**4. Discussions**

We analyzed how the microbial community changes during litter decomposition in the natural forest floor, intermittent stream, and headwater stream by PLFA analyses. We found that habitats didn't affect total PLFA concentrations but strongly influenced microbial community composition during the litter decomposition. Meanwhile, microbial total PLFA increased significantly with time among habitat types. Some specific microbial groups showed different trends with decomposition time in three habitats. Meanwhile, linear regression analysis between PLFA concentration and different environmental variables in the three habitats showed that some PLFA were significantly affected by environmental factors (Table 2). For example, bacteria were significantly affected by temperature and pH while fungi didn't show the same pattern. In general, fungi are assumed to be less sensitive to changes in temperature and moisture than bacteria due to chitinous cell walls [3]. In addition, microorganisms ultimately led to direct litter mass loss. Some communities of microbial decomposers such as bacteria (G+ and G− bacteria), fungi, and eukaryote showed a strong correlation with litter mass loss in the headwater stream, which verifies our hypothesis. This indicated that the environmental differences among habitats will affect the development of microbial communities, and ultimately alter litter decomposition [28,29].


**Table 2.** Linear regression for three habitats between environmental factors (Temperature, Rainfall, pH, Flow rate) and the concentrations of microbial PLFAs.

The sensitivity of microorganisms to the environment is different, which is greatly reflected in our results. First, the total PLFA concentrations in the intermittent stream and the natural forest floor were significantly lower than that in the headwater stream during the 62nd day of decomposition. This may be that we observed two transient droughts in intermittent streams between days 28th and 62nd of decomposition. At this time, the litter was exposed to solar radiation. The disappearance of water means grea<sup>t</sup> physiological stress to microbial decomposers and results in marked decline of its biomass [30]. Moreover, some differences of microbial communities in the three habitats are mainly reflected in fungi, G+ bacteria, and eukaryotes. First, the fungal PLFA concentrations in the natural forest floor were significantly higher than those in other habitats. However, fungal PLFA concentrations were little difference between the headwater stream and the intermittent stream during the whole decomposition period. It is generally accepted that fungi are the main decomposers of refractory substances such as lignin [31]. Once entrained in the ecosystems, litter is rapidly colonized by microorganisms [14]. However, under the conditions of headwater streams and intermittent streams during flow, leaf litter was more vulnerable to leaching [32], which may reduce the colonization of fungi acting on lignin. Some research results also showed that water flow is related to the growth and activity of fungi [17]. Hypoxia conditions in streams may limit the respiration of fungal organisms and inhibit mycelial growth [30]. Meanwhile, the research showed that intermittent stream should be dominated by fungal species with traits of higher desiccation resistance [33]. This makes the fungal concentrations of the intermittent stream similar to that of the headwater stream even under drought. Second, the difference in bacterial concentrations among habitats is mainly reflected in G+ bacteria. Some studies associate G+ bacterial PLFA with stress conditions found that G+ bacteria may be dominant in the condition that the available C is relatively less [34]. Thus, the higher G+ bacterial PLFA in the headwater stream and the intermittent stream suggested lower concentrations of available C received by aquatic ecosystems from land [35]. Meanwhile, the difference in ratios of biomarkers among habitats also showed the different flow of energy and nutrients through the microbial community in different ecosystems [36,37].

Total PLFA concentrations increased significantly with time among habitat types. This may be that some microbial groups (including bacteria, G+ bacteria and actinomycete) also increased significantly with decomposition time in the three habitats. Although the PLFA

concentrations of fungi and eukaryotes decreased significantly at the end of decomposition, their concentrations also increased on the whole. It is generally accepted that fungal biomass increases exponentially, and it is later stabilized or even decreases [38]. However, bacteria play a predominant role in the later stages of litter decomposition [39]. Our results also verify this statement. In addition, Otaki et al. [40] also found similar results in a three-year litter decomposition study. Their study found that the bacterial biomass increased steadily until the end of the experiment while fungal biomass reached its peak in the first year. In a four-month study (from the late summer to the early winter periods), Wilkinson et al. [41] found that bacterial and fungal biomass increased through time on spruce litter in Germany, which is also consistent with our results.

The analysis of individual PLFAs in terms of the whole decomposition period considered revealed a major change in the composition of the microbial communities in the three habitats. It showed that compared with stream conditions, similar strains acting on litter decomposition were more concentrated on the forest floor and intermittent stream. It has been reported that the decomposition environment of stream litter is being influenced by more complex factors such as water chemistry, temperature, velocity, and turbulence, which may result in less individual PLFAs in the whole decomposition period [42]. Under the same decomposition period, we identified more individual PLFAs in the later stage of decomposition. And in the early stage of decomposition, the number of individual PLFA with significant differences among the three habitats was less. This may be due to the large environmental differences among habitats, which makes different microorganisms act on litter decomposition. What is more, their difference was mainly reflected between the forest floor and the aquatic system. The possible reason was that the environmental difference between the two ecosystems may be mainly reflected in the temperature. Temperature is the main driving factor of microbial activity [42,43]. With the increase of temperature, the water temperature is relatively stable, while the atmospheric temperature fluctuates greatly. This led to the difference in individual PLFA between forest floor and water environment habitat. In contrast, in the later stage of decomposition, the difference in individual PLFA is gradually reflected among the three habitats. It was mainly reflected in the headwater stream and the intermittent stream. During this period, the intermittent stream has been in a dry state for a long time, and the litter is exposed to the soil surface. Temporary stress events, such as high temperatures or desiccation, result in dynamic changes in microbial communities over time [44]. This may explain our results.

Litter decomposition is often thought to be regulated by abiotic factors such as microclimate and chemical quality [45]. However, this view changed considerably with the development of molecular tools. Recent studies showed that soil microbial communities differ substantially over the litter decomposition course and space [46,47]. Spatial variation of a rather basic microbial parameter, such as microbial biomass, can be an important determinant of decomposition [48]. Though we did not find direct significant correlations between total PLFA and litter mass loss in three habitats. However, some communities of microbial decomposers showed a strong correlation with litter mass loss. This suggests that, in addition to abiotic factors and potential nutrient transfer among habitat types, the development of some microbial decomposers also drives the decomposition of litter, which further lead to different litter decomposition rates [49]. Among them, the PLFA concentrations of microbial groups showed the highest correlation with mass loss in the headwater stream. Meanwhile, we found fewer common individual PLFAs in the headwater stream throughout the decomposition. This indicated that the diversity of microorganisms is more abundant in the stream environment. A tremendous diversity of microbes contributes to litter decomposition and interacts in a cross-kingdom functional succession of communities [46]. What is more, linking decomposer identity to decomposition could provide a better understanding of the relationship between microbial community structure and leaf litter decomposition.
