Dendroecological reconstructions of forest disturbance history using time-series analysis with intervention detection

The detection of release events in the annual growth increments of trees has become a central and widely applied method for reconstructing the disturbance history of forests. While numerous approaches have been developed for identifying release events, the preponderance of these methods relies on running means that compare the percent change in growth rates. These methods do not explicitly account for the autocorrelation present within tree-ring width measurements and may introduce spurious events. This paper utilizes autoregressive integrated moving-average (ARIMA) processes to model tree-ring time series and incorporates intervention detection to identify pulse and step outliers as well as changes in trends indicative of a deterministic exogenous influence on past growth. This approach is evaluated by applying it to three chronologies from the Forest Responses to Anthropogenic Stress (FORAST) project that were impacted by prior disturbance events. The examples include a hemlock (Tsuga canadensis (L.) Carriere) chronology from New Hampshire, a white pine (Pinus strobus L.) chronology from Pennsylvania, and an American beech (Fagus grandifolia Ehrh.) chronology from Virginia. All three chronologies exhibit a clustering of step, pulse, and trend interventions subsequent to a known or likely disturbance event. Time-series analysis offers an alternative approach for identifying prior forest disturbances via tree rings based on statistical methods applicable across species and disturbance regimes.