Exploiting delayed transitions to sustain semiarid ecosystems after catastrophic shifts

Abstract

Semiarid ecosystems (including arid, semiarid and dry-subhumid ecosystems) span more than 40% of extant habitats and a similar percentage of human population. As a consequence of global warming, these habitats face future potential shifts towards the desert state characterized by an accelerated loss of diversity and stability leading to collapse. Such possibility has been raised by several mathematical and computational models, along with several early warning signal methods applied to spatial vegetation patterns. Here we show that just after a catastrophic shift has taken place an expected feature is the presence of a ghost, i.e., a delayed extinction associated to the underlying dynamical flows. As a consequence, a system might exhibit for very long times an apparent stationarity hiding in fact an inevitable collapse. Here we explore this problem showing that the ecological ghost is a generic feature of standard models of green-desert transitions including facilitation. If present, the ghost could hide warning signals, since statistical patterns are not be expected to display growing fluctuations over time. We propose and computationally test a novel intervention method based on the restoration of small fractions of desert areas with vegetation as a way to maintain the fragile ecosystem beyond the catastrophic shift caused by a saddle-node bifurcation, taking advantage of the delaying capacity of the ghost just after the bifurcation.