Time-dependent decisions under non-stationarity: Applications to the land acquisition problem (ver. 1.0.0)

This code provides a simple quantitative demonstration of an imagined land acquisition problem. Conservation results (i.e., benefit or reward) is evaluated under three competing decision models and varying degrees of ecological non-stationarity. The decision models reflect a spectrum of accommodating time-dependence into decision making: an optimal, fully time-dependent Markov Decision Process (MDP) model; a greedy heuristic; and an intermediary myopic MDP. The cumulative benefit (i.e., cumulative rewards) calculated at the end of all time steps is used to compare model performance, and assess the conditions under which additional levels of decision model complexity is justified. The application of these models is imagined in the Prairie Pothole Region (PPR) of the United States, where land acquisition is critical to the conservation of migratory waterfowl. This analysis was used to assess the impact of incorporating time-dependent decision models into conservation decisions in the associated manuscript (Rubenstein et al, 2026. "Cognitive and Institutional Barriers to Effective Natural Resource Management under Global Change". Conservation Biology).