Regime shifts are large, abrupt, persistent changes in the structure and function of a system(Biggs et al. 2009). A regime is a characteristic behavior of a system which is maintained bymutually reinforced processes or feedbacks. Regimes are considered persistent relative to thetime period over which the shift occurs. The change of regimes, or a regime shift, usually occurswhen a gradual change in an internal process (feedback) or a single disturbance (externalshocks) triggers a completely different system behavior (Scheffer et al. 2001, Beisner et al.2003, Scheffer and Carpenter 2003, Folke et al. 2004). Although such non-linear change havebeen widely studied in different disciplines ranging from atoms to climate dynamics(Feudel2008); regime shifts have gained importance in ecology because they can substantially alter theflow of ecosystem services to societies (Millennium Ecosystem Assessment 2005, Biggs et al.2009); such as provision of food, clean water or climate regulation. Moreover, the frequency ofregime shifts is expected to increase as human domination on the Earth’s processes continuesto expand -the anthropocene-, modifying climate, biogeochemical cycles, land cover, andspecies distribution (Steffen et al. 2007, Rockström et al. 2009).

Abrupt ecological changes or regime shifts have been identified in many ecosystems, but there has been no systemic review to compare regime shifts and their drivers. We use network analysis to perform a comparative analysis of relationships among global change drivers and major ecological regime shifts. We assembled a database of 20 regime shifts and their drivers that spans marine, terrestrial and polar ecosystems. Our analysis reveals that regime shifts typically have multiple drivers (ranging from 2 to 19 in our dataset) that interact in a structured fashion: only 11 drivers cause from 25-60% of the regime shifts and interact with 50-85% of other drivers. Of the 55 drivers recorded, 11 drivers cause 25-60% of the regime shifts and interact with 50-85% of other drivers. Drivers related to agricultural activities and climate change are the most frequent drivers of regime shifts. Although equally important, indirect drivers seem to be underreported. We find that regime shifts in marine systems operate at similar scales due to similar feedback processes, while terrestrial regime shifts occur across a wider range of scales and are more context dependent. Regime shifts differ greatly in the scales at which their drivers can be managed, but most regime shifts are at least partially driven by global change drivers which cannot be managed locally, highlighting the importance of new scale-bridging polycentric approaches to avoid unwanted regime shifts.