Spectral analysis on mountain pine tree-ring chronologies

The present study applies classic spectral analysis techniques to investigate cyclic patterns in four tree-ring chronologies of Pinus montana Miller from the Central Italian Alps (Valle del Gallo). Three of the chronologies were derived from mountain pine populations located in relatively undisturbed areas of the valley bottom and valley slopes, and one from a population located in an area of the valley bottom occasionally affected by sheetfloods. Each chronology consists of raw, standard, and residual data. We estimated power spectra by applying the Blackman-Tukey Method, the Maximum Entropy Method, the Multitaper Method, and the Lomb-Scargle Fourier transform, and tested the results against appropriate red noise models. The power spectra of the standard chronologies from undisturbed areas yielded statistically significant and reproducible interdecadal-scale cyclicities with main peaks closely spaced around a mean value of ∼0.05 cycle/year, in association with statistically non-significant albeit reproducible peaks at higher frequencies. The chronology of trees affected by sheetfloods yielded no statistically significant cyclicities, probably because sheetfloods altered tree growth. Raw chronologies, instead, yielded power spectra dominated by the growth trend, while residual chronologies yielded flat power spectra. Our analysis suggests that tree growth, if not disturbed by external geomorphological factors, was controlled by environmental and/or climatic conditions that oscillated in the last ∼150 years on interdecadal (∼20 years) to decadal scales.