Estimating distance to tipping point from dryland ecosystem images
Pichon, B.; Donnet, S.; Gounand, I.; Kefi, S.
Show abstract
Resource-limited ecosystems, such as drylands, can exhibit self-organized spatial patterns. Theory suggests that these patterns can reflect increasing degradation levels as ecosystems approach possible tipping points to degradation. However, we still lack ways of estimating a distance to degradation points that is comparable across sites. Here, we present an approach to do just that from images of ecosystem landscapes. After validating the approach on simulated landscapes, we applied it to a global dryland dataset, estimated the distance of each of the sites to their degradation point and investigated the drivers of that distance. Crossing this distance with aridity projections makes it possible to pinpoint the most fragile sites among those studied. Our approach paves the way for a risk assessment method for spatially-organized ecosystems.
Matching journals
The top 5 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
- Matching the forecast horizon with the relevant spatial and temporal processes and data sources 94%
- RangeShifter 2.0: An extended and enhanced platform for modelling spatial eco-evolutionary dynamics and species' responses to environmental changes. 93%
- Spatial replication can best advance our understanding of population responses to climate 92%
Similar papers in this journal
- A probabilistic view of forbidden links: their prevalence and their consequences for the robustness of plant-hummingbird communities 95%
- Ranking species based on sensitivity to perturbations under non-equilibrium community dynamics 94%
- Effects of habitat destruction on coevolving metacommunities 93%
Similar papers in this journal
- Multilayer diffusion networks as a tool to assess the structure and functioning of fine grain sub-specific plant-pollinator networks 94%
- Linking intrinsic scales of ecological processes to characteristic scales of biodiversity and functioning patterns 93%
- A theoretical framework for multi-species range expansion in spatially heterogeneous landscapes 93%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.