Results

Decomposition rates and stabilization in 10 mountains

Across the 10 mountains, the decomposition rate (k ) ranged from 0.002 to 0.05 (0.02 on average), and the stabilization factor (S ) ranged from 0.006 to 0.35 (0.13 on average). JFS experienced the fastest mass loss (k = 0.02), which was significantly higher than that in six mountains on its east side (Fig. S2). The highest mean Soccurred in EMS and YMT (0.18), which was significantly higher thanS in GMT and the five eastern mountains (Fig. S3). There were no significant correlations between k and S (except DBS) (Fig. S4).
Although both significant and non-significant correlation existed, there was a consistent pattern in each mountain that k decreased along elevations (but see a slight hump-shape in YMT), and an opposite trend for S (Fig. 2). For k , the significant elevational pattern was mainly found for four western mountains (GMT, YMT, EMS, and JFS) and the northmost mountain (DBS). For S , the statistically significant trend occurred in three western mountains (GMT, YMT, and JFS), and the southernmost mountain (BWL).

The importance of microclimate in decomposition

Among 10 climatic and non-climatic variables, soil microclimate related variables, mainly Temp, TempV, and MoisV, emerged as the most important factors in controlling litter decomposition in the western mountains (Figs 3-4, Table S2). Across all mountains, k was mainly impacted by soil temperature (coef. = 0.48, P < 0.001) and its variation (coef. = 0.36, P < 0.001), while S was subjected to significant effects from all microclimatic factors, with particularly strong relation with soil temperature (coef. = -0.46, P < 0.001) and the variation of soil moisture (coef . = -0.36, P < 0.001) (Fig. 5).

The effects of non-climatic drivers in decomposition

Non-climatic factors, i.e., tree diversity, soil (pH and P), and microhabitat (litter thickness, canopy cover, and slope) emerged as important as soil microclimate in western mountains and the northmost mountain (DBS) (Figs 3-4, Table S2). Among mountains, three representative factors (tree diversity, soil pH, and slope) showed substantially spatial variations in its relations with k andS (Fig. 6). Across all mountains, tree diversity ranged from 4 to 93 and increased k significantly (coef. = 0.25, P< 0.001). Soil pH (ranged from 3.06 – 7.16) and slope (0.25 – 42.5) had no significant effect in this respect. For S , we found no significant effects of tree diversity but a significantly negative association with soil pH (coef . = -0.17, P< 0.001) and slope (coef . = -0.14, P < 0.001).