Detrending tree-ring widths in closed-canopy forests for climate and disturbance history reconstructions

Canopy disturbance events in forests often increase light availability and growth rates for surviving trees. Using ring widths, release-detection methods identify the initiation of rapid growth associated with these events enabling reconstructions of forest disturbance history. Conversely, dendroclimate reconstructions minimize these rapid growth responses by detrending entire ring-width time series to resolve underlying climate signals. Incorporating advantages from both of these approaches, we present a canopy disturbance detrending method that quantifies the subsequent, additional growth from release events as a discrete time series. Our method uses radial-growth averaging to detect release events and then power transformation and age-dependent smoothing splines to detrend individual release events, separating canopy disturbance responses from climatic effects. We test our canopy disturbance detrending method on both a coniferous and a broadleaf ring-width chronology from two contrasting temperate forests in eastern North America with known disturbance histories. The resulting disturbance growth time series agrees with the documented forest histories for each forest. Removing the effects of canopy disturbance from the ring-width series in each chronology improves climate correlations with monthly values of temperature and precipitation and may recover more low-frequency variation with past climate compared to other common detrending methods. Our method provides an alternative approach for detrending disturbance events in closed-canopy forests that should be accessible and useful for both ecological and climatological reconstructions.