Effects of different fencing regimes on community structure of degraded desert grasslands on Mu Us desert, China

Abstract Grazing is one of the major anthropogenic driving factors influencing community structure and ecological function of grasslands. Fencing has been proved to be one of the main measures for rehabilitating degraded grasslands in northwestern China. However, data from combined empirical studies on the effects of different management regimes in desert grasslands are lacking. So we selected long‐term fencing (fenced since 1991), mid‐term fencing and seasonal fencing (fenced since 2002), and adjacent free‐grazing grasslands to investigate vegetation and soil properties on southwest Mu Us desert. Our results showed that fencing increased plant cover, height, aboveground biomass (AGB) of different plant life‐form groups, Shannon–Wiener diversity index, Evenness index, Simpson index, total soil nitrogen, total soil phosphorus, and soil organic matter, but decreased plant density, species richness, Richness index, soil bulk density, water content, and pH. However, 22–24 years of long‐term complete fencing might cause redegradation of vegetation and soil nutrients, characterized by the reduction of some vegetation properties, biodiversity, total AGB, and some soil properties. Seasonal fencing with 11–13 year was more beneficial to vegetation restoration than that with completely fencing measures. Our study suggests that appropriate artificial disturbances, such as seasonal fencing (winter grazing and summer fencing), should be used after long‐term fencing in order to maintain grassland productivity and biodiversity. These findings will help to provide theoretical support for vegetation restoration and sustainable management in grassland under grazing prohibition at Mu Us desert.

inducing a fragile ecological ecosystem that is sensitive to human activities and climate change (Zuo et al., 2014). Moderate grazing can maintain community diversity according to the "intermediate disturbance hypothesis" (Catford et al., 2012;Connell, 1979).
However, overgrazing leads to the excess output of energy and nutrient from plant-soil ecosystems to herbivores which results in degradation of grasslands (Chartier, Rostagno, & Pazos, 2011).
Overgrazing pattern aimed to get higher economic income may not only reduce vegetation cover, aboveground productivity, species diversity, and change the distribution of community spaces, but also exacerbate soil erosion and induce soil degradation, or even enhance desertification (Connell, 1979;Mekuria & Aynekulu, 2013;Nedessa et al., 2005;Wu et al., 2009). Presently, most of grasslands in northwest China show different degrees of degradation due to overuse (Wu & Loucks, 1992). How to restore degraded grasslands has become an urgent problem to be solved in recent decades.
Improving grazing management regimes is considered a main strategy to reestablish degraded grasslands (Lal, 2004;Mekuria & Aynekulu, 2013). Although a variety of biological and engineering practices have been carried out to restore the degraded grassland and protect undegraded grassland, exclosure is one of the most useful measures for restoring vegetation due to its low investment, extensiveness, simple implementation, and quick response (He, Zhao, Liu, Zhao, & Li, 2008;Liu, Zhang, Wang, & Yang, 2015). The effects of excluding livestock grazing on plant communities in arid and semiarid regions mainly depend on the duration of exclosure (Angassa & Oba, 2010) and some environmental factors, especially precipitation (Huxman et al., 2004;Miao, Guo, Xue, Wang, & Shen, 2015). Previous studies had indicated that fencing could significantly change species composition, improve soil physicochemical properties, and promote ecosystems resilience (Li, Cao, et al., 2017;Li, Zhang, et al., 2017;Schmiede, Donath, & Otte, 2009;Shang et al., 2013;Wu et al., 2009) through direct or indirect approaches (Socher, Prati, Boch, Müller, & Fischer, 2012) to restore degraded grassland ecosystem in a short time (Rooyen, Roux, Geldenhuys, Rooyen, & Merwe, 2014), and finally improve the growth and livelihood environment condition for humans (Gao et al., 2009;Shang, Ma, Long, & Ding, 2010). Shang et al. (2013) indicated 3 years of fencing could alter species composition and increase plant cover. Grassland with 6-8 years of fencing had higher aboveground vegetation productivity and cover, but lower plant density and species diversity compared with adjacent grazing grassland (Wu et al., 2009). Some studies showed that the 12 years of fencing significantly increased vegetation cover, height, Richness index, above-and belowground biomass (Zhu, Deng, Zhang, & Shangguan, 2016). However, because of lacking in human interference, the fierce competition of plants, caused by limited resources, may induce intense intra-and interspecific competition in exclosure grasslands (Angassa & Oba, 2010;Catford et al., 2012;Connell, 1979;Oba, Vetaas, & Stenseth, 2001). This may lead to a loss of biodiversity and ultimately cause redegradation of ecosystem after long-term fencing (Shang et al., 2010;Su, Liu, Xu, Wang, & Li, 2015).
However, there is a lack of research on long-term exclosure desert grasslands (e.g., >20 years).
Because of excessive grazing, local vegetation in Mu Us desert, northwest China, has been seriously degraded in late 20th century.
A variety of restoration measures, including fencing, have been implemented to restore the local environment in study area. In order to (a) assess the effects of different grassland management regimes on plant community and soil, and (b) reveal the relationship between community composition, aboveground biomass (AGB), biodiversity, and soil properties, four sites with different fencing time or management regimes were selected in this research. The change observed can be used as indicators of exclosure effectiveness and provide a foundation for sustainable management and utilization of restoring other grasslands in arid and semiarid areas.

| Experimental design and field survey
We used spatiotemporal approach to monitor the effects of fencing.
Four adjacent treatment sites: long-term completely fencing grassland (LG), mid-term completely fencing grassland (FG), mid-term seasonal fencing grassland (SG), and continued free-grazing grassland (CG), were selected in this study according to different management regime and fencing times. Each treatment site was about 20 ha (500 m × 400 m), and less than 10 meters between each plot. LG and FG sites were completely fenced since 1991 and 2002, respectively. SG site was fenced in 2002, however, grazed at the end of growing season (autumn and winter) with a medium livestock density of 0.75 Ovis aries per ha in this area (Liu, Xie, Dai, & Rao, 2008). CG site was with unlimited access for free traditional grazing with a moderate to severe grazing intensity of 1 sheep per ha .
The pore size of fences in the study area was greater than 15 cm which could remove large livestock, while some small herbivores, such as Lepus sinensis and Microtinae, were able to pass through the fences and active in exclosure area. The main characteristics of each site are shown in Table 1. The four sampling sites were in one common pasture under similar livestock grazing intensity before fenced, which ensure a relatively homogeneous vegetation conditions. They were also in the same continuous flat area and adjacent to each other, ensuring a comparable edaphic conditions. This disturbance and geographical conditions effectively reduced the potential for environmental spatial heterogeneity and resulted in a relatively homogeneous soil conditions.
In each treatment site, five 30 m × 30 m plots were established randomly with minimum distance of 50 m. Each plot was laid more than 20 m inside the margin of fencing to avoid edge effects. In these plots, three 2 m × 2 m subplots, treated as parallel set, were randomly selected to represent the vegetation condition. In each 2 m × 2 m subplot, we performed a quantitative vegetation inventory, including plant species, density, coverage, height, and AGB.

| Soil sampling
We sampled soil at the same time of the vegetation surveys. In each 30 m × 30 m plot, five randomly selected soil samples at the depth  of 0-30 cm were taken by bucket auger (7.5-cm inner diameter).
These samples then air-dried and mixed into a single sample. In each site, five samples were collected. All samples were passed through a 0.15-mm sieve for remove root, and the following nutrients were measured (Ma, Ding, & Li, 2016;Wu et al., 2009). Soil organic matter (SOM, g/kg) was analyzed by dichromate oxidation (Nelson & Sommers, 1982), total nitrogen (TN, g/kg) was used by Kjeldahl method (Bremner, 1996), and total phosphorus (TP, g/kg) was determined after digestion of soil with HClO 4 -H 2 SO 4 (Parkinson & Allen, 1975). Soil bulk density (BD, g/cm 3 ) was measured by the volumetric ring method (Wu, Liu, Zhang, Chen, & Hu, 2010). Soil water content (SWC, %) was measured by the oven-dried method (Li, Zhang, et al., 2017;Li, Cao, et al., 2017). Soil pH was measured in a soil: water ratio aqueous extract of 1:5 (PHS-3C pH acidometer, China). Each soil sample was measured with three replicates to ensure data accuracy. where S is the total species number found in each quadrat, N is the summation of plant importance values in limited area, P i is the relative important value represented by ith species.

| Community properties
ANOVAs indicated that fencing regimes had significant effects (p < 0.01) on community properties apart from H (

| Aboveground biomass and species richness of different life-form groups
The grassland sites with different fencing (LG, FG, and SG) had significant (p < 0.05) higher total AGB, while there was significant (p < 0.05) lower total species richness than free-grazing grassland (CG) ( LG, 0.93 in FG, 0.6 in SG, and 1.13 in CG, respectively (Table 4).

| Soil physicochemical properties
Compared with grazing grassland (CG), the exclosure grasslands (LG, FG, and SG) all significantly (p < 0.05) increased soil TN, TP, and SOM, while significantly (p < 0.05) decreased soil BD, SWC, and pH in three years ( LG and FG sites in 2014, and pH between LG and FG sites in all years had no significant difference (p > 0.05) ( Table 5).
Plant cover and height were significantly correlated with soil TN and TP (p < 0.05). Simpson index was significantly related (p < 0.05) with TN, TP, and SOM. There had no significant (p > 0.05) relationships between SWC and plant cover, and height, while had significant (p < 0.05) relationship between SWC and plant density (Table 6).

| D ISCUSS I ON
The restoration of degraded grasslands ecosystem after fencing is a complex and long-term ecological process (Zhu et al., 2016). Previous studies have shown that removing grazing could reduce nutrient and energy cycling from soil-vegetation ecosystem to livestock, and this has significant remarkable impact on vegetation index, diversity index, aboveground biomass, and soil nutrients (Harris, Moretto, Distel, Boutton, & Roberto, 2007;Liu et al., 2015;Wu et al., 2009;Zeng, Liu, Xiao, & Huang, 2017;Zhu et al., 2016). Our results indicated that 11-13 years (moderate time) of fencing would allow vegetation and soil nutrients to recover, while 22-24 years of exclosure might cause a redegradation trend of some community and soil properties. This is consistent with studies in other grasslands (Angassa & Oba, 2010;Liu & Zhang, 2018;Shang et al., 2010;Su et al., 2015;Yan & Tang, 2007). The 11-13 years of seasonal fencing grassland had greater plant density, species richness, biodiversity, and soil physical properties than contemporaneous fencing grassland.
Because there is a lack of disturbance from human and livestock, such as grazing, mowing, and farming (Medinaroldán, Paz-Ferreiro, & Bardgett, 2012), fencing could improve ecological environment of the vegetation for a short time (Catford et al., 2012;Rooyen et al., 2014;Shang et al., 2013). In the early days of exclosure, empty community niches, due to livestock consumption, are gradually taken up by restoring vegetation (Zeng et al., 2017;Zou et al., 2016). The proportion of some palatable perennial grasses increased due to abundant nutrients and water resources (Batoyun, Shinoda, Cheng, & Purevdorj, 2016), but annual herbs trended to decrease (Wu et al., 2009;Zheng, Cao, & Wang, 2005). The colonization capacity of different plants influences the effects of vegetation restoration (Shang et al., 2013). The changing of community composition may induce a further impact on ecosystem (Deng et al., 2014;Mekuria & Aynekulu, 2013 TA B L E 6 Pearson's correlation coefficients between community properties and soil physicochemical properties improves nutrients retention (Zhang, Yang, & Zepp, 2004). There is also an increase in litter input which improves soil nutrient content (Li, Zhang, et al., 2017;Li, Cao, et al., 2017;Medinaroldán et al., 2012;Mekuria & Aynekulu, 2013). The improvement of soil structure and the environment, in turn, promote the growth and development of vegetation (Deng et al., 2014;Wu et al., 2010;Zhu et al., 2016).
However, as fencing time increases, the competition (intra-and interspecific) increased because of the limited natural resources such as light, water, and nutrient (Huston, 1994;Oba et al., 2001;Wal et al., 2004). Some research notes that the restoration process of degraded ecosystem is accompanied with community succession (Deng et al., 2014;Shang et al., 2010;Wu et al., 2009). In restoring grasslands, some tall and strong germinating plants with stronger water and nutrient use, especially Kobresia groups in the study area (Liu & Zhang, 2018), are gradually demonstrating their competitive advantage in the process of community succession after long-term fencing, which inhibit the growth of annual groups in herbaceous layer (Deng et al., 2014). After enclosed for a certain time, the litter accumulation combines with the formation and development of soil crust impedes infiltration of precipitation (Aubert et al., 2011), which deteriorates soil moisture (Oba et al., 2001;Read, Duncan, Vesk, & Elith, 2011;Yang, Chu, Chen, Wang, & Bai, 2014). The decrease in soil moisture directly affects the activity of the soil microbial communities and limits their capacity to decompose certain compounds, leading to the decrease in soil nutrients (Hueso, García, & Hernández, 2012). In addition, lower soil moisture restrains the growing development of seedlings and the decomposition of litters and dead roots (Čatský, 2001;Yan, Tang, Xin, & Wang, 2009), and reduces the circulation velocity of energy in plant-soil ecosystem (Harris et al., 2007).
The increase of annual grasses after long-term fencing results in less root which also has negative effects on the accumulation of soil nutrients (Čatský, 2001). As a consequence, after long-term fencing, grasslands may experience have a degradation trend in plant and soil physicochemical properties (Angassa & Oba, 2010;Su et al., 2015), characterized by lower plant cover, density, total AGB, biodiversity, soil TN, TP, and SOM in LG site than in FG site.
Grazing, as one of the major traction of vegetation succession in northwest China, has some passive effects on germination, growth, development, mortality, and propagation of plant species, and in turn leads to shifts in community quantity characteristics and diversities (Mekuria & Aynekulu, 2013;Wassie, Sterck, Teketay, & Bongers, 2009). Studies have shown that different management regimes had dissimilar influences on grassland ecosystem (Li, Zhang, et al., 2017;Li, Cao, et al., 2017;Mekuria & Aynekulu, 2013). The selective feeding and patchy defecation of livestock increase soil nutrients by accelerating circulation flow of material energy, which accelerates the restoration of degraded soil and causes variation in community composition (Harris et al., 2007;Li, Cao, et al., 2017;Sheppard, Hodge, Paynter, & Rees, 2002). On the one hand, livestock are more likely to mistaken eat the seeds of palatable perennial herbs when ingest leaves. And then, these seeds excrete with feces which aggravates the distribution of perennial herbs (Bertiller & Ares, 2011). Conversely, this behavior provides a better living environment for seed germination, increases soil fertility, and enhances environmental heterogeneity, which provides favorable conditions for the invasion of alien species (Keeley, Lubin, & Fotheringham, 2003;Zuo et al., 2008). The activities of livestock, such as grazing and trampling, not only directly impact aboveground productivity of dominant species, weaken their competitive advantage, and reduce the capacity to utilize resources, but also reduce the barrier effect of surface crust on seeds and promote the growth of lower perennial and annual herbs (Chartier et al., 2011;Gómez et al., 2012;Inderjit, 2005;Miao et al., 2015).
These factors increase the invasiveness of community and lead to increase in species richness and decrease in biodiversity (Brooker et al., 2008;Chambers et al., 2014). Therefore, according to "intermediate disturbance hypothesis," appropriate human interference, such as seasonal grazing, should be taken to maintain plant cover and biodiversity on fencing grasslands after excluding the interference of human and herbivorous livestock for appropriate time.

| CON CLUS IONS
In this study, moderate time fencing (11-13 years) could improve some plant properties and soil nutrients, while 22-24 years of long-term fencing might cause a decreasing trend of those vegetation and soil indexes which indicated the redegradation of grassland ecosystem.
Additionally, seasonal fencing grassland (11-13 years) was better to maintain biodiversity. All these results indicated that different fencing regimes are all effective to recover the degraded desert grasslands in Mu Us desert. Our results lead us to recommend that winter grazing and summer fencing could be used as a beneficial management strategy.

ACK N OWLED G M ENTS
We are grateful to Wanxue You, Cunlin Bai, Zhirong Chen, and other staff of Habahu Nature Reserve of China for valuable assistance with fieldwork and laboratory work. Thanks to Xiaodan Liu, Lili Wang, Xiaoteng Xu, Zhishu, Wang, Zhiru Hao, Zhanchao Fan, and Shuaishuai Niu for vegetation sampling. We also thank the Editor and anonymous reviewers for their helpful comments on improving previous manuscript.

CO N FLI C T S O F I NTE R E S T
None declared.

DATA ACCE SS I B I LIT Y
I agree to deposit my data to a public repository. All data input files: