Pitfalls of applying adaptive management to a wolf population in Algonquin Provincial Park, Ontario.

We examined adaptive management (AM), applied as a science with testable and falsifiable hypothesis, in the context of a large carnivore population, specifically to wolf (Canis lupus lycaon) management in Algonquin Provincial Park, Ontario, Canada. Evidence of a population decline was based upon 12 years of data on 137 different radio-collared wolves. Because human killing accounted for an average of 66% of deaths, and most killing occurred adjacent to the park, a management prescription of complete protection for wolves around the park for 30 months was initiated in January 2001. We evaluated the probability of being able to test the null hypothesis, that protecting wolves adjacent to the park for 30 months would not result in a positive population response. Using preceding variances in population change, yearling recruitment, and mortality rates, we conducted this evaluation in two ways, the former involving a power analysis, the latter involving modeling. Both approaches showed the falsifiability of the hypothesis to be low. The reason, inherent in the application of AM to issues of population biology, especially of large carnivores, was stochasticity of the ecological system and time constraints of the human system. We discuss the political background that led up to the management prescription, and ways to avoid misapplication of a scientific approach to AM in such situations. For the latter, the merit of adjusting the relative probability levels of making Type I or Type II errors are discussed, along with recommendations that in the interests of conservation, avoiding a Type II error holds precedence.