COS 67-7 - How to avoid silent catastrophes by improving early warning signal detection

Wednesday, August 14, 2019: 3:40 PM
L013, Kentucky International Convention Center
Amy C. Patterson1, Alexander G. Strang2 and Karen C. Abbott1, (1)Department of Biology, Case Western Reserve University, Cleveland, OH, (2)Department of Mathematics, Case Western Reserve University, Cleveland, OH
Background/Question/Methods

Ecological systems can experience catastrophic change in response to certain gradual parameter changes such as increasing nutrient input or changing temperature. Once catastrophic change has occurred, a large reversal in parameter change may be required to restore the system to its original state, or restoration may not be possible. The parameter values where these drastic changes occur are called tipping points and are of great importance in ecology. Recent work has begun on the development of a system of statistical early warning signals (EWS) that can predict tipping points. However, EWS are only sometimes detected before critical transitions, leading to the possibility of silent catastrophes. Silent catastrophes can occur if some species in a system are not monitored for EWS. Our work addresses how to avoid a silent catastrophe without measuring every species, by focusing on how the strength of EWS depends on three key directions in state space: the system’s direction of critical slowing down (CSD, the phenomenon responsible for EWS), the primary noise direction, and the observation direction (i.e. which species is monitored). First, we find analytically how the strength of EWS depends on these three directions. We then explore the implications for monitoring in hypothetical two-species systems.

Results/Conclusions

We find that the strength of observed EWS increases with the alignment of the CSD direction with the direction of observation as the square of the cosine of the angle between them. The strength of EWS also increases with the alignment of the CSD direction and the primary noise direction, though this dependency is moderated by how much noise occurs in the primary direction. In a two species system, we determine that EWS are most likely observed in the species experiencing smaller intraspecific effects, and in the species experiencing larger interspecific effects. The magnitude of intraspecific effects matters more when the system is dominated by competition or mutualism, and the magnitude of interspecific effects matters more when the system is dominated by predator-prey interactions. Practically, our findings can be used to reduce the monitoring effort required to detect EWS. They also show that it may be difficult to monitor complex systems for EWS, as our work shows explicitly that detecting strong EWS becomes more difficult in systems with greater numbers of species.