Both ecological and social systems have been observed to change states abruptly as the result of crossing some critical threshold. Theories of ecological multistability have long described this phenomenon and explored how ecological management impacts stability landscapes, but with underlying tipping dynamics assumed to stem from ecological processes. Similarly, examples from social science suggest that tipping points may result from complex features of human systems, from the collapse of societies to social networks dynamics such as the spread of innovations. Despite widespread interest in the causes and location of tipping points in integrated socio-ecological systems, it has generally been assumed that the underlying dynamical complexity must be ascribed either to social process or natural phenomena alone. Using a Markov Decision Process to represent forward looking decision making in a stochastic ecological environment, we explore how rates of ecological processes, in particular ecosystem service build up, interact with decision making processes to result in long term multistable trajectories of adoption of diversified agroecological practices. We then leverage a large-scale empirical study of adoption of diversified agroecological practices across California, Oregon, and Washington as a comparison to our simulated results.
Results/Conclusions
Temporal feedbacks between a farmer's investment choice based on their perceived utility over a given time horizon and probabilistic changes in the ecological services derived from the environment can result in alternate stable states. Because benefits of ecosystem services take time to accrue, farmers in environments with degraded land are unlikely to invest in diversified agroecological practices, while farmers who benefit from ecosystem services are more likely to bolster those services further through choices of such practices. This path dependency leads to a bifurcation into either a more-simplified (conventional) or more-diversified (agroecological) farming approach, which echoes empirical findings. We show that these alternate stable states need not be an inherent feature of either ecological or decision dynamics but can emerge as a general pattern by dynamically coupling a simple ecosystem model with a rational decision process over time. We suggest that a better understanding of such tipping points has important implications for agricultural and land use policy design across a range of domains, including land tenure and agricultural subsidies.