**3. Results**

#### *3.1. Snow Depth, Temperature and Freeze–Thaw Cycle*

The snow depth in the shallow snow plot was 79% lower than that in the deep snow plot (Figure 1a). The snow depths were higher at the ends of the snow formation and snow coverage stages relative to those at the ends of the snowmelt stages in 2013 and 2014. The daily temperature was higher in the deep snow plot than that in the shallow snow plot, particularly during winter (Figure 1b). The mean temperatures at snow formation and snow coverage stages were below 0, but those at the snowmelt stages were above 0 during the winters in 2013 and 2014. The mean temperatures were significantly (*p* < 0.05) higher in the deep snow plots at all stages during the four years of decomposition, except those at the snowmelt stage and growing season in 2013 (Figure 1c).

The average freeze–thaw cycles in winters during the four years of decomposition were 0.41 and 0.57 times per day for the deep and shallow snow plots, respectively. For individual stages, the freeze–thaw cycles were significantly (*p* < 0.05) lower in the deep snow plots than those in the shallow snow plots at the snow coverage and snowmelt stages in 2014 as well as during the winter in 2016 (Figure 1d).

## *3.2. Mass Remaining*

Litter mass remaining varied greatly among litter species (*F* = 35.9, *p* < 0.001) but did not differ between deep and shallow snow plots (*F* = 1.8, *p* = 0.25, Table 3) over time. After four years of decomposition, 53%, 66%, 58% and 61% of litter mass were lost at the deep snow plots, respectively, and 55%, 65%, 55% and 60% of litter mass were lost at the shallow snow plots for fir, cypress, larch and birch litter, respectively (Figure 2).

**Table 3.** Results of repeated measures ANOVA testing for the effects of litter species and variation in snow depth over time.


Bold *p* values are significant (*p* < 0.05). AHC: acid hydrolysable components.

**Figure 2.** Mass remaining in decomposing litter in the deep and shallow snow plots. (**a**) Fir, (**b**) cypress, (**c**) larch, and (**d**) birch. SF: snow formation stage, SC: snow coverage stage, SM: snowmelt stage, and GS: growing season. \* *p* < 0.05, \*\* *p* < 0.01 and \*\*\* *p* < 0.001.

#### *3.3. Gravimetric Water Content*

The initial gravimetric water contents in the fir, cypress, larch and birch litters were 8%, 9%, 10% and 13%, respectively, but varied greatly over time (*F* = 775.7, *p* < 0.001; Table 3). The gravimetric water content in all the foliar litters showed an increasing tendency in 2013 but greatly decreased at the snow formation and snow coverage stages in 2014 (Figure 3). For all the litter species, the gravimetric water content notably increased in 2016 to 40%, 30%, 23% and 24%, respectively, in the deep snow plots after four years of decomposition. The variation in snow depth significantly (*F* = 13.7, *p* < 0.001; Table 3) changed the gravimetric water content in the decomposing litter, and higher (*p* < 0.05) gravimetric water contents were observed in deep snow plots for all litter species (Figure 3a–d).

**Figure 3.** Gravimetric water contents (±SE, *n* = 6) in decomposing litter in the deep and shallow snow plots. (**a**) Fir, (**b**) cypress, (**c**) larch, and (**d**) birch. SF: snow formation stage, SC: snow coverage stage, SM: snowmelt stage, and GS: growing season. \* *p* < 0.05, \*\* *p* < 0.01 and \*\*\* *p* < 0.001.

#### *3.4. Content of Acid Hydrolysable Components*

The initial contents of the acid hydrolysable components were 19%, 19%, 21% and 16% for the fir, cypress, larch and birch litters, respectively (Figure 4a–d). The content of acid hydrolysable components significantly (*F* = 274.1, *p* < 0.001; Table 3) changed over time, but this change did not differ among the litter species (*F* = 1.5, *p* = 0.23). During the four years of decomposition, the content of the acid hydrolysable components greatly decreased at the snow formation stage in 2012 and the growing season in 2016. The contents of the acid hydrolysable components also decreased during the growing season in 2013 to the snow coverage stage in 2014, as well as at the snowmelt stage and growing season in 2015. However, the content of acid hydrolysable components increased at both the snow coverage and snowmelt stages in 2013, while only at the snowmelt stages in 2014 and 2016

for all litter species. After four years of decomposition, the content of the acid hydrolysable components was only 1.8–2.7% in the deep snow plots across the litter species.

**Figure 4.** Contents of acid hydrolysable components (±SE, *n* = 6) in decomposing litter in the deep and shallow snow plots. (**a**) Fir, (**b**) cypress, (**c**) larch, and (**d**) birch. SF: snow formation stage, SC: snow coverage stage, SM: snowmelt stage, and GS: growing season. \* *p* < 0.05, \*\* *p* < 0.01 and \*\*\* *p* < 0.001.

Overall, the variation in snow depth did not significantly (*F* = 3.0, *p* = 0.083; Table 3) change the content of the acid hydrolysable components for the decomposing litter across species. However, for individual litter species, the effect of variation in snow depth on the content of acid hydrolysable components was greater for birch litter (*F* = 6.0, *p* = 0.016) compared with those for other foliar litters (*p* > 0.05; Table 4). During the four years of decomposition, the contents of the acid hydrolysable components were significantly higher (*p* < 0.05) in shallow snow plots than those in deep snow plots for all litter species at certain stages. At the end of the experiment, higher (*p* < 0.01) contents of acid hydrolysable components were also observed in the shallow snow plots for cypress and larch litters.

The path analysis results showed that snow depth was positively (*p* < 0.001; Figure 5) correlated with litter temperature (*r* = 0.38), gravimetric water content (*r* = 0.18) and MBC (*r* = 0.43) but negatively (*r* = −0.21, *p* < 0.001) correlated with the litter C/N ratio. The litter temperature and gravimetric water content did not have significant effects on either the C/N ratio or MBC (all *p* > 0.05), but both were negatively correlated with the content of acid hydrolysable components (*r* = −0.11, *p* < 0.01 for litter temperature and *r* = −0.25, *p* < 0.001 for gravimetric water content). The litter C/N ratio and MBC had, respectively, negative (*r* = −0.34) and positive (*r* = 0.15; both *p* < 0.001) effects on the content of the acid hydrolysable components.


**Table 4.** Results of repeated measures ANOVA testing for the effects of variation in snow depth over time for each litter species.

a Bold *p* values are significant (*p* < 0.05).

**Figure 5.** Path analysis shows direct and indirect effects of the variation in snow depth on the release of acid hydrolysable components. The color (blue: positive, and red: negative) and gray arrows denote significant and non-significant effects, respectively. The widths of the arrows and associated numbers (standard regression weights) represent the strength of the path coefficients, and the bold coefficients with asterisks are significant (\*\* *p* < 0.01 and \*\*\* *p* < 0.001). T: temperature, GWC: gravimetric water content, C/N: carbon to nitrogen ratio, MBC: microbial biomass carbon, and AHC: acid hydrolysable components. *n* = 576, *R*<sup>2</sup> = 0.531, *p* < 0.001 for the model.

#### *3.5. Remaining Acid Hydrolysable Components*

Overall, the remaining acid hydrolysable components consistently decreased over time (*F* = 333.7, *p* < 0.001; Table 3), and marked declines were found at the snow formation stage in 2012 and during the growing seasons in 2012 and 2016 for all litter species. However, the remaining acid hydrolysable components increased during the snow coverage and snowmelt stages in 2013 as well as the snowmelt stages in 2014 and 2016 (Figure 6a–d). After four years of decomposition, the remaining acid hydrolysable components were only 4–5% across the litter species.

The variation in snow depth had a significant (*F* = 6.8, *p* = 0.0094; Table 3) effect on the remaining acid hydrolysable components across litter species, and these effects were greater for cypress (*F* = 6.6, *p* = 0.011; Table 4) and birch litters (*F* = 6.4, *p* = 0.013) compared with fir and larch litters (both *p* > 0.05). During the four years of decomposition, the remaining acid hydrolysable components were significantly higher (*p* < 0.05) in shallow snow plots than those in deep snow plots for all litter species at certain stages. At the end of the experiment, higher (*p* < 0.001) remaining acid hydrolysable components were also observed in the shallow snow plot for larch litter (Figure 5c). The remaining acid hydrolysable components

in the decomposing litter were significantly correlated with the remaining litter mass for all litter species (all *p* < 0.001; Figure 7).

**Figure 6.** Remaining acid hydrolysable components (±SE, *n* = 6) in decomposing litter in the deep and shallow snow plots. (**a**) Fir, (**b**) cypress, (**c**) larch, and (**d**) birch. SF: snow formation stage, SC: snow coverage stage, SM: snowmelt stage, and GS: growing season. \* *p* < 0.05, \*\* *p* < 0.01 and \*\*\* *p* < 0.001.

**Figure 7.** Remaining litter mass versus acid hydrolysable components for each litter species. Adjusted *R*<sup>2</sup> values from Gaussian functions are shown for each litter species, and all *p* < 0.001 (*n* = 144 for each litter species).
