Woody encroachment is occurring worldwide (Archer 1995), with negative effects on biodiversity in North America (Ratajczak et al. in press) and uncertain effects on ecosystem functioning (Barger et al. 2011). The spatial properties of tree and grass dominance are well-studied and strongly suggest that woody encroachment of grasslands represents a shift to an alternative stable state (Archer et al. 1995, Shaver et al. 2001, Briggs et al. 2005, Young et al. 2007, Bond 2008, Sankaren et al. 2005, Browning et al. 2008, D’Odorico et al. 2011, Staver et al. 2011). In contrast, the temporal properties of tree-grass bi-stability have not been rigorously analyzed and thus, are not well-understood. Yet, temporal dynamics are largely responsible for generating the spatial patterns of bi-stability and are integral to understanding when and why state-shifts will and have occurred. We propose to use long-term tree and shrub cover data from 7 LTER sites and 3 non-LTER sites (Santa Rita experimental range NEON site, Edwards Plateau and Stillwater, OK) to determine if woody encroachment constitutes a state shift, and to develop and parameterize a mathematical, mechanistic model of tree-grass bi-stability (Fig 1). This analysis will take advantage of the LTER network’s unique long-term, high-resolution datasets of tree/shrub cover and environmental data. The mechanics of the resulting model can then be compared to models developed in marine, aquatic and other terrestrial systems (e.g. Scheffer et al. 2001, Bestelmeyer et al. 2011) to improve our general understanding of ecosystem bi-stability. The proposed working group is directly relevant to the LTER network initiative ‘Future Scenarios: Landscape vulnerability and resilience to climate and land-use change’.