Early warning signals of regime shifts from cross-scale connectivity of land-cover patterns

Abstract Increasing external pressures from human activities and climate change can lead to desertification, affecting the livelihood of more than 25% of the world's population. Thus, determining proximity to transition to desertification is particularly central for arid regions before they may convert into deserts, and recent research has focused on devising early warning signals for anticipating such regime shifts. We here draw the attention to some emerging land-cover cross-scale patterns with a historical characteristic sequence of different regimes in arid or semi-arid Mediterranean regions that could indicate an impending transition to the tightening and extension of desertification processes. Inflexibility of land administration may, in turn, reinforce desertification processes, erode the resilience and promote regime shifts and collapse instead of the adaptability required to counter surprises due to climate change. Various theoretical studies have designated the increase in spatial connectivity as the leading indicator of early warning for an impending critical transition of regime shifts. We show that a potential way to address early warning signals of regime shifts to monitor and predict changes is to look at current land-cover regime within a simple framework for interpreting cross-scale spatial patterns. We provide examples of this approach for the Apulia region in southern Italy with desertification processes in place, and discuss what a cross-scale land-cover pattern could mean, what it says about the condition of socio-ecological landscapes, and what could be the effects of changing observed conditions ought to, for instance, climate change. We took advantage of the rich information provided by cross-scale pattern analysis in the pattern transition space provided by classic neutral landscape models. We show potentially dramatic shifts of connectivity at low land-cover composition below certain thresholds, and suggest that the degree to which the observed pattern departs from a particular neutral model can indicate early warning signals of regime shifts, and how those landscapes might evolve/react to additional land-cover variation. Moreover, as the land-cover pattern mostly depends on social-economic factors, we argue that we have to change societal values at the root of inflexibility.