Introduction

Nitrogen (N) and phosphorus (P) are limiting factors in the early stage of litter decomposition process (Bargali et al. 2015a, b; Güsewell and Gessner 2009). Nitrogen is usually retained in the initial stage of litter decomposition, whereas P concentration is more variable than N (Moore et al. 2006). The dynamics of N and P during litter decomposition can be strongly affected by soil fauna, and such effects could be moderated by environmental nutrient availability or micro-environmental conditions (Manzoni et al. 2010). Soil fauna are important decomposers that colonize litter and excrete nutrient-rich waste during litter decomposition, thus regulating the nutrient replenishment that occurs with mineralization and accelerating the decomposition process (Hättenschwiler et al. 2005).

Generally, the effects of local-scale environmental factors such as temperature and moisture on litter decomposition and the accompanying nutrient release process at a regional scale may be greater than litter quality (Bradford et al. 2016; Liski et al. 2003). Temperature was traditionally considered to have a significant effect on soil fauna populations, either directly or indirectly, via altering species-specific habitat suitability for soil biota (Simmons et al. 2009). Therefore, the contribution of soil fauna to nutrient release may be reduced by the low temperature in winter but promoted by the rising temperature in the following growing season (Li et al. 2014). Soil moisture availability is another key factor influencing the fauna community composition and function (Sylvain et al. 2014). Temperature and moisture have been identified as important moderators of nutrient dynamics during litter decomposition (García-Palacios et al. 2016). These moderators can affect faunal contribution to nutrient release during litter decomposition, but their effects may vary substantially across different types of ecosystems, where the environmental conditions (e.g., temperature and moisture) can vary strongly (Wang et al. 2010). Moreover, the effects of fauna on the nutrient release pattern may also be moderated by plant species (García-Palacios et al. 2013), as many soil faunas usually prefer species with higher litter quality (e.g., lower C:N and lignin:N ratios) (Gessner et al. 2010). However, as far as we know, the effects of soil fauna on N and P release patterns during litter decomposition and the factors controlling these processes are not yet well-assessed, limiting our knowledge of nutrient cycling during litter decomposition at both local and regional scales.

As a functional transition region and buffer zone of environmental changes, the ecotone between a dry valley and montane forest hinders the upward advance of arid regions (Bailey 2004). Montane forest ecosystems on the eastern Tibetan Plateau are usually subject to long periods of seasonal snow cover and frequent freeze–thaw cycles (He et al. 2015; Wu et al. 2014), which could lead to substantial variations in temperature and moisture availability between winter and the following growing season. In dry valley, however, the low precipitation and large differences in temperature in different seasons are likely to become the main limiting factors for soil fauna activities during litter decomposition. Compared with the adjacent dry valley and montane forest, the ecotone (the interface between these two ecosystems) exhibits milder freeze–thaw characteristics and greater moisture to support fauna activities. Previous studies have recorded higher fauna diversity in ecotones than in montane forest and dry valley (e.g., Wang et al. 1997), which may make a greater contribution to litter decomposition. We therefore hypothesized that the effect of soil fauna on N and P release during litter decomposition would be greater in the ecotone than in the montane forest and dry valley.

To test this hypothesis, a field experiment using litterbags with two different mesh sizes (3 mm and 0.04 mm) was conducted in montane forest, ecotone, and dry valley ecosystems along an elevation gradient on the eastern Tibetan Plateau. The effects of soil fauna on the dynamics of N and P were evaluated based on multiple sampling events during litter decomposition over a 1-year experiment. Two types of litter from the dominant plant species [fir (Abies faxoniana) and birch (Betula albosinensis) in montane forest; oak (Quercus baronii) and cypress (Cupressus chengiana) in ecotone; and cypress and clovershrub (Campylotropis macrocarpa) in dry valley] in each ecosystem were used in this study, with the aim of determining the actual effects of soil fauna on the N and P release patterns during litter decomposition in each ecosystem. The objectives of this study were (1) to determine whether the N and P dynamics in decomposing litter show differences between different types of ecosystems and whether soil fauna have effects on these processes; and (2) to evaluate the influence of plant species and local-scale environmental factors on these processes within each ecosystem. We focused only on the effects of soil fauna in this study, while the potential effects of microbial community and activities were not assessed. The results may provide insights about soil fauna effect on nutrient release during litter decomposition and the influencing factors on this process, further helping us to better understand nutrient fluxes from plant to soil under a scenario of expansion of arid land areas.

Materials and methods

Study sites

The study was conducted in the Long-term Research Station of Alpine Forest Ecosystems, which is located in Lixian County, Sichuan Province, southwest China (31.14°N–31.32°N, 102.53°E–103.26°E). Sampling sites in three types of ecosystems (i.e., montane forest, ecotone, and dry valley) were selected at altitudes ranging from 1450 to 3000 m a.s.l. The details of the study sites were described in Peng et al. (2015), and are briefly summarized here (Table 1).

Table 1 Summary for the study sites in the montane forest, ecotone, and dry valley ecosystems
Full size table

Methodology and experimental design

The effects of soil fauna on N and P release were determined using the widely employed litterbag method, utilizing litterbags with two different mesh sizes (0.04 and 3 mm). The small-mesh bags could exclude almost all fauna except for very small members of the micro-fauna (Swift et al. 1979). Litterbags with a 5 mm mesh are usually used to explore the effect of the whole fauna during litter decomposition (Swift et al. 1979; Wang et al. 2010). However, our previous experiments indicated that it is difficult to maintain complete samples placed in the litterbags using such a coarse mesh size, and small leaves of the selected litter species (e.g., fir, cypress, and clovershrub) together with the poor road conditions for driving in this alpine region will increase errors in calculating the litter mass loss. Moreover, because a 2 mm-mesh-size litterbag can include meso- and macro-fauna (Swift et al. 1979) and few fauna larger than 3 mm were observed in our previous investigation (Tan et al. 2010), we used 3.00 mm-mesh-size litterbags in the present study.

In October 2013, samples of naturally senesced plant foliar litter were collected from the uppermost layer in the sampling plots in the montane forest (fir and birch), ecotone (oak and cypress) and dry valley (cypress and clovershrub) areas. The foliar litter samples were air-dried for at least 2 weeks at room temperature to prevent structural damage during subsequent oven-drying. Subsamples of the foliar litter from each species were oven-dried at 65 °C for 72 h to calculate the moisture correction factor, and were then milled to pass through a 0.3-mm mesh to analyze the initial chemistry. The air-dried foliar litters (equivalent to 10 g oven-dry mass per litterbag) were placed in two types of 20 × 20 cm nylon litterbags: with a small mesh size (0.04 mm on both sides) or large mesh size (0.04 mm on the bottom side and 3.00 mm on the reverse side). The litterbags were then sealed.

We explored 3 plots (5 × 5 m) in each of the ecosystem for placement of the litterbags, and the distance between each individual plot was approximately 100 m. In total, 30 litterbags (two mesh types × three replicates × five sampling dates) containing each species (except for cypress, for which 60 litterbags were filled, with 30 used in the ecotone and 30 in the dry valley plots) were placed on the floor of the three selected sampling plots on November 12, 2013. The air temperature 20 cm above of the ground and the temperature in the litterbags were measured every 2 h in each of the three ecosystems between November 12, 2013 and October 23, 2014 (see Fig. S1) using DS1923-F5 iButton loggers (Maxim Integrated Products, Inc., Sunnyvale, CA, USA). To characterize the effects of temperature, we calculated the daily average temperature (DAT), the positive accumulated temperature (PAT), and the negative accumulated temperature (NAT) (Yue et al. 2016b). Soil moisture was measured in soil core samples (5 cm below soil surface) with a diameter of 5 cm collected at each sampling time from each plot (i.e., 3 replicates for each ecosystem) during the experiment period. The soil samples were dried for 12 h at 105 °C to determine soil moisture: soil moisture (%) = (wet weight − dry weight)/wet weight × 100% (see Fig. S2). We divided the entire year into five periods based on the changes in freeze–thaw cycles and seasonal dynamics determined from our previous field observations in the montane forest (Wu et al. 2010; Yue et al. 2016a), and litterbags were collected at the end of each period. Specifically, litterbags were randomly retrieved from each plot on December 22, 2013 (the pre-freezing period, 41 day exposure), March 9, 2014 (the freezing period, 118 day exposure), April 24, 2014 (the thawing period, 164 day exposure), August 20, 2014 (the growing season, 282 day exposure) and October 24, 2014 (the late growing season, 347 day exposure). The retrieved small-mesh-size litterbags were air dried upon arrival at laboratory, while large-mesh-size litterbags were immediately used to extract fauna, and were then, together with the air-dried small-mesh-size litterbags, oven-dried at 65 °C for 72 h to determine the dry mass and calculate the remaining dry mass, and the remaining portion was used to test element concentrations. For soil fauna collection, briefly, macro-fauna were directly picked up by hand and meso-/micro-fauna were extracted from the foliar litter using the modified Tullgren extractors (Tan et al. 2010). All collected soil faunas were placed in 75% alcohol and then identified to family level under microscope (Yin et al. 1998).

Chemical analyses and calculations

The oven-dried foliar litter was ground through a 0.3 mm sieve, and the concentrations of total organic C, total N, and total P were then determined. Briefly, C was determined using the dichromate oxidation-ferrous sulfate titration method; N was determined using the Kjeldahl method; and P was determined using the molybdenum-blue colorimetric method (He et al. 2016a; Lu 1999). Details of the initial litter quality were described in Peng et al. (2015). For each litter species, the remaining mass (RMt), N or P release rate for each period (Rt), and the rate of N or P release mediated by the fauna for each period (Cfau) were calculated as follows (Xin et al. 2012; Yang and Chen 2009):

$$ RM_{t } \left( \% \right) = \frac{{M_{t} }}{{M_{0} }} \times 100\% $$
$$ R_{t} \left( \% \right) = \frac{{M_{t - 1} C_{t - 1} - M_{t} C_{t} }}{{M_{0} C_{0} T_{t} }} \times 100\% $$
$$ C_{fau} \left( \% \right) = R_{lt} - R_{st} $$

where M0 is the initial litter dry mass; Mt is the dry mass at sampling time t (t = 1, 2, 3, 4, 5); C0 is the initial concentration of N or P; Mt and Mt−1 are the remaining dry mass at sampling time t−1 and t, respectively; Ct−1 and Ct are the concentration of N or P at sampling time t−1 and t, respectively; ΔTt is the time interval in day between sampling time t−1 and t (t = 1, 2, 3, 4, 5); and Rlt and Rst are the release rate of N or P in large-size litterbags and small-size litterbags at sampling time t, respectively (t = 1, 2, 3, 4, 5). The N or P release rate for the entire winter (Rw), the entire growing season (Rg), and the entire year (Ry) were calculated as follows:

$$ R_{w} = \frac{{M_{0} C_{0} - M_{3} C_{3} }}{{M_{0} C_{0} T_{w} }} \times 100\% $$
$$ R_{g} = \frac{{M_{3} C_{3} - M_{5} C_{5} }}{{M_{0} C_{0} T_{g} }} \times 100\% $$
$$ R_{y} = \frac{{M_{0} C_{0} - M_{5} C_{5} }}{{M_{0} C_{0} T_{y} }} \times 100\% $$

where M0 and C0 are the initial litter dry mass and the initial concentration of N or P, respectively; M3 and M5 are the litter dry mass at the third and fifth sampling time, respectively; C3 and C5 are the concentrations of N or P at the third and fifth sampling times, respectively; Tw is the time interval in days for the entire winter (i.e., from the beginning to the third sampling time), Tg is the time interval in days for the entire growing season (i.e., between the third and fifth sampling time), and Ty is the days of the whole study year. The N or P release rate mediated by the fauna for the entire winter (C′fau), the entire growing season (C′′fau), and the entire year (C′′′fau) were calculated as follows:

$$ C_{fau}^{{\prime }} \left( \% \right) = R_{lw} - R_{sw} $$
$$ C_{fau}^{{{\prime \prime }}} \left( \% \right) = R_{lg} - R_{sg} $$
$$ C_{fau}^{{{\prime \prime \prime }}} \left( \% \right) = R_{ly} - R_{sy} $$

where Rlw and Rsw, Rlg and Rsg, and Rly and Rsy are the release rates of N or P in large-size litterbags and small-size litterbags for the entire winter, the entire growing season, and the entire year, respectively. We used individual density (I) and Shannon–Wiener index (H) to describe fauna community, which were calculated as follows:

$$ I = {\raise0.7ex\hbox{$n$} \!\mathord{\left/ {\vphantom {n M}}\right.\kern-0pt} \!\lower0.7ex\hbox{$M$}} $$
$$ H = - \mathop \sum \limits_{i = 1}^{S} P_{i} \ln P_{i} \quad \left( {P_{i} = {\raise0.7ex\hbox{${n_{i} }$} \!\mathord{\left/ {\vphantom {{n_{i} } N}}\right.\kern-0pt} \!\lower0.7ex\hbox{$N$}}} \right) $$

where n and M are the number of fauna and the remaining litter dry mass in each litterbag, respectively; ni are individuals of group at family level (e.g., Isotomidae, Prostigmata, and Oribatida etc.; Table S1) i and N totals of the groups at family level (Whittaker 1972), and S is the number of groups.

Statistical analyses

An independent sample T test was performed to compare N or P release rates between the large- and small-mesh litterbags in each period for a specific species within each ecosystem. Three-way analysis of variance (ANOVA) was used to test the effects of litter species, litterbag mesh size, the decomposition period, and their interactions on the remaining litter dry mass, N concentration, P concentration, and N/P ratio within each ecosystem. We used a repeated-measure ANOVA with Greenhouse–Geisser adjustment to test the individual and interactive effects of litter species and decomposition period on N or P release rate mediated by soil fauna within each ecosystem. When the ANOVA results were significant at P = 0.05, differences between means were subsequently explored using Tukey’s honestly significant difference (HSD). A one-sample T test was employed to determine whether the rate of N or P release mediated by the fauna during each decomposition period was significantly different from zero. Pearson correlation or linear regression analyses were conducted to test the relationship between the litter initial chemistry or environmental factors and the rate of N or P release mediated by soil fauna within each ecosystem. All statistical analyses were performed using the SPSS 20.0 software package for Windows (SPSS Inc., Chicago, IL, USA).

Results

Remaining litter mass and fauna community

Foliar litter in the dry valley generally decomposed more rapidly than in the ecotone and the montane forest during comparable periods regardless of litterbag mesh sizes (Fig. S3). The percentages of remaining mass in the dry valley, ecotone, and montane forest were 44.7–72.5%, 51.1–71.7%, and 74.7–80.9%, respectively, according to litter species and litterbag mesh size. The individual density of fauna in all litterbags showed an increase pattern with the proceeding of litter decomposition except in clovershrub litterbags, which was the highest in the pre-freezing period and then decreased until the growing season (Fig. 1a). The Shannon–Weiner diversity index of fauna in the litterbags of each litter species was the highest in the growing season except for that in clovershrub litterbags, which was the lowest during the growing season (Fig. 1b).

Fig. 1
figure 1

Dynamics of fauna a individual density and b Shannon–Weiner diversity index in the litterbags of each litter species (mean ± SE, n = 3). Asterisks indicate significant (P < 0.05) differences between different litter species at each sampling event

Full size image

Release rates of N and P mediated by soil fauna

Foliar litter N concentration increased in all the plant litter species regardless of the ecosystem and litterbag mesh size, whereas the P concentration and N/P ratio fluctuated during the 1-year experiment (Fig. 2). The effects of litterbag mesh size, litter species, and decomposition period on the concentrations of N and P or N/P ratio were commonly significant with each ecosystem (Fig. 2). Significant differences of N and P release rate were observed between large and small mesh size litterbags (Fig. S4), and the repeated measures ANOVA analysis suggested that both the species and decomposition period could significantly (P < 0.05) influence the contribution of soil fauna to N or P release rate, but such effects varied among different ecosystems (Table 2). Overall, the effects of soil fauna on both N and P release rate were most strong in the dry valley, followed in the ecotone and the montane forest (Fig. 3). Over 1 year of incubation, fauna in the dry valley, the montane forest, and the ecotone contributed release percentages of 15.6–37.3%, 0.54–6.4%, and − 3.7–37.3% of the initial N content, respectively, depending on the litter species. Likewise, the release of P mediated by fauna ranged from − 12.4 to 40.9% of the initial content, depending on the plant litter species and ecosystem type (Fig. 3). Moreover, our correlation analyses suggested that the effects of soil fauna on N release rate were remarkably (P < 0.05) related to litter initial quality, but not the P release rate in the montane forest and ecotone (Table 2). However, the release rate of N mediated by soil fauna showed an insignificant relationship with environmental factors such as soil pH, moisture and temperature, but temperature appeared to be an important moderator of the effects of fauna on the P release rate meditated by soil fauna (Table 3).

Fig. 2
figure 2

Dynamics of foliar litter N (a–c) and P (d–f) concentrations and the N/P ratio (g–i) during litter decomposition in the montane forest, ecotone, and dry valley throughout the entire year (mean ± SE, n = 3). The results of three-way ANOVA with the factors of decomposition period, litter species, and litterbag mesh size within each ecosystem are shown. For simplification, only significant (P < 0.05) effects or interactions are shown. Asterisks indicate statistically significant (*P < 0.05; **P < 0.01; ***P < 0.001) difference between the large- and small-mesh-size litterbags for a given litter species in a specific period. PP pre-freezing period, FP freezing period, TP thawing period, GS growing season, LGS late growing season

Full size image
Table 2 Results of repeated-measures ANOVA for the effects of litter species and the decomposition period on the N or P release rates mediated by the fauna (Cfau) over the 1-year experiment in different types of ecosystems
Full size table
Fig. 3
figure 3

N (a–c) and P (d–f) release rates mediated by the fauna(Cfau, % day−1) during litter decomposition in the montane forest, ecotone, and dry valley in different periods over the 1-year experiment (mean ± SE, n = 3). Different lowercase letters indicate significant (P < 0.05) differences between different periods for a given litter species, and different uppercase letters indicate significant (P < 0.05)differences for a specific litter species between the winter, entire growing season and entire year. Asterisks indicate differences in the rate of N or P release mediated by the fauna that are significantly different from zero (*P < 0.05; **P < 0.01; ***P < 0.001) in each period. PP pre-freezing period, FP freezing period, TP thawing period, GS growing season, LGS late growing season, EW entire winter, EGS entire growing season, EY entire year

Full size image
Table 3 Determination coefficient (R2) for regression analyses with N or P release rate mediated by the fauna (Cfau) of the entire year as response variable and fauna community or environmental factors as predictor variable in different types of ecosystems
Full size table

Discussion

Inconsistent with the original hypothesis, our results suggested that the effects of soil fauna on the average release rates of N and P during litter decomposition in the dry valley were greater than in the montane forest and ecotone across the entire year. Such different responses of N and P release rates to the impacts of soil faunas may be attributed to the fact that the fauna effects can be significantly moderated by plant species, ecosystem type, and decomposition period, which was approved by the strong relationships between the effects of fauna on N and P release rates and ecosystem types and seasonal dynamics. Specifically, the N release rate mediated by fauna was more sensitive to the effects of the plant species (i.e., initial litter chemical traits), while the P release rate mediated by fauna was subject to the effects of local-scale environmental factors (e.g., temperature and moisture) to a greater extent.

Through synthesizing global data using litterbag method, García-Palacios et al. (2013) found that soil faunas increase litter decomposition rate by an average of 37% at the global scale, but their effect varies significantly among different types of ecosystems that have various plant species and different temperature and moisture regimes. Likewise, in this study, we found that plant species, which is characterized by different initial chemical traits, had significant but different influences on N and P release rates (Tables 2, 4). Litter quality is a function of the chemical composition of the litter, which is usually reflected as different plant species and varies at different decomposition phases (He et al. 2016b; Yu et al. 2015). Species with high litter quality could supply abundant resources for fauna, maintaining a high total diversity of soil fauna and, thus, increasing the decomposition rate of litter (Mueller et al. 2016). Consistent with these previous studies, the influence of soil fauna on the N release rate in decomposing litter was significantly (P < 0.05) moderated by initial litter quality, especially initial C/N ratio and lignin/N ratio (Table 4). However, the Cfau of P was seldom affected by the initial litter quality in the montane forest and ecotone, and only significant relations between Cfau of P and litter initial quality were found in dry valley (Table 4). A plausible explanation for this inconsistency may be that the relative control over litter decomposition and the accompanying release of elements due to biotic and abiotic factors can change dramatically during the process of decomposition (García-Palacios et al. 2016), and the influence of initial quality on element release patterns can vary at temporal scales, as the local-scale environmental factors such as temperature and soil moisture can be of great importance for the decomposition process (Bradford et al. 2014, 2016).

Table 4 Pearson correlation coefficients (ρ) between N or P release rate mediated by soil fauna (Cfau) for the entire year and initial litter chemistry within each type of ecosystem
Full size table

Physicochemical conditions can determine soil abiotic characteristics, which influence the activity and composition of soil fauna (Wardle and Lavelle 1997). Temperature is one of the most important environmental factors that moderate the effects of soil fauna on litter decomposition directly and/or indirectly. Temperature can influence activities in decomposing litter not only directly but also indirectly through its effect on plant physiology, soil biogeochemistry and geographic patterns of soil substrate (Wardle et al. 2004). Along with the elevation gradient, ecosystem structure and abiotic and biotic factors vary remarkably with changes in temperature. The Cfau of P was negatively corrected with the DAT but positively related with the PAT, indicating that a higher temperature in an ecotone may promote the effects of the fauna on P release. As well, the different soil biogeochemistry conditions can also be important regulators. For example, the relatively abundances of soil N and P in different ecosystems can be limitations for litter decomposition and its associated nutrient release process (Hobbie and Vitousek 2000). Soil P concentration and the proportions of P with other nutrients such as C/P and N/P ratios were major moderators that can inhibit or promote litter decomposition rate by providing a balanced nutritional pool for decomposers such as soil fauna (Moore et al. 2011; Xu and Hirata 2005). This was supported by our results that soil fauna community groups and diversities varied significantly among different ecosystems (Fig. 1, Table S1).

Likewise, soil moisture has also been found to be an important factor related to litter decomposition at both spatial and temporal scales (García-Palacios et al. 2016). In cold biomes, such local-scale environmental factors are likely to be especially important (Aerts 2006), and warmer and wetter conditions are expected to promote more rapid decomposition and release of carbon and element found in such biomes (Hobbie and Stuart Chapin III 1996) via greater fauna stimulating effects. However, no significant effect of soil moisture on the release rate of either N or P mediated by fauna was observed in the present study. Nevertheless, we found that the release rates of N and P in decomposing litter were highest in the dry valley, especially in winter. A plausible explanation for this inconsistency may be related to the differences in the microclimates and the litter species that would be more preferable for fauna community in winter, as fauna individual density was highest in the clovershrub litters in winter. Although the release patterns of N and P appear to be in line with this explanation, the contribution of the fauna to these release rates showed little response to the investigated environmental factors; only the soil and air temperature in the montane forest exhibited significant correlations with rate of P release mediated by the fauna. Furthermore, the recycles of wet and dry as well as snowmelt in winter dry valley may directly contribute to foliar litter nutrient releases which indirectly provide a source of nutrient for soil fauna, leading to a higher N and P release rate mediated by soil faun in winter (Cleveland et al. 2004). However, as we used different litter species within each type of ecosystem, it is thus hard to disentangle the exact influence of environmental factors such as temperature and moisture on soil fauna effects on N and P release rate in the present study.

Moreover, the three ecosystems investigated in this study show distinct seasonal dynamics, particularly in winter, which may have significant influences on N and P release rates mediated by soil faunas. The montane forest is subject to long periods of snow cover and frequent freeze–thaw cycles (He et al. 2015; Wu et al. 2014), whereas the dry valley is characterized by frequent alternation between wet and dry periods (Peng et al. 2015). However, the ecotone is the interface of the montane forest and the dry valley and displays both types of characteristics. Such typical environmental factors cause the effects of soil fauna on N or P release within each ecosystem not only vary significantly at a temporal scale but also to differ between different ecosystems in the same period. However, the effects of litter species may confounding the actual effects of ecosystem type in the present study, as we used different litter species within each type of ecosystem. Furthermore, another potential important moderator on soil fauna impacts on N and P release rates is microbial communities and activities (Heděnec et al. 2013). For example, the most abundant soil meso-fauna oribatid mites in terrestrial ecosystems can increase microbial activity (Heděnec et al. 2013), thus stimulate litter decomposition and nutrient release process. This was approved by our study, as Prostigmata and Oribatida were found to be the dominant fauna communities (Table S1). However, as we only focused on soil fauna in this study, the impacts of soil microbial activity need further investigations.

Conclusions

The soil fauna showed obvious effects on the N and P release rates in different decomposition periods and ecosystem types. The N release rate mediated by soil fauna was likely to be more pronounced in the dry valley, while the effects of soil fauna on P release rate were manifest both in the montane forest and the dry valley. Moreover, the impacts of soil fauna can vary remarkably between different decomposition periods. Our study also showed that foliar litter initial quality had significant effects on Cfau of N, but had either limited or weak effects on Cfau of P. Furthermore, local-scale environmental factors, such as temperature, were important moderators of the N and P release rates mediated by the fauna, but varied between different ecosystems. However, as we used different litter species within each ecosystem, the exact effects of ecosystem type expressed by different environmental factors was hard to disentangle, which needs further investigations. Overall, our results suggested that fauna can significantly affect litter mass loss and the associated N and P release dynamics, and the N release rate mediated by fauna was likely to be more sensitive to the effects of plant species (i.e., initial litter chemical traits), while the P release rate mediated by fauna may be subject to the effects of local-scale environmental factors (e.g., temperature) to a greater extent.