Predicting catastrophic shifts

Catastrophic shifts are known to pose a serious threat to ecology, and a reliable set of early warning indicators is desperately needed. However, the tools suggested so far have two problems. First, they cannot discriminate between a smooth transition and an imminent irreversible shift. Second, they aimed at predicting the tipping point where a state loses its stability, but in noisy spatial system the actual transition occurs when an alternative state invades. Here we suggest a cluster tracking technique that solves both problems, distinguishing between smooth and catastrophic transitions and to identify an imminent shift in both cases. Our method may allow for the prediction, and thus hopefully the prevention of such transitions, avoiding their destructive outcomes.