All statistical analyses were performed using the 3.1.1 version of
R (The R Foundation for Statistical Computing, Vienna, Austria).
ANOVA or the KruskaleWallis rank sum test were used when
checking for significant differences in parameter values, depending
on whether or not, respectively, normality and homoscedasticity
assumptions were met. Depending on the same assumptions, two
by two comparisons were made using either Tukey's honest significant
differences tests or pairwise Wilcoxon rank sum tests.
Multivariate datasets were analyzed using linear discriminant
analysis (LDA) for the CLPP data, non-metric multidimensional
scaling (NMDS) with the BrayeCurtis distance metric for the ARISA
fingerprints, and canonical correspondence analysis (CCA) for an
integration of all responses. In addition, analyses of similarities
(ANOSIM) were performed when checking for significant differences
between groups with multivariate data, and Mantel tests
were used to test for correlations between distance matrices
(BrayeCurtis for fingerprint data, Euclidean in the other cases).