*2.5. Data Analysis*

We first examined the effect of snow reduction and its variation over time among the litter species using a three-way analysis of variance (ANOVA). Differences between the control and reduced snow cover plots were assessed via paired t-tests or Wilcoxon signed-rank tests in MATLAB R2012a (MathWorks Inc., Natick, MA, USA) if the sample sizes were unequal. Then, we compared the relationships between the remaining litter mass and the contents of the total humus, humic acid and fulvic acid. We also monitored certain approximate chemical compounds and mineral elements in this long-term experiment and used a partial least squares analysis [39] in SIMCA 14.0 (MKS, Umeå, Sweden) to distinguish the importance of the factors that may impact the humus content. The relative influence of each independent explanatory variable (chemical compounds and mineral elements) were estimated using the variable importance of projection (VIP) and values greater than 1 were considered relevant and significant for explaining the humus content [40].
