**3.5 Decomposition traits of leaf litters of three dominant** *Stipa* **species**

The remaining mass of leaf litters decreased with decomposition time and showed significant differences among three *Stipa* species (**Figure 7**). At the end of decomposition experiment, the remaining masses of leaf litters of *S. bungeana*, *S. grandis* and *S. przewalskyi* were 64.47%, 61.53% and 65.78%, respectively (**Table 1**).

Different lowercase letters in the same column indicate significant differences (*P <* 0.05).

During 2 years' decomposition process, variations of nutrient concentration were affected by the nutrient type (**Figure 8**). In detail, concentrations of carbon and nitrogen showed species-specific fluctuations with decreasing tendency among three *Stipa* species. In contrast, phosphorus concentrations in leaf litters were averaged doubled. There were significant differences in C:N ratio and nutrient accumulation index (NAI) of leaf litters among three *Stipa* species (**Table 2**).

### **Figure 5.**

*Correlations between ANPP and annual precipitation (a) and mean annual temperature (b) during 1992–2011 at Yunwushan. AP means annual precipitation and MAT represents mean annual temperature.*

**143**

**Table 1.**

**Figure 7.**

**Figure 6.**

*variables.*

*Responses of Community Structure, Productivity and Turnover Traits to Long-Term Grazing…*

*Results of partial least squares (PLS) regression correlating grassland productivity at Yunwu Mountain during 1992–2011 with 15-day running means of (a) daily mean temperature and (b) daily precipitation previously from September to august. Blue bars in the top row indicate that VIP values are greater than 0.8, the threshold for variable importance. In the middle row, red colour means model coefficients are negative and important, while green colour indicates important positive relationships between grassland productivity and climate variables. The black lines in the bottom panel stand for daily mean temperature and precipitation, while grey, green and red areas represent the standard deviation of daily climate* 

*The remaining mass dynamics of leaf litters of three Stipa species during 2 years' field decomposition process.*

**Species Remaining mass k-Value**

*Comparisons of litter decomposition traits after 1 and 2 years' decomposition between three Stipa species.*

*S. bungeana* 70.05 ± 3.91 b 64.47 ± 3.66 ab 0.360 0.236 *S. grandis* 73.97 ± 1.81 ab 61.53 ± 5.24 b 0.320 0.242 *S. przewalskyi* 79.18 ± 1.49 a 65.77 ± 1.80 a 0.237 0.225

**First year Second year First year Second year**

*DOI: http://dx.doi.org/10.5772/intechopen.85306*

*Responses of Community Structure, Productivity and Turnover Traits to Long-Term Grazing… DOI: http://dx.doi.org/10.5772/intechopen.85306*

### **Figure 6.**

*Plant Communities and Their Environment*

nity biomass of 3240.2 g m<sup>−</sup><sup>2</sup>

more complex impacts (**Figure 6b**).

(*P <* 0.05).

with aboveground community biomass of 520.5 g m<sup>−</sup><sup>2</sup>

(**Figure 4**).

**3.4 Responses of aboveground productivity to climate variation**

(**Figure 5b**) but was little influenced by AP variations (**Figure 5a**).

**3.5 Decomposition traits of leaf litters of three dominant** *Stipa* **species**

lation index (NAI) of leaf litters among three *Stipa* species (**Table 2**).

*Correlations between ANPP and annual precipitation (a) and mean annual temperature (b) during 1992–2011 at Yunwushan. AP means annual precipitation and MAT represents mean annual temperature.*

Aboveground biomasses of four plant groups increased with restoration time after grazing exclusion. Aboveground biomass of Gramineae and Compositae peaked at the 20th year, while that of Leguminosae peaked at the 25th year, and that of Weeds families peaked at the 15th year during restoration process after grazing exclusion. Considering the reduced biomasses of weed families, long-term grazing exclusion improved forage quality of grassland. Meanwhile, aboveground and belowground community biomasses were both increased by grazing exclusion. Since grassland mainly consisted of plants belonging to Gramineae and Compositae, peaks of the total above- and belowground community biomass both occurred at the 20th year,

Regression analysis showed that ANPP was significantly correlated with MAT

The VIP and standardised model coefficients of the PLS analysis showed that impacts of warming on grassland productivity varied with season periods (**Figure 6a**). Different with the clear-cut impacts of temperature on ANPP, precipitation showed

The remaining mass of leaf litters decreased with decomposition time and showed significant differences among three *Stipa* species (**Figure 7**). At the end of decomposition experiment, the remaining masses of leaf litters of *S. bungeana*, *S. grandis* and *S. przewalskyi* were 64.47%, 61.53% and 65.78%, respectively (**Table 1**). Different lowercase letters in the same column indicate significant differences

During 2 years' decomposition process, variations of nutrient concentration were affected by the nutrient type (**Figure 8**). In detail, concentrations of carbon and nitrogen showed species-specific fluctuations with decreasing tendency among three *Stipa* species. In contrast, phosphorus concentrations in leaf litters were averaged doubled. There were significant differences in C:N ratio and nutrient accumu-

and belowground commu-

**142**

**Figure 5.**

*Results of partial least squares (PLS) regression correlating grassland productivity at Yunwu Mountain during 1992–2011 with 15-day running means of (a) daily mean temperature and (b) daily precipitation previously from September to august. Blue bars in the top row indicate that VIP values are greater than 0.8, the threshold for variable importance. In the middle row, red colour means model coefficients are negative and important, while green colour indicates important positive relationships between grassland productivity and climate variables. The black lines in the bottom panel stand for daily mean temperature and precipitation, while grey, green and red areas represent the standard deviation of daily climate variables.*

### **Figure 7.**

*The remaining mass dynamics of leaf litters of three Stipa species during 2 years' field decomposition process.*


### **Table 1.**

*Comparisons of litter decomposition traits after 1 and 2 years' decomposition between three Stipa species.*

### **Figure 8.**

*Dynamic of carbon (a), nitrogen (b), phosphorus (c), concentrations and C:N ratio (d) of leaf litters of three Stipa species during 2 years' field decomposition process.*


### **Table 2.**

*Analysis of variance of decomposition time, species for nutrient concentration, C:N ratio and NAI.*

NAI, nutrient accumulation index; ns indicates no significant effects (*P >* 0.05). \*\* and \*\*\* indicate significant effects at *P <* 0.01 and *P <* 0.001 level, respectively.

Different with nutrient concentrations, nutrient accumulation indices in **Figure 9** indicated that C, N and P were all mineralised into soils during the decomposition process. There was no significant difference between species for carbon-releasing pattern (**Figure 9**).

**145**

**4. Discussion**

*field decomposition process.*

**Figure 9.**

*Responses of Community Structure, Productivity and Turnover Traits to Long-Term Grazing…*

Anthropogenic activities and climate changes have made important impacts on terrestrial ecosystem structure and functions in the last century [30]. Global surface temperatures during the twentieth century was increased by 0.56–0.92°C, while temperatures are predicted to have an increment of 2.0–4.5°C in the twentyfirst century [61]. Annual mean air temperature was increased by 1.17°C from 1982 to 2011 in this study, having similar temperature changing trends with study in Xilingol steppe of Inner Mongolia [61]. In detail, temperature rises differentiated with seasons, with temperature rises of 1.01°C and 1.68°C in growing season and nongrowing season, respectively. Thus, the nongrowing season experienced a higher temperature rise than the growing season. In contrast with mean annual temperature, mean annual precipitation showed a decreasing trend and larger intra- and interannual variations in our study, indicating the warmer and drier climate. Previous researches have shown that vegetation characteristics could be improved using grazing exclusion in the degraded sandy grasslands, alpine meadow and wetlands in China [5, 62]. However, many of these restoration studies were based on a relatively short-term scale and the research strategy focusing on the spatial series substitute for temporal series methods [5, 63]. In this study, community coverage, plant species richness, plant density and Shannon-Wiener index had similar variation tendencies during the three-decade restoration process. After 20 years' restoration, they reached peak values, but these four index values

*NAI dynamics for carbon (a), nitrogen (b), phosphorus (c) of leaf litters of three Stipa species during 2 years'* 

*DOI: http://dx.doi.org/10.5772/intechopen.85306*

*Responses of Community Structure, Productivity and Turnover Traits to Long-Term Grazing… DOI: http://dx.doi.org/10.5772/intechopen.85306*

**Figure 9.**

*Plant Communities and Their Environment*

**144**

**Figure 8.**

Time× Species

**Table 2.**

*Stipa species during 2 years' field decomposition process.*

**Variables df Concentration (g·kg<sup>−</sup><sup>1</sup>**

pattern (**Figure 9**).

NAI, nutrient accumulation index; ns indicates no significant effects (*P >* 0.05). \*\* and \*\*\* indicate significant effects at *P <* 0.01 and *P <* 0.001 level, respectively. Different with nutrient concentrations, nutrient accumulation indices in **Figure 9** indicated that C, N and P were all mineralised into soils during the decomposition process. There was no significant difference between species for carbon-releasing

*Analysis of variance of decomposition time, species for nutrient concentration, C:N ratio and NAI.*

*Dynamic of carbon (a), nitrogen (b), phosphorus (c), concentrations and C:N ratio (d) of leaf litters of three* 

Time 6 0.575 ns 4.701 \*\* 39.564 \*\*\* 3.877\*\* 49.738 \*\*\* 23.944 \*\*\* 53.070 \*\*\* Species 2 0.613 ns 18.860 \*\*\* 2.991 ns 9.074\*\* 0.560 ns 11.026 \*\*\* 50.008 \*\*\*

**) C/N NAI**

**C N P C N P**

12 1.163 ns 1.843 ns 1.224 ns 1.889 ns 1.663 ns 1.014 ns 1.185 ns

*NAI dynamics for carbon (a), nitrogen (b), phosphorus (c) of leaf litters of three Stipa species during 2 years' field decomposition process.*

## **4. Discussion**

Anthropogenic activities and climate changes have made important impacts on terrestrial ecosystem structure and functions in the last century [30]. Global surface temperatures during the twentieth century was increased by 0.56–0.92°C, while temperatures are predicted to have an increment of 2.0–4.5°C in the twentyfirst century [61]. Annual mean air temperature was increased by 1.17°C from 1982 to 2011 in this study, having similar temperature changing trends with study in Xilingol steppe of Inner Mongolia [61]. In detail, temperature rises differentiated with seasons, with temperature rises of 1.01°C and 1.68°C in growing season and nongrowing season, respectively. Thus, the nongrowing season experienced a higher temperature rise than the growing season. In contrast with mean annual temperature, mean annual precipitation showed a decreasing trend and larger intra- and interannual variations in our study, indicating the warmer and drier climate. Previous researches have shown that vegetation characteristics could be improved using grazing exclusion in the degraded sandy grasslands, alpine meadow and wetlands in China [5, 62]. However, many of these restoration studies were based on a relatively short-term scale and the research strategy focusing on the spatial series substitute for temporal series methods [5, 63]. In this study, community coverage, plant species richness, plant density and Shannon-Wiener index had similar variation tendencies during the three-decade restoration process. After 20 years' restoration, they reached peak values, but these four index values

decreased in the following years. These decreases mainly resulted from accumulation of litter, which reduced the access to light for plant seedlings [64, 65]. Overall, 30 years' restoration made plant species richness increase from 9.5 species m<sup>−</sup><sup>2</sup> to 28 species m<sup>−</sup><sup>2</sup> and make grassland coverage increase from 25 to 85%. In addition, plants were categorised into four groups: Gramineae, Leguminosae, Compositae and weeds. Considering the reduced biomasses of weeds, long-term grazing exclusion improved forage quality of grassland. Meanwhile, aboveground and belowground community biomasses were both increased by grazing exclusion.

The rapid recovery due to grazing exclusion played a more important role than climatic variations in regulating grassland ecosystem. Therefore, datasets of aboveground grassland biomass and climate variables during 1992–2011 were used to examine the impacts of climate variations on aboveground net primary productivity (ANPP). Regression analysis showed that ANPP was significantly correlated with MAT and was little influenced by AP variations, while precipitation is regarded as the most important determinant of grassland productivity in arid and semiarid regions [19, 21, 66]. Considering the neglected temporal variation of annual climate variables, more attentions should be paid to studies at higher temporal resolution attributing impacts of climate variation on grassland productivity to seasonal or even daily variation in climatic variables rather than to annual variation [26, 27, 28, 29, 31]. A low root-mean-square error (RMSE) of 8.13 g m<sup>−</sup><sup>2</sup> indicated a good fit of the data for the resulting PLS model. The VIP and standardised model coefficients of the PLS analysis showed that impacts of warming on grassland productivity varied with season periods. Since model coefficients in April and May were always positive and VIP values mostly exceeded 0.8, warming in this period had a positive impact on grassland productivity. The positive impacts of warming in spring on grassland productivity may result from increased water absorption, N mineralisation, accelerated snowmelt and advanced spring greening for plants, which may lengthen the growing season and increase photosynthesis and carbon acquisition for plants [13, 67–69].

Warming in summer (June–July) depressed productivity, forming a striking contrast with the impacts of spring warming. The results can be explained by physiological stress for plant growth generated by warming in summer coinciding with drought [70]. Moreover, warming in summer may reduce soil moisture by increasing evapotranspiration [71]. It is believed that climate variations make impacts on grassland productivity through changes of soil moisture [24, 72, 73]. Furthermore, continuous warming and drought in summer reduced productivity by limiting soil resource availability [74, 75]. And, temperature variation in August had no apparent impacts on grassland productivity.

The majority of published studies have focused on productivity responses to climate variability during the growing season. However, the importance of winter climate is getting more and more attentions [76–80]. Considering the majority of model coefficients during September–March, high temperature at that time was unfavourable for productivity of the following year. Temperature increases during September–October delay the senescence of grassland, which may increase soil nutrient and water depletion, inhibiting biomass production in the following year [36, 69, 81]. Our results were similar with warming experiments in two limestone grasslands in the UK, which showed that winter heating combined with drought reduced the biomass of both communities [11]. Besides, warmer winter can accelerate snowmelt, resulting in declines of snow cover accompanied with increases of frequency of freezing events, which exerted negative impacts on plant growth [76, 82]. Also, warming in winter may delay the fulfilment of chilling requirements of plants for resuming growth in the following spring or even delay onset of spring phenology [58, 59, 77–79, 83].

**147**

*Responses of Community Structure, Productivity and Turnover Traits to Long-Term Grazing…*

The daily precipitation values between the previous September and August were

. Different with the

also used as independent variables in the PLS analysis. The resulting model still

clear-cut impacts of temperature on ANPP, precipitation showed more complex impacts. Precipitation increases in June and July had positive impacts on productivity, while increasing precipitation during the senescence period (September– October) and the early growing season (April–May) was correlated with low productivity. In contrast to studies reporting the positive impacts of precipitation during April–May on grassland productivity [29, 30], results in the present study can be explained by the site hydrology, with frequent winter snow providing sufficient soil water for plant growth, making sporadic precipitation during April–May (with an average of 59.5 mm during 1992–2011) which has less important direct impacts on grassland productivity. Similarly, there was also no significant relationship between grassland ANPP and precipitation in August. Similar results have also been reported for grasslands in Kansas, USA [13]. During the dormancy period, positive impacts of precipitation were almost offset by negative ones; thus, precipitation seemed to have little impacts on grassland productivity during this period. Investigating the decomposition traits of dominant *Stipa* species' (*S. bungeana*,

*S. grandis* and *S. przewalskyi*) litters can reveal the ecosystem cyclic process under grazing exclusion and climatic changes. The remaining mass of leaf litters decreased with decomposition time and showed significant differences among three *Stipa* species. At the end of decomposition experiment, the remaining mass of leaf litters of *S. bungeana*, *S. grandis* and *S. przewalskyi* were 64.47%, 61.53% and 65.78%, respectively. Therefore, *S. grandis* decomposed fast, and *S. przewalskyi* had a slow decomposition rate. Additionally, leaf litters decomposed faster in growing season (6–12 month and 18–24 month) than in nongrowing season (0–6 month and 12–18 month). The decomposition rate (k) was calculated based on the regression of negative exponential decay function, with k-values of 0.360, 0.320 and 0.237 after 1 year's decomposition for *S. bungeana*, *S. grandis* and *S. przewalskyi*, respectively. Similarly, k-values after 2 years' decomposition of *S. bungeana*, *S. grandis* and *S. przewalskyi* were 0.236, 0.242 and 0.225, respectively. Since higher k-values indicate higher decomposition rates, we concluded that litter's decaying progress became difficult as decomposition time increases, mainly due to the depletion of soluble compounds and easily decayed parts at the beginning of decomposition

process, leaving hard parts such as lignin to decay slowly [39].

The variations of nutrient concentration were affected by nutrient type during 2 years' decomposition process during 2013–2014(**Figure 8** and **Table 2**). In detail, concentrations of carbon and nitrogen showed species-specific fluctuations with decreasing tendency among three *Stipa* species (**Figure 8a** and **b**). In contrast, phosphorus concentrations in leaf litters were averaged doubled (**Figure 8c**), indicating immobilisation of P in the leaf litters, possibly due to microbial immobilisation through the uptake of P from soil solution and translocation of P from fungal hyphae [84]. There were significant differences in C:N ratio of leaf litters among three *Stipa* species (**Figure 8d**). *S. przewalskyi* had higher C:N ratio than *S. bungeana*, which

proved to be a good fit for the data, with an RMSE of 6.53 g m<sup>−</sup><sup>2</sup>

Interestingly, some short intervals with positive coefficients during 1 November–29 March were detected during 1992–2011, indicating a complex physiological and ecological process in dormancy period of grassland. Taking a broader view at model coefficients and aiming at consistency with established phenological phases, we interpreted the entire period (November–March) as another relevant period during which temperature increases appeared to reduce grassland productivity. Therefore, we recommend that more scientific attention should be paid to impacts of winter warming on grassland productivity and the timing of spring

*DOI: http://dx.doi.org/10.5772/intechopen.85306*

phenology events.

### *Responses of Community Structure, Productivity and Turnover Traits to Long-Term Grazing… DOI: http://dx.doi.org/10.5772/intechopen.85306*

Interestingly, some short intervals with positive coefficients during 1 November–29 March were detected during 1992–2011, indicating a complex physiological and ecological process in dormancy period of grassland. Taking a broader view at model coefficients and aiming at consistency with established phenological phases, we interpreted the entire period (November–March) as another relevant period during which temperature increases appeared to reduce grassland productivity. Therefore, we recommend that more scientific attention should be paid to impacts of winter warming on grassland productivity and the timing of spring phenology events.

The daily precipitation values between the previous September and August were also used as independent variables in the PLS analysis. The resulting model still proved to be a good fit for the data, with an RMSE of 6.53 g m<sup>−</sup><sup>2</sup> . Different with the clear-cut impacts of temperature on ANPP, precipitation showed more complex impacts. Precipitation increases in June and July had positive impacts on productivity, while increasing precipitation during the senescence period (September– October) and the early growing season (April–May) was correlated with low productivity. In contrast to studies reporting the positive impacts of precipitation during April–May on grassland productivity [29, 30], results in the present study can be explained by the site hydrology, with frequent winter snow providing sufficient soil water for plant growth, making sporadic precipitation during April–May (with an average of 59.5 mm during 1992–2011) which has less important direct impacts on grassland productivity. Similarly, there was also no significant relationship between grassland ANPP and precipitation in August. Similar results have also been reported for grasslands in Kansas, USA [13]. During the dormancy period, positive impacts of precipitation were almost offset by negative ones; thus, precipitation seemed to have little impacts on grassland productivity during this period.

Investigating the decomposition traits of dominant *Stipa* species' (*S. bungeana*, *S. grandis* and *S. przewalskyi*) litters can reveal the ecosystem cyclic process under grazing exclusion and climatic changes. The remaining mass of leaf litters decreased with decomposition time and showed significant differences among three *Stipa* species. At the end of decomposition experiment, the remaining mass of leaf litters of *S. bungeana*, *S. grandis* and *S. przewalskyi* were 64.47%, 61.53% and 65.78%, respectively. Therefore, *S. grandis* decomposed fast, and *S. przewalskyi* had a slow decomposition rate. Additionally, leaf litters decomposed faster in growing season (6–12 month and 18–24 month) than in nongrowing season (0–6 month and 12–18 month). The decomposition rate (k) was calculated based on the regression of negative exponential decay function, with k-values of 0.360, 0.320 and 0.237 after 1 year's decomposition for *S. bungeana*, *S. grandis* and *S. przewalskyi*, respectively. Similarly, k-values after 2 years' decomposition of *S. bungeana*, *S. grandis* and *S. przewalskyi* were 0.236, 0.242 and 0.225, respectively. Since higher k-values indicate higher decomposition rates, we concluded that litter's decaying progress became difficult as decomposition time increases, mainly due to the depletion of soluble compounds and easily decayed parts at the beginning of decomposition process, leaving hard parts such as lignin to decay slowly [39].

The variations of nutrient concentration were affected by nutrient type during 2 years' decomposition process during 2013–2014(**Figure 8** and **Table 2**). In detail, concentrations of carbon and nitrogen showed species-specific fluctuations with decreasing tendency among three *Stipa* species (**Figure 8a** and **b**). In contrast, phosphorus concentrations in leaf litters were averaged doubled (**Figure 8c**), indicating immobilisation of P in the leaf litters, possibly due to microbial immobilisation through the uptake of P from soil solution and translocation of P from fungal hyphae [84]. There were significant differences in C:N ratio of leaf litters among three *Stipa* species (**Figure 8d**). *S. przewalskyi* had higher C:N ratio than *S. bungeana*, which

*Plant Communities and Their Environment*

and carbon acquisition for plants [13, 67–69].

impacts on grassland productivity.

of spring phenology [58, 59, 77–79, 83].

28 species m<sup>−</sup><sup>2</sup>

decreased in the following years. These decreases mainly resulted from accumulation of litter, which reduced the access to light for plant seedlings [64, 65]. Overall, 30 years' restoration made plant species richness increase from 9.5 species m<sup>−</sup><sup>2</sup>

plants were categorised into four groups: Gramineae, Leguminosae, Compositae and weeds. Considering the reduced biomasses of weeds, long-term grazing exclusion improved forage quality of grassland. Meanwhile, aboveground and belowground community biomasses were both increased by grazing exclusion.

The rapid recovery due to grazing exclusion played a more important role than climatic variations in regulating grassland ecosystem. Therefore, datasets of aboveground grassland biomass and climate variables during 1992–2011 were used to examine the impacts of climate variations on aboveground net primary productivity (ANPP). Regression analysis showed that ANPP was significantly correlated with MAT and was little influenced by AP variations, while precipitation is regarded as the most important determinant of grassland productivity in arid and semiarid regions [19, 21, 66]. Considering the neglected temporal variation of annual climate variables, more attentions should be paid to studies at higher temporal resolution attributing impacts of climate variation on grassland productivity to seasonal or even daily variation in climatic variables rather than to annual variation [26, 27, 28, 29, 31]. A low root-mean-square error (RMSE) of 8.13 g m<sup>−</sup><sup>2</sup> indicated a good fit of the data for the resulting PLS model. The VIP and standardised model coefficients of the PLS analysis showed that impacts of warming on grassland productivity varied with season periods. Since model coefficients in April and May were always positive and VIP values mostly exceeded 0.8, warming in this period had a positive impact on grassland productivity. The positive impacts of warming in spring on grassland productivity may result from increased water absorption, N mineralisation, accelerated snowmelt and advanced spring greening for plants, which may lengthen the growing season and increase photosynthesis

Warming in summer (June–July) depressed productivity, forming a striking contrast with the impacts of spring warming. The results can be explained by physiological stress for plant growth generated by warming in summer coinciding with drought [70]. Moreover, warming in summer may reduce soil moisture by increasing evapotranspiration [71]. It is believed that climate variations make impacts on grassland productivity through changes of soil moisture [24, 72, 73]. Furthermore, continuous warming and drought in summer reduced productivity by limiting soil resource availability [74, 75]. And, temperature variation in August had no apparent

The majority of published studies have focused on productivity responses to climate variability during the growing season. However, the importance of winter climate is getting more and more attentions [76–80]. Considering the majority of model coefficients during September–March, high temperature at that time was unfavourable for productivity of the following year. Temperature increases during September–October delay the senescence of grassland, which may increase soil nutrient and water depletion, inhibiting biomass production in the following year [36, 69, 81]. Our results were similar with warming experiments in two limestone grasslands in the UK, which showed that winter heating combined with drought reduced the biomass of both communities [11]. Besides, warmer winter can accelerate snowmelt, resulting in declines of snow cover accompanied with increases of frequency of freezing events, which exerted negative impacts on plant growth [76, 82]. Also, warming in winter may delay the fulfilment of chilling requirements of plants for resuming growth in the following spring or even delay onset

and make grassland coverage increase from 25 to 85%. In addition,

to

**146**

explained the differences of decomposition rates between them. As the dominant species in late succession stage of grassland, C:N ratio of *S. przewalskyi* litters did not show a lower value as predicted from other studies [51], possibly due to the divergences of climate and species between two regions. C:N ratio has been proven to be negatively correlated with decomposition rate. Besides, the lower k-value after 2 years' decomposition process could be explained by the increased C:N ratio of leaf litters. Compared with nutrient concentrations, nutrient accumulation indices indicated that C, N and P were all mineralised into soils during the decomposition process. There was no significant difference between species for carbon-releasing pattern. Still, NAI value for C of *S. przewalskyi* was higher than two other *Stipa* species after 2 years' decomposition (**Figure 9**). The lower NAI values for N and P of *S. bungeana* indicated that *S. bungeana* released more N and P to soil than the two other *Stipa* species. From this perspective, replacement of *Stipa* species after long-term grazing exclusion might inhibit nutrient cycling of grassland ecosystem, due to the lower nutrient mineralisation in leaf litters of two *Stipa* species at middle and late succession stage.
