Premium
Active adaptive management for reintroduction of an animal population
Author(s) -
Runge Michael C.
Publication year - 2013
Publication title -
the journal of wildlife management
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.94
H-Index - 111
eISSN - 1937-2817
pISSN - 0022-541X
DOI - 10.1002/jwmg.571
Subject(s) - wildlife , population , markov decision process , adaptive management , value of information , iterated function , framing (construction) , computer science , biology , markov process , demography , statistics , environmental resource management , mathematics , economics , geography , ecology , artificial intelligence , mathematical analysis , archaeology , sociology
ABSTRACT Captive animals are frequently reintroduced to the wild in the face of uncertainty, but that uncertainty can often be reduced over the course of the reintroduction effort, providing the opportunity for adaptive management. One common uncertainty in reintroductions is the short‐term survival rate of released adults (a release cost), an important factor because it can affect whether releasing adults or juveniles is better. Information about this rate can improve the success of the reintroduction program, but does the expected gain offset the costs of obtaining the information? I explored this question for reintroduction of the griffon vulture ( Gyps fulvus ) by framing the management question as a belief Markov decision process, characterizing uncertainty about release cost with 2 information state variables, and finding the solution using stochastic dynamic programming. For a reintroduction program of fixed length (e.g., 5 years of releases), the optimal policy in the final release year resembles the deterministic solution: release either all adults or all juveniles depending on whether the point estimate for the survival rate in question is above or below a specific threshold. But the optimal policy in the earlier release years 1) includes release of a mixture of juveniles and adults under some circumstances, and 2) recommends release of adults even when the point estimate of survival is much less than the deterministic threshold. These results show that in an iterated decision setting, the optimal decision in early years can be quite different from that in later years because of the value of learning. © 2013 The Wildlife Society.