Checking the stability of correlation of chronologies over time: an example on Pinus pinea L. rings widths

An exploratory study concerning the variation over time of multidimensional time series allows to check to what extent factors and dendrograms, issued by ordination and classification methods, keep stable over time or change – even dramatically. In this paper, using five chronologies of Pinus pinea L. growth rings from literature, principal component analysis and hierarchical factor classification are applied on a ten years window, moving along time-series. These may be resumed through graphics showing the variation of the eigenvalues issued by the principal component analyses and of the correlations between time-series and principal components, through their corresponding time series, as well as through an animation and a compact representation of the time series of dendrograms. The results show that the studied period could be partitioned in seven intervals different in both correlations and groups structure, some of them highly stable: this suggests a second study, where two time intervals, identified as more homogeneous, showed really different structures.