prilagođeno pretraživanje po punom tekstu

ŠUMARSKI LIST 7-8/2020 str. 25     <-- 25 -->        PDF

Akaike Criterion and combined across all series using biweight robust estimation of the mean to exclude the influence of the outliers. ARSTAN produces two types of output chronologies – a standard chronology (STD) and a residual chronology (RES) containing only high-frequency variations (no autocorrelation) (Cook, 1985, Cook et al., 1990). For the purpose of this research, we used RES chronology, which represents a robust estimate of the arithmetic mean and contains no autocorrelation (Cook, 1985). After detrending, the Expressed Population Signal (EPS) was calculated to assess the common forcing (e.g. climate) in tree-ring width chronologies. An EPS of 0.85 or higher is generally accepted to be high enough to show that analysed tree ring series represent the common forcing mechanisms of a larger population of trees (Briffa and Jones, 1990).
Data analyses – Analiza podataka
Defoliation trend was tested by Mann-Kendall test (Mann, 1945, Kendall, 1948) and the slope was estimated according to Sen (1968). In order to establish if defoliation statistically differed during the investigated period we performed one-way analysis of variance. Prior to the analysis, defoliation data was checked for normality and homoscedasticity. Since the data did not adhere to the latter assumption we applied the Welch’s ANOVA (Welch, 1951). To specify differences between years we conducted the Games-Howell test (Games and Howell, 1976). For the trend analysis we used the “EnvStats” package (Millard, 2013), while the package “userfriendlyscience” (Peters, 2017) was used to conduct ANOVA and the subsequent post-hoc test.
Bootstrapped Pearson’s correlation coefficient calculated in the package “treeclim” (Zang and Biondi, 2015) was used to identify dependency between the residual chronology and climate variables for the period 1950-2008 and to identify dependency between defoliation and climate variables for the period 1996-2007.
Because foliar nutrition data followed a non-normal distribution we used Spearman’s rank correlation coefficient to assess how foliar nutrition relates to climate variables and other vitality indicators. We also tested if previous year defoliation was correlated to current year residual chronology and vice versa. Additionally, we tested if previous year climate variables were correlated to current year vitality indicator values. All analyses were conducted in R statistical