Long-term Responses of Canopy-understorey Interactions to Disturbance Severity in Primary Picea abies Forests

Abstract

In this paper, we investigate how past disturbances (whose data were obtained by carefull reconstruction based on core dendrochronological analysis) influence current stem density in primary Picea abies forests. The investigation was based on semiparametric GAM modeling, using both time to the main past disturbance and its severity, as well as their (parsimonious) interaction. To represent substantial data features (overdispersion, spatial trend and autocorrelation), we used negative binomial GAM with smooth spatial effect and location-specific random intercepts. Effects of main interest (time, severity and their interaction) were modeled nonparametrically (as penalized splines with roughness penalty estimated via crossvalidation). The interaction was modelled by a tensor product spline. Model was fitted via penalized likelihood, allowing for both estimation and testing of various hypotheses of interest. The present investigation greatly extends the timescale of studies monitoring regeneration dynamics following disturbance. We have produced a quantitative model of understory cohort densities, which reinforces that shade-tolerant understory cohorts reduce the amount of time required to recover pre-disturbance conditions. Treating disturbance severity as a continuous variable avoided the a priori assignment of forest dynamics to either even or uneven aged, which should be treated as the two extremes on a continuum of potential age structures.