Terrestrial ecosystems are currently undergoing unprecedented climate and human-induced disturbances, which are likely to push these systems towards changes in their physiognomies, structure, and functioning. It has been hypothesized that these new configurations may be alternative states of systems comprising vegetation-climate-disturbance interactions (Fig. 1).
The majority of the studies reporting ecosystem switches considers vegetation-climate-disturbance systems confined to certain spatial scales (local to continental) without accounting for multi-scale interactions and are unable to detect out-of-range changes and/or regime shifts in vegetation due to difficulties in collecting sufficiently long time series to define standard behavior of the system.
In this context, we propose to investigate novel machine learning and image processing techniques aiming to support the use of multi-scale ecological knowledge in the analysis of vegetation-climate-disturbance systems. We also propose the use of the theory of dynamical complex systems as a novel way of filling up the gaps in evaluating ecosystem transitions, transients, and alternative states under current land use and climate change trends.
Study area: Brazilian transitional forest and savannah ecosystems (See Fig. 3).
Project website: E-tribes project - http://e-tribes.com.br
Funding & acknowledgements:
A joint call on E-science applied to environmental studies, funded by Microsoft Research and FAPESP (São Paulo funding agency for researching). Grant number is 2013/50169-1.