**4. Discussion**

SOM sequesters twice as much carbon as the total carbon contained in the atmosphere and terrestrial plants [41], and forecasting its feedback to climate change requires a clear understanding, not only of SOM decomposition [12] but also of SOM formation. Recent studies have changed our traditional understanding of the mechanisms underlying SOM formation [7,13]. A considerable amount of carbon is known to be transferred from plant litter to mineral soils at the early stages of litter decomposition [8]. However, the mechanisms underlying the accumulation of humus in the decomposing litter substrate remain unclear.

#### *4.1. Is Humus Really Formed at a Very Late Stage of Litter Decomposition?*

The conventional view suggests that plant litter decomposes to a "limit value", beyond which, litter cannot be decomposed and recalcitrant residues are humified and retained in stable SOM; therefore, humus is considered to form at a very late stage of litter decay [1]. In this study, we found that the initial humus content accounted for 8–13% of the litter mass (Figure 2), suggesting a high degree of humification in freshly senesced foliage. Furthermore, the humus content consistently increased as decomposition proceeded ( *R*<sup>2</sup> = 0.40, *p* < 0.001; Figure 3a) and eventually reached 16–21% after four years of humification. Our results demonstrated that considerable humus accumulation occurred, and it sequestered carbon in the remaining litter substrate and did not allow for the net release of carbon from litter during the early stages of decomposition (four years) in this high-altitude forest.

Increasing evidence suggests that recalcitrant compounds (e.g., AUR) are not selectively preserved during litter decomposition [42]. In contrast, labile components increase microbial substrate use efficiency and promote SOM formation and persistence [9,13]. In this experiment, we monitored certain chemical compounds and mineral elements during litter decomposition and performed a partial least squares analysis to distinguish the importance of these factors that may impact the humus content. We found that the effect of AUR was greater than that of labile carbon, although neither was significant (Figure 4b). However, this result from our litterbag experiment was insufficient to support the microbial efficiency-matrix stabilization theory [13]. Our findings indicated that the litter Mn concentration was the dominant factor (VIP = 2.2) and positively related to the humus content (the coefficient was 0.35; Figure 4a), which suggested that Mn not only stimulated litter decomposition [43] but also promoted humus accumulation, thus supporting the hypothesis proposed by Berg et al. [44]. Our results also implied that bonding between humus and mineral elements (e.g., calcium and iron) may be more important than the approximate chemical compounds (e.g., water-soluble substances and AUR) in maintaining the organo-mineral stabilization of newly formed humus during the early stages of litter decomposition.

Fulvic acid (5.6–10.4%; Figure 6) formed earlier than humic acid (1.9–2.4%; Figure 5) in the decomposing foliar litter. However, the newly formed fulvic acid was not stable and degraded at earlier stages of humification (remaining mass >70%; Figure 3c), whereas humic acid greatly increased before the remaining mass reached 50–60%. It is not surprising that FA and HA were most present in the alkali-extractable substances since individual, lower molecular weight organic substances are extracted from fresh plant sediment. A study by Qualls et al. [45] also found that humic acid increased from 2.1% to 15.1% and that fulvic acid decreased from 7.5% to 6.1% in 3-year-old pine (*Pinus strobus* L.) litter compared with the fresh sample. These results show that humus accumulation during early litter decay was primarily dependent on the increase in humic acid. Our findings emphasized that more long-term ecological research should be performed to address how carbon is sequestered in soils instead of how carbon is lost from plant litter in a changing climate.

#### *4.2. Will Reduced Snow Cover Alter This Carbon Sequestration Process in High-Altitudinal Ecosystems?*

Winter snow cover has decreased in cold biomes [17], and numerous studies assessed its influence on the decomposition of plant litter; however, the results indicated high uncertainty. For example, experiments found that reduced snow cover decreases the litter decomposition rates by 5–47% in forests [18,46] and tundra [47], and a surprising threefold lower decomposition rate was reported under reduced snow cover in a subalpine meadow [19]. However, other experiments suggested that reduced snow cover has only a minor [28,48] or even null effect [49,50] on litter decomposition in subarctic and alpine ecosystems. Indeed, in our mid-latitudinal alpine forest, we also found that reduced snow cover decreased the loss of litter mass but only during the first two years and not during the subsequent periods.

In fact, the humified and not the decomposed components in the decomposing litter are more important for SOM formation and stabilization. Sequestering more carbon into the SOM with a longer mean resistance time [51] rather than allowing it to decompose is more effective under a changing climate [5]. Unfortunately, the mechanisms involved in SOM formation and its feedback on climate change are largely unknown. In this study, we found that reduced snow cover decreased the humus content in decomposing litter during the first three years but increased it in the fourth year (Figure 2). This result suggested that shortterm assessments may erroneously estimate the real influence of the reduction in winter snow cover, which has a more complex effect on soil carbon sequestration, as indicated by previous snow manipulation studies. Thus, long-term ecological studies should be performed to decrease the uncertainty and accurately evaluate the SOM–climate feedback.
