Asymmetric effects of litter accumulation on soil temperature and dominant plant species in fenced grasslands

Excess litter accumulates on the soil surface of fenced grasslands and alters the abiotic environment and plant population dynamics. However, little is known about the effect of litter accumulation on the interaction between environmental factors and plant population characteristics in fenced grasslands, especially over different time scales. We applied a three-year litter removal experiment to two kinds of fenced grasslands in Inner Mongolia, China. We measured soil temperature in situ and plant phenology and population characteristics of three dominant species (Stipa grandis, S. krylovii, and Leymus chinensis). During the growing season, litter accumulation (i.e., the control) significantly decreased soil temperature, with a larger effect in the daytime than at night. The diurnal negative effect gradually weakened across the growing season, whereas the negative effect in the nighttime shifted to a positive effect on soil temperature in the late growing season. The decreased soil temperature delayed plant phenology, with longer delays in S. grandis and S. krylovii than L. chinensis. Litter accumulation also significantly increased the height, cover, root biomass, and relative dominance of L. chinensis but decreased cover, density, root biomass, and relative dominance of both Stipa, driving replacement of S. grandis or S. krylovii by L. chinensis in two grasslands. Our findings emphasize the critical function of litter in grassland management and provide a new insight to elucidating the mechanism of how litter accumulation regulates the abiotic environment, community composition and structure, and successional change in fenced grasslands.


INTRODUCTION
Wilted, abscised, and decomposing plant material (i.e., plant litter) is a critical component of many arid and semiarid grasslands and often influences community composition, structure, and productivity (Wang et al. 2011, Letts et al. 2015. Plant litter acts to mediate the interactions between the abiotic environment and the plant community. Litter accumulation decreases soil temperatures (Facelli andPickett 1991, Weltzin et al. 2005) that regulate seed germination (Eckstein and Donath 2005), plant growth (Facelli andPickett 1991, Song et al. 2018), and the mineralization rates of soil nutrients (Weltzin et al. 2005). Litter accumulation is a mechanical barrier to heat exchange between the soil and air through two contrasting effects (Facelli andPickett 1991, Weltzin et al. 2005). For example, in the daytime, litter shades the soil, reducing the absorption of solar radiation and keeping soil temperature lower than bare soil (Facelli andPickett 1991, Song et al. 2018). But in the nighttime, litter prevents heat loss from soil, keeping soil temperatures higher (Facelli andPickett 1991, Weltzin et al. 2005). However, these two effects of litter accumulation on soil temperatures are not well understood. In addition, most previous studies of the effects of litter accumulation on soil temperatures usually integrate over the entire day (e.g., daily mean, daily maximum, and daily minimum; Song et al. 2018), neglecting the differences between daytime and nighttime soil temperatures. Furthermore, as air temperature and community characteristics change during the course of the growing season, the efficiency of these two effects could also change. Xia et al. (2010) reported that community composition, structure, and function can be altered by slight changes in soil temperature. Unfortunately, little is known about how litter accumulation regulates soil temperatures across different time scales, which is important for understanding the effects of litter accumulation on plant populations and communities.
Litter accumulation affects the plant population characteristics and interspecific relationships (Facelli and Pickett 1991) that are driving forces of community composition and structure (Ruprecht et al. 2008). For example, litter accumulation alters plant phenology via effects on soil temperature and moisture (Zhu et al. 2016). Additionally, species with different sensitivities to litter accumulation have diverse responses, in terms of aboveground and belowground morphology and population characteristics (Facelli and Pickett 1991). For instance, litter accumulation increases the cover and aboveground biomass of some species but decreases that of others (Weltzin et al. 2005, Kelemen et al. 2013. Litter accumulation also acts as a mechanical barrier, preventing seeds from reaching the soil surface (Jensen andGutekunst 2003, Loydi et al. 2013), as well as decreasing the light quantity and quality reaching the soil (Hou et al. 2019), inhibiting germination of some species (Jensen and Gutekunst 2003). Some secondary metabolites in leaf litter also inhibit population renewal (Hovstad andOhlson 2008, Ruprecht et al. 2008). These changes in plant phenology, form, and population dynamics further influence community composition, structure, and function (Xiong and Nilsson 1999, Amatangelo et al. 2008, Letts et al. 2015, and even productivity (Willms et al. 1993, Deutsch et al. 2010, Wang et al. 2011. Ultimately, the altered interspecific relationships under litter accumulation could drive alternate community successional trajectories (Jing et al. 2013, Hou et al. 2019. Grassland ecosystems are one of the important terrestrial ecosystems in China, covering 40% of its land area (Kang et al. 2007). Most grasslands in China have suffered from severe degradation due to long-term overgrazing over the past decades (Kang et al. 2007, Wiesmeier et al. 2009). To restore these degraded grasslands, fencing that limits grazing is regarded as an effective and simple method, which has been widely applied in northern China and especially in Inner Mongolia (Han et al. 2009. Plant height, cover, aboveground biomass, and soil nutrients of degraded grasslands have been improved after long-term fencing in Inner Mongolia (Liu et al. 2017, Sagar et al. 2019. However, compared with other grassland managements (e.g., grazing and mowing), long-term fencing also results in remarkable litter accumulation on the surface of grasslands. The effects of this litter accumulation on the abiotic environment, plant population characteristics, and community structure and function are still unclear.
Bunchgrass grasslands, dominated by Stipa grandis or S. krylovii, are typical of Inner Mongolia. The species S. grandis, S. krylovii, and Leymus chinensis are widely distributed in these two kinds of grasslands, playing important roles in regulating interspecific relationships and community structure and productivity. However, the three species have different characteristics in terms of morphology, plant functional group, and ecological adaptations . For example, S. grandis and S. krylovii are bunch grasses with deeper roots, adapted to a relatively dry environment . In contrast, L. chinensis is a typical rhizomatous grass with long, strong rhizomes and clonal propagation (Bai et al. 2009) that grows in a relatively moist environment . However, there are few reports on the adaptability of these three species to litter accumulation in fenced grasslands.
The aim of this study was to examine the effects of litter accumulation on the soil temperature, plant phenology, and population characteristics of the dominant species in fenced grasslands. During 2015-2017, soil temperature v www.esajournals.org was continuously monitored and plant phenology and population characteristics of S. grandis, S. krylovii, and L. chinensis were investigated in two fenced grasslands in Inner Mongolia, China. We addressed the following questions: (1) How does litter accumulation regulate soil temperature at different time scales? (2) How do the three dominant species, with different ecological traits, respond to litter accumulation?
In the S. grandis grassland, the mean annual temperature is 0.3°C, the monthly mean temperature ranges from −21.6°C in January to 19.0°C in July, and the mean annual precipitation is 351 mm. In the S. krylovii grassland, the mean annual temperature is 0.1°C, the monthly mean temperature ranges from −19.0°C in January to 21.4°C in July, and the mean annual precipitation is 300 mm. In both grasslands, 80% of the precipitation occurs between May and August, and the soil type is a chestnut soil (Chinese Soil Taxonomic Classification). The growing season starts in late April and lasts to mid-September in both sites.

Experimental design
A three-year field experiment was conducted during 2015-2017 with a random complete block design. Within each fenced grassland, three blocks (50 × 22 m) were established and the distance between two blocks was 200 m. In each block, the half of block (50 × 10 m) was randomly assigned as the litter removal treatment and the other half (50 × 10 m) was the control treatment, separated by a 2-m buffer. All the litter (mainly including dead, aboveground leaves, and stems of plants) in the litter removal treatment was cut near the soil surface and removed at the end of the growing season in late October 2014-2016 (Hou et al. 2019). The all litter in the control treatment was retained throughout the experiment.

Soil temperature
We continuously measured soil temperature (°C) using ECH 2 O 5TE sensors (Decagon Devices, Pullman, Washington, USA). In each treatment half-block, ECH 2 O 5TE sensors were installed in the soil at 2.5 cm and 12.5 cm depths. Soil temperature was automatically recorded at 10-min intervals, from 1 May 2015 to 30 September 2017.

Plant population characteristics
L. chinensis is a typical rhizomatous grass, and S. grandis and S. krylovii are bunch grasses. The aboveground biomass of S. grandis and L. chinensis was 63.4-73.5% of the total aboveground biomass in the S. grandis grassland and that of S. krylovii and L. chinensis was 77.4-95.2% in the S. krylovii grassland. In the S. grandis grassland, we monitored plant phenology, population characteristics, and root biomass of S. grandis and L. chinensis. In the S. krylovii grassland, we monitored S. krylovii and L. chinensis.
Leaf-out day, first flowering day, and first withering day indicated the returning green, reproductive, and withering stages, respectively, which were the main stages of plant growth. A total of fifteen individual plants of each species (five individuals per block × three blocks) were randomly selected in each treatment of both grasslands and labeled before each growing season. Plant phenology was monitored every 3-5 d from 15 April to 15 October over the growing season of 2015-2017 (Ji et al. 2017). During phenological changes, the frequency of observation increased to every two days. The leaf-out day was the day that the first leaf was observed for each marked individual (Ji et al. 2017). The first v www.esajournals.org flowering day and the first withering day were also recorded.
Nine 1 × 1 m quadrats (three quadrats per block × three blocks) were randomly set in each treatment of both grasslands on 1 August 2015-2017. For each quadrat, the cover (%), plant height (cm), and density (individual per m 2 ) of each species were recorded. In this study, the density of L. chinensis was counted by individual and the density of S. grandis and S. krylovii by clumps. To determine aboveground biomass (g), the aboveground parts of each individual were clipped near the soil surface and dried at 65°C for 48 h to a constant weight.
According to Chen et al. (2001)'s study, the root biomass of S. grandis concentrated in the soil depth of 0-30 cm and that of S. krylovii and L. chinensis in 0-25 cm and 0-20 cm, respectively (Chen et al. 2001). In each treatment of both grasslands, nine individual plants of each species (three individuals per block × three blocks) were randomly selected and dug up on 1 August 2015 and 2016 (the weather was too dry in 2017), removing all the soil and roots in a radius of 30 cm and depth of 50 cm. According to root morphological characteristics of the three species (Appendix S1: Fig. S1), roots were washed from topsoil to deep soil carefully to remove soil, separated, and placed into envelopes and then dried at 65°C for 48 h to a constant weight.

Data analyses
Soil temperature in the daytime and nighttime was indicated by the mean soil temperature between 6:00 am-6:00 pm and 6:00 pm-6:00 am (local time), respectively (Xia et al. 2010). The effects of litter accumulation on soil temperature were indicated by the soil temperature difference between the control treatment and the litter removal treatment, calculated by the following equation (Hou et al. 2019): where ST C is the mean soil temperature in the control treatment and ST E is the mean soil temperature in the litter removal treatment. To examine differences in the mean soil temperature in the daytime and nighttime among different months and during the growing season, we used repeated-measures ANOVA.
To test for the effect of litter accumulation on plant phenology, the phenological date was first transformed into the Julian day. Then, we compared the mean leaf-out day, first flowering day, and first withering day of each species using a general linear model with treatment and block as factors. In this study, block had no significant effect on the plant phenology (P > 0.05).
Population-level characteristics of the three species were indicated by the means of plant height, cover, and density in the nine quadrats of each treatment. Root characteristics were indicated by the mean of root biomass of the nine individuals in each treatment of both grasslands. The relative dominance of the three species was indicated by the mean relative aboveground biomass in the nine quadrats of each treatment. It was calculated by the following equation: where B i is aboveground biomass of a species and B is the total aboveground biomass in this quadrat.
To test for the effect of litter accumulation on cover, plant height, density, root biomass, and relative dominance of the three species, we used a general linear model with treatment and block as factors. In this study, block had no significant effect on the population characteristics (P > 0.05). All statistical analyses were conducted in SPSS 21.0 (IBM, Chicago, Illinois, USA).

Asymmetric effects of litter accumulation on soil temperature
Compared with the litter removal treatment, litter accumulation significantly decreased soil temperature in the control plots of both grasslands, with a larger effect in the daytime than in the nighttime ( Fig. 1; Appendix S1: Fig. S2; P < 0.05). This negative effect was greater at 2.5 cm depth than 12.5 cm ( Fig. 1; Appendix S1: Fig. S2; P < 0.0001). For example, in the S. grandis grassland, litter accumulation decreased soil temperature at 2.5 cm depth in the daytime by 2.3°-6.3°C and in the nighttime by 0.0°-0.8°C. The decrease was much lower at 12.5 cm depth (1.6°-2.2°C in the daytime and 1.4°-3.1°C in the nighttime; Appendix S1: Fig. S2). In the S. v www.esajournals.org krylovii grassland, litter accumulation decreased soil temperature at 2.5 cm depth in the daytime by 1.0°-6.3°C and in the nighttime by −0.8°-0.6°C. Again, the decrease was less at 12.5 cm depth (0.3°-2.4°C in the daytime and 0.6°-2.0°C in the nighttime; Appendix S1: Fig. S2). The negative effect of litter accumulation on soil temperature gradually weakened during the growing season ( Fig. 1; P < 0.05), and in the late growing season even shifted to a positive effect during the nighttime ( Fig. 1; P < 0.05).

Asymmetric effects of litter accumulation on plant phenology
Litter accumulation in the control plots delayed the leaf-out day, first flowering day, and first withering day of the three species of both grasslands compared with the removal treatment ( Fig. 2; P < 0.05). The delay in S. grandis or S. krylovii was significantly longer than L. chinensis in the control plots of both grasslands, respectively ( Fig. 2; t-test: P < 0.05). For example, the leaf-out days of S. grandis and S. krylovii were postponed by 16.9-24.3 d and 9.3-11.7 d (Fig. 2a  and c), but that of L. chinensis was only postponed by 5.1-6.7 days in the S. grandis grassland and 2.1-3.2 d in the S. krylovii grassland (Fig. 2b  and d).
Furthermore, the earlier phenological stages were more delayed than the later ones (Fig. 2). For instance, in the control of the S. grandis grassland, the delay of leaf-out (in the early growing Fig. 1. Monthly dynamics of soil temperature difference in the daytime and nighttime between the two treatments. Data were mean + 1 SE; different lowercase and uppercase letters, respectively, indicated that there were significant differences among the monthly soil temperature difference in the daytime and nighttime at 0.05 level. v www.esajournals.org season) in S. grandis was 16.9-24.3 d and in L. chinensis was 5.1-6.7 d (Fig. 2a and c), whereas the first withering day in S. grandis (delayed 9.3-9.7 d) was similar to that in L. chinensis (5.3-9.9 d) in the late growing season (Fig. 2i-l).

Asymmetric effects of litter accumulation on population characteristics
Litter accumulation had no significant effect on the plant height of S. grandis or S. krylovii (Fig. 3a and c), but significantly increased the height of L. chinensis by 66.7-77.5% in the S. grandis grassland and 30.2-54.2% in the S. krylovii grassland ( Fig. 3b and d; P < 0.05). Litter accumulation reduced the cover of S. grandis or S. krylovii in the control plots of both grasslands ( Fig. 3e and g; P < 0.01). In contrast, the cover of L. chinensis increased in the control plots of both grasslands ( Fig. 3f and h; P < 0.05). Litter accumulation decreased the density of S. grandis and S. krylovii (Fig. 3i and k; P < 0.05), but had no significant effect on L. chinensis ( Fig. 3j and l).
In 2015, litter accumulation had no significant effects on the root biomass of S. grandis or S. krylovii ( Fig. 4a and c), but in 2016 the root biomass was significantly greater in the litter removal treatments of both grasslands ( Fig. 4a and c; P < 0.05). The root biomass of L. chinensis reduced in the litter removal treatments of both grasslands in 2015 and 2016 (P = 0.059 and 0.051; Fig. 4b and d).
Litter accumulation altered the relative dominance of the three species in both grasslands (Fig. 5). The relative dominance of S. grandis or S. krylovii was significantly lower in the control plots of both grasslands (Fig. 5a and c; P < 0.01), but the relative dominance of L. chinensis was greater ( Fig. 5b and d; P < 0.01).

DISCUSSION
Litter accumulation asymmetrically altered soil temperature, plant phenology, and population characteristics of three dominant species in Fig. 2. Plant phenology of the three species between the two treatments. (a-d) indicated leaf-out day of the three species; (e-h) indicated first flowering day of the three species; (i-l) indicated first withering day of the three species; time series were used the Julian day; data were mean + 1 SE andN = 15; and *, **, and *** indicated that there were significant differences in leaf-out day, first flowering day, and first withering day of the three species between the two treatments at 0.05, 0.01, and 0.0001 levels, respectively. fenced grasslands in Inner Mongolia, China. Litter accumulation decreased soil temperatures, more so in the daytime than the nighttime, except late in the growing season when litter accumulation increased nighttime soil temperatures. The decreased soil temperature delayed plant phenology, longer in S. grandis or S. krylovii than L. chinensis. These abiotic and phenological changes asymmetrically altered population characteristics of the three species, thereby changing population dynamics.

Effects of litter accumulation on soil temperature
Litter accumulation decreased soil temperature in the control plots of both experimental grasslands, consistent with many previous studies (Facelli andPickett 1991, Weltzin et al. 2005). More importantly, we found that this negative effect was greater in the daytime than in the nighttime, particularly at shallow (2.5 cm) depths ( Fig. 1; Appendix S1: Fig. S2). This result indicated that litter accumulation has complicated and asymmetric effects on soil temperature depending on the time of day, with the transfer of thermal energy from air to the soil during the day reversing during the night (Xia et al. 2010). The shading effect of litter reduces the daytime heat absorption of soil, but the physical barrier of litter creates a warming effect to inhibit heat loss from soil during the night (Facelli and Pickett 1991). Overall, the shading effect of litter accumulation was greater than the warming effect. In addition, it also shows that litter accumulation decreased soil temperature, owing to the decrease in soil temperature in the daytime. Compared with the warming effect, the shading effect of litter accumulation had a dominant effect on soil temperature in fenced grasslands. Thus, the strong temporal variation in these shading and warming effects of litter must be accounted for in future studies.
Over longer time scales, such as a growing season, the negative effect of litter accumulation on daytime soil temperatures gradually weakened, but the negative effect in the nighttime shifted to a positive effect by in the late growing season (Fig. 1). This result indicated that litter accumulation has different effects on soil temperature in Fig. 3. Population characteristics of the three species between the two treatments. (a-d) indicated the plant height of the three species; (e-h) was the cover of the three species; (i-l) was the density of the three species; data were mean + 1 SE andN = 9; and *, **, and *** indicated that there were significant differences in plant height, cover, and density between the two treatments at 0.05, 0.01, and 0.0001 levels, respectively. v www.esajournals.org Fig. 4. Root biomass of the three species between the two treatments. Data were mean + 1 SE andN = 9; and * and ** indicated that there were significant differences in root biomass between the two treatments at 0.05 and 0.01 levels, respectively. Fig. 5. Relative dominance of the three species between the two treatments. Data were mean + 1 SE andN = 9; and ** and *** indicated that there were significant differences in relative dominance of the three species between the two treatments at 0.01 and 0.0001 levels.
v www.esajournals.org 8 November 2020 v Volume 11(11) v Article e03289 the day versus the night at different stages of the growing season. In the early growing season, litter accumulation severely postponed increase in soil temperature through reducing the heat absorption of soil, whereas had a weak warming effect on soil temperature in the nighttime because of lower soil temperature. Therefore, litter accumulation had a negative effect on soil temperature in the early growing season, supporting some previous studies (Facelli and Pickett 1991). Over the growing season, soil temperature gradually increased in both grasslands with the increase in solar radiation over the year. In contrast, the shading effect of litter gradually reduced across time due to the increase in plant growth. Increasing insolation and plant growth might therefore decrease the effectiveness of litter shading on soil temperature over the growing season. However, the accumulation of litter later in the growing season decreases the heat exchange between soil and air in the nighttime, resulting in a positive effect on soil temperature, consistent with previous studies (Facelli and Pickett 1991). Soil temperature is a critical ecological factor that affects plant growth and regulates biological processes in terrestrial ecosystems (Field et al. 1992). Slight changes in soil temperature can alter soil moisture, soil respiration rate, and nutrient cycling (Xia et al. 2010). These soil processes might well be altered by the effect of litter accumulation on soil temperatures in fenced grasslands, probably more in the daytime (e.g., Fig. 1). These soil processes were not measured in this study but should be considered in future research.
Soil temperature was strongly coupled with soil moisture, such that litter accumulation could indirectly increase soil moisture in fenced grasslands (Appendix S1: Fig. S3) by decreasing soil temperature (Deutsch et al. 2010, Xia et al. 2010. Moreover, a decrease in soil temperature could delay thawing of soil in the early growing season (Facelli and Pickett 1991). But an increase in soil temperature and moisture later in the year could extend the growing season and postpone soil freezing (Watt 1970, Facelli andPickett 1991). For plant communities, lower soil temperatures could inhibit seed germination, delay plant phenology (Fig. 2), and reduce growth rates, all of which may affect the life history, phenology, and function of affected plants.

Effect of litter accumulation on plant phenology
Plant phenology is an indicator of plant growth that is sensitive to environmental changes such as air temperature, precipitation, and soil temperature and moisture (Cleland et al. 2007, Shen et al. 2018. Soil temperature is an important driver of plant phenology (Ji et al. 2017), an increase in which can advance leaf unfolding but delay leaf coloring in many regions (Peñuelas and Filella 2001). Similarly, an increase in soil moisture can advance plant phenology (Zhu et al. 2016). In this study, litter accumulation significantly delayed leaf-out, first flowering, and first withering of the three species in control plots of both grasslands (Fig. 2). The delay in leaf-out may have postponed first flowering and first withering. Alternatively, increased soil temperature and moisture ( Fig. 1; Appendix S1: Figs S2 and S3) may have extended the growing season . Differences in plant phenology between the species may have been driven by root morphology. Leaf-out was delayed more in the deeper-rooted S. grandis and S. krylovii than in the shallower-rooted L. chinensis (Fig. 2), maybe because of a greater availability of soil moisture to L. chinensis. In addition, denser litter appeared to have more negative effects on leaf-out of S. grandis and S. krylovii (Hou, personal observation).
Interestingly, our results were opposite to some previous studies. For example, Zhu et al. (2016) reported that plant phenology was significantly advanced in a fenced alpine meadow in Tibet, likely due to soil moisture rather than soil temperature. However, compared with the dominant species in the alpine meadow (e.g., Kobresia pygmaea and Potentilla saundersiana), L. chinensis, S. grandis, and S. krylovii have relatively deeper roots that are less sensitive to increases in shallow soil moisture. Additionally, the arid and semiarid grasslands in Inner Mongolia have lower water requirements and sensitivity than the Tibetan alpine meadow. Moreover, Zhu et al. (2016) stressed the reproductive stage of K. pygmaea was delayed after five years fencing. It indicated that fencing time might be another critical factor to regulate plant phenology. Our fencing time was longer than the previous study. Thus, v www.esajournals.org the effects of fencing and litter accumulation on plant phenology may vary depending on habitat and time since fencing, as well as plant habit, life form, and other traits.

Effects of litter accumulation on population characteristics
Species showed different sensitivities to litter accumulation in characteristics such as plant height, cover, and biomass. In this study, litter accumulation increased the height, cover, and root biomass of L. chinensis, but decreased cover, density, and root biomass of S. grandis and S. krylovii in the control plots of both grasslands (Fig. 3), indicating that L. chinensis has a higher tolerance to litter accumulation than S. grandis and S. krylovii. During the experiment, the abiotic environment of the control plots became shaded and moister, because of reduced light (Jensen andGutekunst 2003, Hou et al. 2019), lower soil temperature ( Fig. 1; Appendix S1: Fig. S2), and higher soil moisture (Appendix S1: Fig. S3). Plants may grow taller as they compete for light in more shaded environments, aided by more soil moisture (Wang et al. 2011). The shallower roots of L. chinensis may have allowed it to take advantage of the greater soil moisture (Bai et al. 2009); this species also tends to be better adapted to the moister environments , Wang et al. 2011) created by litter accumulation. It seems likely that litter accumulation alters asymmetric resource acquisition patterns of different species that then alters the original interspecific competitive relationships.
Fencing and subsequent litter accumulation changed the relative dominance of S. grandis or S. krylovii to L. chinensis in terms of aboveground biomass (Fig. 5). This phenomenon, where litter accumulation changes community composition and structure, has been reported in many fenced herbaceous plant communities, including in the Inner Mongolia (Hou et al. 2019), the Loess Plateau (Jing et al. 2013), the Mediterranean (Golodets et al. 2010), America (Weltzin et al. 2005), and England (Medina-Roldan et al. 2012). Litter accumulation has these effects through changes to the abiotic environment (such as light and soil temperature and moisture, detailed above), as well as direct effects on recruitment and growth through the physical barrier to seed dispersal and germination (Jensen andGutekunst 2003, Loydi et al. 2013). These effects on the population dynamics of species ultimately determine species richness and community evenness, composition, and structure, and are likely to drive community successional change in fenced communities (Hou et al. 2019). These findings emphasize the direct role of litter accumulation in plant population and community dynamics and suggest that litter should be taken into account in the management of grasslands.

Implications of litter on grassland management
Litter is regarded as one indicator of the health of grassland ecosystems (Wang et al. 2011). We have shown that fencing accelerates litter accumulation that then alters the abiotic environment, plant phenology, and plant population characteristics (Figs. 1-4), often in complex ways. In grassland management, the function of litter should be fully considered and it may be better to avoid excess litter accumulation. Additionally, although the grassland ecosystem service and function were improved (indicated by the increase in plant height, cover, and aboveground biomass of plants), nutrition and quality of forage grass may be reduced due to the mixture and decay of litter. Long-term fencing may also result in a decrease in the use efficiency of grassland resources. Therefore, grasslands should be appropriately used when enclosed to avoid any negative effects of long-term fencing.

CONCLUSIONS
Many grasslands in northern China have suffered from severe degradation in the last few decades. Fencing was regarded as a useful method for restoring degraded grasslands, due to the low investment and quick response of grasses. However, excess litter accumulates on the soil surface, generally decreasing soil temperature, although these effects vary over daily cycles and the growing season. These changes in the abiotic environment delayed leaf-out, flowering, and wilting of S. grandis, S. krylovii, and L. chinensis, with bigger effects on the deeperrooted S. grandis and S. krylovii, as well as on the performance and relative abundance dominance of all grasses. Changes to the successional dynamics mean that S. grandis and S. krylovii could potentially even be replaced by L. chinensis v www.esajournals.org in both fenced grasslands, respectively. Our findings elucidate the mechanisms of interaction between the abiotic environment and plant community after litter accumulation and suggest that litter should be fully considered in grassland management. Degraded grasslands should be appropriately managed during enclosure to avoid having to maintain fencing over the long term.