of dollars each year in North America. The Atlantic population of Canada Geese (Branta Canadensis) has historically been a major target for hunters throughout the eastern parts of the United States and Canada, although numbers declined significantly in the 1990s. Resident (non-migratory) populations of geese and migratory populations during the non-breeding season can damage crops and cause public nuisance complaints. Thus, management of goose populations can be a balance between providing high harvest opportunity while not allowing populations to get so large as to cause damage. We investigate optimal control of the Atlantic population of Canada Geese. We seek to maximize harvest, while maintaining the population within acceptable upper and lower bounds. Control is obtained by setting the harvest rate on breeding adult birds each year. The optimal harvest strategy is informed by the population state, and must incorporate a range of uncertainties regarding population dynamics and the ability to control the population. The first uncertainty is environmental variation. This is unpredictable and uncontrollable, and represented by stochasticity in breeding productivity from year to year. The second uncertainty considered in this paper is a limited ability to control the population. Given a finite number of hunters, there may be an upper limit on the total number of birds that can be harvested each year. We explore a range of limits to total annual harvest. Structural uncertainty is the third uncertainty considered. Models constructed to represent population dynamics are not a perfect description of the true dynamics. In particular, there is disagreement about the strength of density dependence underlying Canada Goose dynamics. We pose two reasonable but contrasting models of density dependence in this study. Both models of population dynamics include age structure. Canada Geese, like other goose populations, exhibit life-history attributes that differ by age. Thus, a structured model may be necessary to fully capture the dynamics of this population. Annual harvest decisions are made using the estimated number of birds in each age group. While age-structured harvest has been investigated in the past, the objective has usually been only to maximize yield. In this study we have the additional goal of maintaining population size within set bounds. Stochastic dynamic programming has rarely been used to optimize the harvest of structured populations, probably due to the increased dimension of the state space required to describe population structure. Simulations of the optimal harvest under each model show a range of strategies. Under the density independent model, annual harvest is maximized by holding the population size as close as possible to the upper acceptable limit, while ensuring that stochastic fluctuations rarely exceed this limit. When there is limited control, however, the population is optimally maintained at a lower level, to ensure that it does not grow beyond harvest capacity and continue indefinitely with an unacceptably large abundance. Under the density dependent model, the maximum sustainable yield may be obtained by keeping population size at some level between the minimum and maximum acceptable thresholds. Limits to control do not significantly change this optimal population size, although the amplitude of fluctuations may increase under very limited control. This result depends critically on the fact that the population size that achieves maximum sustainable harvest falls within the desired bounds. If this were not the case, the limits to control could play a more central role. It is clear that the strength of density dependence and constraints on harvest significantly affect the optimal harvest strategy for this population. Model discrimination might be achieved in the long term, while continuing to meet management goals, by adopting an adaptive management strategy.
|
|---|