Litter accumulation alters the abiotic environment and drives community successional changes in two fenced grasslands in Inner Mongolia

Abstract Fencing is an effective and practical method for restoring degraded grasslands in northern China. However, little is known about the role of excess litter accumulation due to long‐term fencing in regulating abiotic environment and driving changes in community structure and function. We conducted a three‐year field experiment in two fenced grasslands in Inner Mongolia, and monitored light quantity, soil temperature, and soil moisture continuously, and determined community height, community aboveground net primary productivity (ANPP), and the relative dominance of different plant functional groups. Litter accumulation reduced light quantity and soil temperature but increased soil moisture. The regulating effects of litter accumulation on soil temperature and soil moisture fluctuated temporally and gradually weakened over the growing season. Litter accumulation also altered community vertical structure and function by increasing community height and ANPP. The increase in soil moisture increased the relative dominance of rhizome grasses but suppressed bunch grasses, thereby shifting bunch grass grasslands to rhizome grass grasslands. Our findings provide a potential mechanism for community succession in the context of litter accumulation in fenced grasslands and indicate that the vegetation and ecosystem services of degraded grasslands are improved after appropriate fencing.


2002
) and seeding (Barr, Jonas, & Paschke, 2017). However, these restoration methods not only need huge manpower and resources but also have limited applicability. Of all restoration methods, fencing appears to be an effective and practical method for restoring degraded grasslands, especially for those subjected to overgrazing (Jing, Cheng, & Chen, 2013;Wu, Du, Liu, & Thirgood, 2009).
However, although long-term fencing can facilitate the restoration of vegetation and soil in degraded grasslands (Liu, Wu, Su, Gao, & Wu, 2017;Wang et al., 2018), it can also create new issues. For example, long-term fencing results in excess litter accumulation on the soil surface, particularly in arid and semiarid grasslands.
Those alterations, in turn, modify plant-soil interactions (Brearley, Press, & Scholes, 2003). Unfortunately, limited information is available for understanding the role of those changes in the context of litter accumulation.
Litter accumulation can redistribute light, heat, and water, all of which have complex impacts on abiotic environment (Facelli & Pickett, 1991a, 1991bJensen & Gutekunst, 2003). Litter acts as a mechanical barrier, intercepting light, and altering the spectral structure (Facelli & Pickett, 1991b;Jensen & Gutekunst, 2003). It also reduces soil temperature during the day by decreasing solar radiation absorption, but increases soil temperature at night through reducing heat loss (Facelli & Pickett, 1991a). Additionally, litter accumulation may delay the freezing of soil in winter and thawing in spring (Facelli & Pickett, 1991a). Decreased soil temperature also indirectly improves soil moisture (Deutsch et al., 2010a). Litter can directly increase soil moisture by reducing water evaporation (Deutsch, Bork, & Willms, 2010b). In addition, litter accumulation can increase snow capture and retention in winter and soil moisture in early spring (Naeth & Chanasyk, 1995;Wikeem, Newman, & Ryswyk, 1989), but the interception effect of litter reduces infiltration by rainfall (Naeth, Bailey, Chanasyk, & Pluth, 1991). Therefore, understanding the modified microenvironment is crucial to elucidate the role of litter accumulation in regulating plant communities in fenced grasslands. Previous studies in fenced grasslands failed to continuously measure abiotic factors (Deutsch et al., 2010a;Facelli & Pickett, 1991a;Wang et al., 2011), and little is known about the effects of continuous changes in abiotic factors at different stages of the growing season.
The grasslands in Inner Mongolia are typical of grasslands in northern China. Bunch grass and rhizome grass grasslands are the most common types and are widely distributed in this region (Kang et al., 2007). However, because overgrazing has caused severe grassland degradation over the past decades, (Wang, Deng, Song, Li, & Chen, 2017), long-term fencing has been widely implemented.
Community structure and function have been significantly altered in some region due to excess litter accumulation, particularly in these bunch grass and rhizome grass grasslands.
The purpose of this study was to elucidate the role of excess litter accumulation in regulating light quantity, soil temperature, and soil moisture, and driving community successional changes in two fenced grasslands in Inner Mongolia. To this end, we performed a three-year field experiment in two contrasting fenced grasslands.
We hypothesized that excess litter accumulation could directly affect light, heat, and water regimes, with subsequent effects on community height, ANPP, and the relative dominance of different functional groups in fenced grasslands.

| Study area
The study area was located in Xilin Gol League, Inner Mongolia.
Grasslands are the typical vegetation, most of which have been fenced for a long time. Stipa grandis (bunch grass) and Leymus chinensis (rhizome grass) grasslands were selected as experimental communities. The S. grandis grassland was located at the Inner Mongolia The study area belongs to a temperate continental monsoon climate (cold and dry in winter and hot and wet in summer). The mean annual temperature of the S. grandis grassland is 0.3°C, with mean monthly temperature ranging from −21.6°C in January to 19.0°C in July. The mean annual precipitation is 351.0 mm, and 80% of the precipitation usually occurs from May to August. The mean annual temperature of the L. chinensis grassland is 0.1°C. The temperature of the coldest and hottest month is −19.0°C in January and 21.4°C in July; the mean annual precipitation is 300.3 mm. The soil type of the two study sites is a chestnut soil (Chinese Soil Taxonomic Classification), and the clay content is higher in the L. chinensis grassland than in the S. grandis grassland. The growing season in late April and lasts to mid-September.
In the S. grandis grassland, the common species include L. chinensis, Agropyron cristatum, Cleistogenes squarrosa, Achnatherum sibiricum, Carex duriuscula, Allium condensatum, and Allium tenuissimum. In the L. chinensis grassland, the common species include C. squarrosa, Stipa krylovii, C. duriuscula, Lappula myosotis, and A. tenuissimum. The number of species is lower in the L. chinensis grassland than in the S. grandis grassland.
Based on climate data at the two study sites, air temperature was higher in 2017 than in 2015 and 2016, and the precipitation was higher in 2015 and 2016 than in 2017 (Table 1). We also observed that plants suffered from drought stress in the early growing season of 2017.

| Measurements of soil temperature and soil moisture
Soil temperature and soil moisture were measured in situ with

| Measurements of light quantity
Light quantity was measured with an array illuminometer (ZL2016 2 1344510.1) designed by us. This design can reduce the heterogeneity in light estimates due to litter accumulation. The array illuminometer was composed of five light quantity sensors arranged in a row at 10-cm intervals. An adjustable shelf allowed us to measure light quantity at different heights. First, the array illuminometer was placed at a random location within the control plot and the light TA B L E 1 Climate data over the growing season at the two study sites during 2015-2017 quantity at the soil surface was measured. Then, the array illuminometer was elevated at 5-cm intervals to measure the light quantity at different heights. Finally, the light quantity that was unshaded by litter was measured (i.e., full light quantity). Light quantity at a given height was recorded with five light quantity sensors (five replicates per height) when the reading of the array illuminometer was stable.
Light quantity was measured between 14:00 and 15:00 with cloudless weather and at 6-day intervals from mid-April to mid-May during 2015-2017.

| Plant community sampling
Plant community characteristics were sampled using three quadrats (

| Data analyses
Because the litter in the litter removal treatment was removed dur- where Li C is the light quantity at different heights in the control. Li E is the light quantity at the same height in the litter removal treatment, which was full light.
The effects of litter accumulation on soil temperature and soil moisture were indicated by the difference in soil temperature and soil moisture between the litter removal treatment and control (Yan et al., 2018). The larger the difference, the greater the effects of litter accumulation. We used daily mean soil temperature and soil moisture at the same depth to calculate the difference (1 May to 30 September of 2015-2017). The two equations were as follows: where ST C is the daily mean of soil temperature in the control, and ST E is the daily mean of soil temperature in the litter removal treatment at the same depth. Similarly, SM C is the daily mean of soil moisture in the control and SM E is the daily mean of soil moisture in the litter removal treatment at the same depth. where B i is the dry biomass of a plant functional group in a given quadrat, and B is the total dry biomass of the same quadrat.
One-way analysis of variance with a post hoc Tukey test was used to test for differences in litter accumulation and percent light interception at each height during 2015-2017. A general linear mixed effects model was used to test the effects of litter accumulation on community height and ANPP with treatment as a fixed effect and block as a random effect. The linear recursive analysis was selected to quantify relationships between the relative dominance of two plant functional groups and both soil temperature and soil moisture.
All statistical analyses were performed using SPSS 21.0 (IBM).

| Litter accumulation characteristics
Litter biomass increased rapidly and significantly in the control plots of the L. chinensis grassland (Figure 1; p < .05) during 2015-2017, but did not change in the S. grandis grassland (Figure 1; p > .05).

| Light quantity characteristics
Litter accumulation reduced the light quantity in the control plots of both grasslands (

| Soil temperature characteristics
Litter accumulation decreased soil temperature during the growing season at both 2.5 cm and 12.5 cm depths in the control plots of both grasslands (Figure 2). The difference in soil temperature between the litter removal treatment and control was greater at 2.5 cm than at 12.5 cm (Figure 2). The difference in soil temperature gradually decreased over time ( Figure 2, the original data were supplemented in Figure S1). Specifically, the soil temperature difference was greater earlier in the growing season than later (Figure 2). This difference indicated that the regulating effect of litter accumulation on soil temperature fluctuated, and importantly, gradually weakened over the growing season. In addition, compared with 2015, soil temperature differences in 2016 and 2017 were higher in the L. chinensis grassland (Figure 2).

| Soil moisture characteristics
Unlike soil temperature, litter accumulation increased soil moisture at both 2.5 cm and 12.5 cm depths in the control plots of both grasslands during the growing season ( Figure 3, the original data were supplemented in Figure S2). The variation in soil moisture at 2.5 cm was greater than at 12.5 cm ( Figure 3). The soil moisture difference gradually decreased over the growing season, particularly at 2.5 cm ( Figure 3). Litter accumulation altered the relative dominance of the two plant functional groups in both grasslands (Figure 4c,f). In the control plots of the S. grandis grassland, litter accumulation significantly increased the relative dominance of rhizome grasses but decreased that of bunch grasses (Figure 4c). However, litter accumulation had no effect on the relative dominance of rhizome grasses but slightly decreased that of bunch grasses in the control plots of the L. chinensis grassland (Figure 4f).

| Plant community characteristics
Across both grasslands, the relative dominance of rhizome grasses and bunch grasses had no significant relationship with soil temperature (Figure 5a; p > .05). However, greater soil moisture led to an increase in the relative dominance of rhizome grasses and an decrease in bunch grasses (Figure 5b; p < .0001). A total of 68% of the variation in rhizome grasses and 70% of the variation in bunch grasses were explained by soil moisture (Figure 5a,b). Thus, the increase in soil moisture due to litter accumulation could be a driving force to shift the relative dominance of two plant functional groups in fenced grasslands.

| Effects of litter accumulation on soil temperature and soil moisture
In this study, litter accumulation decreased soil temperature but increased soil moisture, which is consistent with previous studies (Deutsch et al., 2010a(Deutsch et al., , 2010bFacelli & Pickett, 1991a). Early in the growing season, the shading effect of litter accumulation decreased soil temperature by preventing the absorption of solar radiation but increased soil moisture by inhibiting evaporation (Facelli & Pickett, 1991a). Solar radiation increased by the middle and late growing season, soil temperature increased, and the soil temperature difference between the control and litter removal treatment gradually diminished, especially later. The combined growth of plants and the rise in air temperature increased the loss of soil moisture due to evaporation and transpiration (Lauenroth & Bradford, 2006). However, because of concentrated rainfall in this region, water was continuously input into the soil in the middle and late growing season. Further, the accumulated litter also trapped and retained more snow in winter, which could increase soil moisture in spring (Naeth & Chanasyk, 1995). The balance of these factors resulted in the decrease in soil moisture in the litter removal treatment in the late growing season.
The effects of litter accumulation on soil temperature and soil moisture were strongest in the early growing season and had a negative effect on plant growth. Lower soil temperature can delay seed germination, decrease the growth rate of plants (Deutsch et al., 2010b), and even reduce community ANPP in the control plots of both grasslands (Figure 4b,e). But because of the increase in plant growth over the growing season, the effect of litter accumulation on these plants was slowly reduced. Later in the year, the soil surface covered by litter maintained a warm and stable environment, extending the growing season (Facelli & Pickett, 1991a;Watt, 1970).
Similarly, relatively higher soil moisture allowed plants to resist drought stress in the middle and late growing season, thereby increasing community ANPP (Figure 4b (Figures 2 and 3). This result indicates that some threshold of litter accumulation may determine the degree of its regulating effects (Deutsch et al., 2010b;Loydi, Eckstein, Otte, & Donath, 2013).

| Effects of litter accumulation on light quantity
Light quantity was reduced where the litter was not removed (Table 2), which agrees with previous studies (Facelli & Pickett, 1991b;Jensen & Gutekunst, 2003;Weltzin et al., 2005). In our study, percent light interception decreased rapidly in the control plots of the S. grandis grassland but decreased slowly in the control plots of F I G U R E 5 Relationships between the relative dominance of two plant functional groups and soil temperature and soil moisture. Soil temperature and soil moisture were monthly means between the depths of 2.5 cm and 12.5 cm during the growing season; dotted lines are the 95% confidence intervals of the fitting lines the L. chinensis grassland (Table 2). These contrasting changes might be due to differences in the litter in the two grasslands. Most of the litter lay flattened on the soil surface in the S. grandis grassland but remained standing for a long time in the L. chinensis grassland. In the presence of wind and snow, litter was also more likely to concentrate on the soil surface in the S. grandis grassland than the L. chinensis grassland. Further, tall plants are often better competitors for light than dwarf plants, especially when litter accumulation could intensify this competition (Letts et al., 2015). In addition, reduced light quantity delayed the increase in soil temperature, especially in the early growing season.

| Effects of litter accumulation on plant community
In our study, we found that the increase in soil moisture led to rhizome grasses expanding rapidly and bunch grasses declining, particularly in the S. grandis grassland (Figure 4c,f). These changes might alter interspecific competition between rhizome and bunch grasses, which was due to the fact that different plant functional groups respond differentially to water availability ( Figure 5). Rhizome grasses are often moisture-tolerant species, and bunch grasses are usually drought-tolerant species (Chen, Bai, Zhang, & Han, 2005). Thus, the moister microenvironment where litter accumulated benefited the rhizome grasses more than the bunch grasses. With the expansion of rhizome grasses and decline of bunch grasses, grassland resources could be improved because livestock in this region prefer to consume rhizome grasses.
In addition, litter accumulation can negatively affect the sexual reproduction of plants (Deutsch et al., 2010b), potentially impacting the population growth of species with sexual reproduction, such as S. grandis and C. squarrosa, whereas asexually reproducing species such as L. chinensis and C. duriuscula might be less affected.
Moreover, the lack of external stimuli, such as grazing and mowing, might also inhibit the tillering of bunch grasses, further limiting their growth and reproduction.
Our findings suggest that litter accumulation potentially drove community successional changes in bunch grass grasslands and that long-term fencing facilitates this shift. Succession usually occurs over long-time scales in arid and semiarid grasslands, but litter accumulation might act as a medium to indirectly alter interspecific competition and accelerate this process.
Litter accumulation significantly increased community height Litter accumulation could promote community ANPP ( Figure 4b,e), which agrees with some previous studies (Weltzin et al., 2005;Willms, McGinn, & Dormaar, 1993). This increase in ANPP was likely due to the increase in soil moisture (Deutsch et al., 2010b;Wang et al., 2011). Interestingly, the community ANPP of the S. grandis grassland decreased in the early growing season. Compared with the L. chinensis grassland, the litter was denser ( Figure 1) and more concentrated on the soil surface in the S. grandis grassland. These differences could delay seed germination and decrease plant growth in the S. grandis grassland in the early growing season. The increase in community ANPP was higher in the L. chinensis grassland than the S. grandis grassland (Figure 4b,e). This result was consistent with rapid litter accumulation in the L. chinensis grassland during 2015-2017 ( Figure 1) and indicates that rhizome grass grasslands may be more suitable for litter accumulation than bunch grass grasslands.
In a relatively dry year (2017), we observed that litter accumulation promoted community ANPP, supporting previous findings (Deutsch et al., 2010b).
Note that the litter removal was achieved via mowing at the end of the growing season, and the litter in the control plots was completely retained in this study. Ideally, the litter in these two treatments should be removed at the same time and the corresponding litter was re-applied to the control plots of two grasslands. However, this manipulation could strongly destroy the natural structure of litter layer in the control plots, thereby altering the real light, heat, and water regimes. In this study, we used mowing to remove litter for the following reasons. First, our blocks (20 m × 50 m) were much larger than those (2 m × 6 m) used in previous studies (Wang et al., 2011). It was hardly possible to remove litter from such large blocks without external interference. In order to uniformly remove litter, mowing might be the most feasible method. Secondly, this method has been widely applied in similar studies in Inner Mongolia (DJ Hou, personal observation).
To minimize the effects of mowing, litter removal is commonly conducted at the end of growing seasons when all plants are dormant (Deutsch et al., 2010b;Wang et al., 2011). However, it should be noted that litter removal by mowing has some limitations. First, mowing can affect plants due to the presence of mechanical disturbances, such as increasing soil compaction and trampling plants. Secondly, the mechanical disturbance due to mowing should be applied in the control plots, but this manipulation could alter the natural structure of litter layer, leading to the absence of real control. In the future, similar studies about litter removal by mowing should consider the effects of mechanical disturbances as much as possible, and the effects of litter removal and mechanical disturbances should be dissected.
In summary, our three-year field experiment provides insights into the role of litter accumulation in regulating abiotic factors and plant communities in fenced grasslands. Our findings will advance our understanding of community succession in the context of litter accumulation. Litter accumulation regulated light quantity, soil temperature, and soil moisture, increased community height and ANPP, and shifted the relative dominance of different plant functional groups. The vegetation and ecosystem services of degraded grasslands were improved after appropriate fencing. In addition, forage palatability was increased because of the increase in rhizome grasses.

ACK N OWLED G M ENTS
We are grateful to Prof. Yongfei Bai and Prof. Taogetao Baoyin for providing two experimental platforms and also thank the staff of two field stations for their logistical support during the experiment. We would like to thank Prof. Simon Queenborough at Yale University for his assistance with the English language editing of the manuscript.
This study was supported by The National Basic Research Program of China (2014CB138802).

CO N FLI C T O F I NTE R E S T
No conflict of interest.

AUTH O R CO NTR I B UTI O N
Ke Guo, Changcheng Liu, and Dongjie Hou conceived and designed this experiment. Dongjie Hou and Xianguo Qiao performed the field experiment and processed the data. Dongjie Hou and Weiming He analyzed the data and wrote the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data supporting the results are archived on Dryad (https ://doi. org/10.5061/dryad.6r99k25).