Detection and removal of disturbance trends in tree-ring series for dendroclimatology

Nonclimatic disturbance events are an integral element in the history of forests. Although the identification of the occurrence and duration of such events may help to understand environmental history and landscape change, from a dendroclimatic perspective, disturbance can obscure the climate signal in tree rings. However, existing detrending methods are unable to remove disturbance trends without affecting the retention of long-term climate trends. Here, we address this issue by using a novel method for the detection and removal of disturbance events in tree-ring width data to assess their spatiotemporal occurrence in a network of Scots pine (Pinus sylvestris L.) trees from Scotland. Disturbance trends "superimposed" on the tree-ring record are removed before detrending and the climate signals in the precorrection and postcorrection chronologies are evaluated using regional climate data, proxy system model simulations, and maximum latewood density (MXD) data. Analysis of subregional chronologies from the West Highlands and the Cairngorms in the east reveals a higher intensity and more systematic disturbance history in the western subregion, likely a result of extensive timber exploitation. The method improves the climate signal in the two subregional chronologies, particularly in the more disturbed western sites. Our application of this method demonstrates that it is possible to minimise the effects of disturbance in tree-ring width chronologies to enhance the climate signal.