Background/questions/methods:
Ecosystem transitions resulting from climatic drivers are observed at multiple time scales. In the arid southwest United States, the role of climate variation in the collapse and recovery of grassland communities is poorly understood. Here we use a quality-controlled vegetation monitoring dataset initiated in 1915 to ask what climate components were most closely correlated to perennial grass dynamics in the 20th century. We constructed a suite of models relating grass cover to climate variables known to influence dryland ecosystem behavior at multiple time scales, including annual precipitation, annual drought indices, El Nino Southern Oscillation (ENSO) index, and Pacific Decadal Oscillation (PDO) index.

Results/conclusions:
We found that dynamics of perennial grass collapse and recovery during the 20th century were most closely correlated to the Pacific Decadal Oscillation (PDO) index. The relationship out-performed models correlating grasses to yearly precipitation and drought indices, suggesting that ecosystem transitions attributed only to local disturbances were instead influenced by climate teleconnections. We also found that an incidence of grass recovery in the 1960s-70s was accompanied by a significant shift from core species (common, abundant species) to transient species (species that are patchy in space or time). Finally, the relationship between PDO and vegetation cover broke down after 1995, indicating that warming or land degradation overwhelmed multi-decadal variation driven by PDO.