Persistence of growth variation in tree-ring chronologies

We infer baseline response to climate of the growth of trees in California dendroclimatological samples, adopting linear and nonlinear approaches and using the expected accumulated impact of a state change as one measure of p ersistence. Many samples exhibit two quite distinct growth states. Timing o f high and low growth states is highly heterogeneous across locations, and regional aggregation seriously understates the size and persistence of loca l growth changes. If climate is a dominant factor in local tree growth, the n local climate is not well represented by an aggregate regional index.