The recognition that ecosystems can undergo sudden shifts to alternate, less desirable stable states has led to the desire to identify early warning signs of these impending collapses. Motivated by the mathematics of bifurcations, the search for early warning signs seeks to detect subtle patterns that may precede these shift. Faced with limited and imperfect data, such forecasting is fraught with uncertainty. I will present work illustrating how we can quantify the risks of false alarms and missed detection of these early warning signals, and illustrate how this information can be used to inform robust decision-making that may avoid collapse.
The most pressing issues of our time are all characterized by sudden regime shifts: the collapse of marine fisheries or stock-markets, the overthrow of governments, shifts in global climate. Despite their importance, predicting these shifts has long been more the domain of soothsayers than scientists, an area where experts perform no better than lay observers. Such transitions by definition lie outside the familiarity of our experience and the description of our models. By tapping the data deluge of information technological innovations have made available and by identifying a common mathematical framework around some sudden shifts, I will suggest how we can begin to break free of this mold.