Grazing exclosures solely are not the best methods for sustaining alpine grasslands

Study site

The study area was located on the Sanding Village in the northeastern edge of the Qinghai-Tibet Plateau, Kangle Town, Sunan County, Zhangye City of Gansu Province, in China (99°48′E, 38°45′N, and 3,200 m above sea level). The annual average precipitation was 255 mm (1985–2017), with ∼85% occurring during the growing season (May–September) (Fig. 1B). The average annual temperature was approximately 3.8 °C (1985–2017) (Fig. 1A). The annual cumulative temperature (≥0 °C) was approximately 2323.9 °C (1985–2017) (Fig. 1C). The vegetation growth period was from June to September i.e., approximately 4 months. The type of grassland belonged to alpine meadow. The species richness was high, with 12–24 species per m² in this vegetation meadow. The vegetation was divided into five classes that included (gramineous grasses (GG); sedge grasses (SG); leguminous grasses (LG); forbs grasses (FG) and noxious species (NG)). The dominant species included: Kobresia humilis, Polygonum viviparum, Potentilla fruticosa and Caragana sinica (Yang et al., 2012) (Species are shown in Table 1). The soil of the alpine meadow belonged to meadow soil with high calcium content (Kemp & Michalk, 2011).

Experimental design

The study area, including all research sites, was slightly degraded because it was used as summer pasture and mainly grazed from June to September by Gansu Alpine Fine-wool sheep before 2005. Two treatments, i.e., non-grazed and grazed treatments, were started in 20th June, 2005 under the same conditions, while sample collection was started from 2015. In each treatment three sites, each site occupied 4 ha area (approximately 100 m away from each other, to insure an unification of experiment conditions, all sites had similar slope gradient, aspect, elevation and soil type (Cheng et al., 2016; Li et al., 2017; Wen et al., 2018)), were randomly selected over a homogeneous area (total 24 ha area). At each site, using the line transects method, three 100 m ×100 m monitoring blocks (almost at a distance of 50 m) with the same conditions were selected, altogether eighteen sample blocks for two treatments were chosen. The non-grazed sites had been excluded from livestock grazing for 10 years. The grazed sites were fenced for the whole year and were freely grazed from 20th June to 20th August with a 4.5 livestock units/ha (Gansu Alpine Fine-wool sheep, 18-month-old female sheep, average 35 kg live weight) during the grazing period. The grazing experiment started on 20th June, 2005. Animal welfare and experimental procedures were carried out in accordance with the Guide for the Care and Use of Laboratory Animals, Ministry of Science and Technology of China and were approved by College of Animal Science and Technology of Gansu Agricultural University. Every effort was made to minimize animal pain, suffering and distress and to reduce the number of animals used. Gansu Alpine Fine-wool sheep management followed traditional practice, in which grazing sheep were kept in the grazing sites day and night with freely available drinking water. The paddock was owned by a local farmer, who agreed to its use for this experimental study. There were no endangered or protected species within the paddock.

Aboveground vegetation community survey, plant and soil sampling

Successive samples were collected on mid-August of 2015, 2016 and 2017 when aboveground biomass at the peak of biomass production. In every experimental block, five random sampling of quadrats (1 m ×1 m) were done, having distance between each quadrat over 1 m from edge to edge to eliminate the effect of margin. Species composition, life type, edibility, coverage, aboveground biomass and density were recorded in each quadrat and each plant species was clipped to 1-cm stubble height. Additionally, the plant functional types were divided into five classes that included GG (grass species type), SG (sedge species type), LG (leguminous species type), FG (forbs species type) and NG (noxious species type) (Wu et al., 2009; Zhang et al., 2018). Edible plants species mean those plants that can be eaten by animals while noxious species mean noxious plants. Noxious species means plant species that are classified as undesirable, noxious, exotic, injurious, or poisonous, pursuant to local government law, which are of foreign origin, and can directly or indirectly injure crops, other useful plants, livestock, or poultry (Parker, 1949). Dry matter in each functional group under the constant weight of every quadrat drying at 70 °C for 48 h was determined. A total of 270 quadrants were recorded during 3 years of experiment for long-term non-grazed and grazed treatments. Total coverage, aboveground biomass (dry matter) and species density of alpine meadow plant functional types were recorded. The information of all species is shown in Table 1. Species abundance was calculated by using the number of species in each square. Species richness (S), Shannon diversity index (H) and Evenness index (E) of the vegetation was calculated using the formula:

${\displaystyle \mathrm{Species richness}\left(R\right):R=S}$ ${\displaystyle \mathrm{Shannon diversity index}\left(H\right):H=-\sum _{\mathrm{i}=1}^{\mathrm{s}}\left(Pi\ln Pi\right)}$ ${\displaystyle \mathrm{Evenness index}\left(E\right):E=\frac{H}{\ln S}.}$ Where S, H, P_i represents total species of alpine meadow vegetation community, Shannon diversity index and density proportion of i species, respectively (Shannon, 1948).

Along with this, surface soil was also removed from five spots of each quadrat. The center and four diagonal corners of each sampling quadrat were selected using the earth boring auger in three soil layers: 0–10, 10–20 and 20–30 cm. Soil samples from five spots of each quadrat were mixed together to use as one soil sample. There were 270 soil samples in total from same soil layer for long-term non-grazed and grazed treatments during 3 years of experiments.

Nutrients chemical analysis

Nutritional values of four edible functional types were evaluated. Plant samples were dried, crushed and passed through 1 mm mesh screening by using Foss Tecator Cyclotec 1,093 sample mill. The plant nutrition parameters included crude protein (CP), neutral detergent fiber (NDF) and in vitro ture digestibility (IVTD). Plant nitrogen (N) concentration was measured using Foss fully automated Kheltec 8400 (Feldsine, Abeyta & Andrews, 2002) and utilized to calculated crude protein (CP) by using formula: % CP = % N ×6.25. NDF was analyzed using an ANKOM 2000 Fiber Analyzer on the basis of the two-stage method (Tilley & Terry, 1963) while in vitro true digestibility (IVTD) was determined using an Ankom F57 filter bag (Goering & Van Soest, 1970; Van Soest, Robertson & Lewis, 1991).

Soil samples were passed through 0.14 mm mesh screening after air-drying. The measurement of soil samples according to methods (Miller & Keeney, 1982). Total nitrogen (TN) was obtained by the semi-micro Kjeldahl method. Total phosphorus (TP) and total potassium (TK) were determined with an inductive coupled plasma (ICP) emission spectrometer after digestion of the samples in concentrated HNO₃. Available nitrogen (AN) was determined using the continuous alkali-hydrolyzed reduction diffusion method. Available phosphorus (AP) was extracted with sodium bicarbonate and determined by Olsen method. Available potassium (AK) was determined by H₂SO₄-HCLO₄ digestion and the molybdenum antimony-ascorbic acid colorimetric method.

Data analysis

Mixed Model option in SPSS version 19.0 (IBM Corp., Armonk, New York, USA) based on an autoregressive covariance structure through ANOVA analyzing data was used. There were 270 observations [2 treatments ×9 blocks ×5 quadrants ×3 years] for each functional type variables (Coverage, Plant density, Aboveground biomass, CP, IVTD and NDF) and soil property variables in each soil depth (Total N, Total P, Total K, Available N, Available P and Available K). Repeated measurement analysis for each functional type variables and soil property variable in every soil depth were performed using a mixed-effects model, including grazing treatment (non-grazed, grazed), and blocks were selected to study the fixed effects with year (2015, 2016, 2017) as a repeated effect and their interactions. The ANOVA analysis was followed by least significant difference (LSD) tests (P < 0.05).

Vegetation characteristics response to non-grazed treatment and year variations

The vegetation coverage and aboveground biomass of non-grazed treatment were significantly higher than grazed treatment (Fig. 2; Table 2). The vegetation density, Species richness (R), Shannon diversity index (S) and evenness index (E) of non-grazed treatment were considerably lower than grazed treatment (Fig. 2; Table 2). The sampling year notably influenced the vegetation coverage and density (Fig. 2; Table 2). The effect of non-grazed treatment on the coverage and density were highly influenced by sampling year (Fig. 3; Table 2).

In comparison to grazed treatment, non-grazed treatment considerably increased the coverage of all five vegetation types (Fig. 3; Table 2). Significant increase in the aboveground biomass of GG, SG was observed during non-grazed alpine meadows. Non-grazed treatment also increased the density of all five vegetation types (Fig. 3; Table 2). The sampling year only increased the coverage of LG and NG. The sampling year also increased the density of all vegetation types, except FG. However, significant effects of interaction between non-grazed treatment and sampling year on the aboveground biomass of SG, LG and FG was observed (Fig. 3; Table 2).

Non-grazed treatment showed a significant decreasing trends over a period of time on the density of all five vegetation types (Fig. 4; Table 2) displaying minimum inclination in 2017 (Fig. 4; Table 2). In case of grazed treatment, the density of all five vegetation types first decreased and then showed an upward trend during the experimental period (Fig. 4; Table 2) with minimum reading during 2016 (Fig. 4; Table 2). The non-grazed treatment displayed first decreasing and then an upward trend in the aboveground biomass of all five vegetation types (Fig. 4; Table 2) with minimum value occurring during 2016 (Fig. 4; Table 2).

Vegetation nutritional values response to non-grazed treatment and year variations

In comparison to grazed treatment, non-grazed treatment significantly decreased the CP content of all vegetation types, except NG (Fig. 5; Table 3), it also decreased the IVTD of all vegetation types, except NG (Fig. 5; Table 3). A considerable decrease in the NDF content of all vegetation types, except NG (Fig. 5; Table 3) during non-grazed treatment was also recorded. The sampling year showed positive effect on the CP content of GG and SG (Fig. 5; Table 3). The significant increase in the IVTD of GG and LG were recorded during non-grazed treatment (Fig. 5; Table 3). The NDF content of GG and SG increased were observed during the sampling year (Fig. 5; Table 3). The relation of non-grazed treatment and sampling year showed significant effect on the CP content of all vegetation types, except NG (Fig. 5; Table 3). While in case of IVTD, only SG showed significant effect (Fig. 5; Table 3). The NDF contents of SG and LG were observed effected during the study of interaction of non-grazed treatment and sampling year (Fig. 5; Table 3) were seen.

The CP and IVTD contents of all vegetation types, except NG in non-grazed alpine meadows showed a gradually decreasing trend (Fig. 6; Table 3) while in case of grazed alpine meadows, the CP and IVTD contents of all vegetation types, except NG firstly showed decreasing trend and then an upward trend was observed during the experiment period (Fig. 6; Table 3) both showing lowest value during 2016 year (Fig. 6; Table 3). However, the NDF content of all vegetation types, except NG showed an increasing trend in non-grazed alpine meadow during experiment, displaying highest value during year 2017 (Fig. 6; Table 3). In case of grazed alpine meadows, the NDF content of all vegetation types, except NG first increased and then a downward trend was recorded (Fig. 6; Table 3), the highest value was observed in year 2016 (Fig. 6; Table 3).

Soil properties response to non-grazed treatment and year variations

As compared to grazed treatment, non-grazed treatment significantly increased all six measured soil properties in the 0–10 cm soil layer while an increase in all measured soil properties, except TP were recorded in the 10–20 cm soil layer. In 20–30 cm soil layer all measured soil properties, except TN and TK considerably increased (Fig. 7; Table 4). Sampling year also effected the soil properties such as significant increase in the soil TK, AN and AK in the 0–10 cm soil layer was observed and an increase in the soil AN and AK in the 10–20 cm soil layer (Fig. 7; Table 4) was also recorded. The interaction of non-grazed treatment and sampling year significantly affected the soil TN, TP, AN and AK in the 0–10 cm soil layer. A significant effect on all measured soil properties, except TK in the 10–20 cm soil layer and on all measured soil properties, except TP and TP in the 20–30 cm soil layer were observed (Fig. 7; Table 4).

In non-grazed alpine meadows, the all measured soil properties, except TK of three soil layers i.e., 0–10, 10–20 and 20–30 cm showed a gradually increasing trend during the whole experimental period, the lowest value occurred in 2017 year (Fig. 8; Table 4). While in case of grazed alpine meadows, the TP, TK and AP of three soil layers (0–10, 10–20 and 20–30 cm) first decreased and then an upward trend was recorded during the whole experimental period with the lowest value displayed in 2016 (Fig. 8; Table 4). Overall TN, AN and AK of three soil layers (0–10, 10–20 and 20–30 cm) showed a gradually decreasing trend during the whole experimental period, and displayed the lowest value in 2017 year (Fig. 8; Table 4).

Vegetation characteristics response to non-grazed treatment and year variations

The restoration of degraded grassland ecosystem is a complex and long-term ecological process (Cheng et al., 2016; Török & Helm, 2017; Török et al., 2018). Non-grazed treatment is generally regarded as a useful tool to restore the productivity of degraded grasslands ecosystem (Spooner, Lunt & Robinson, 2002). Many studies have shown that the restoration of grassland by non-grazed grasslands can be assessed by biomass, coverage, density and diversity of vegetation (Wilkins, Keith & Adam, 2003; Cheng et al., 2016). In our research, we selected aboveground biomass, coverage, density, biodiversity and richness to assess the impact of non-grazed treatment on vegetation. Research has shown that climatic factors are also the main driving force of degradation, greater than overgrazing for alpine grassland (Niu, Ma & Zeng, 2008). The variation of vegetation biomass between years is primarily influenced by local rainfall, temperature and sunshine radiation (Akiyama & Kawamura, 2007; Niu, Ma & Zeng, 2008; Wu et al., 2009; Miao et al., 2015; Ren et al., 2016). In our study area, the annual average temperature and ≥0 °C accumulative temperature gradually increased, while annual rainfall decreased (Fig. 1). That indicates that there was a warmer and dry trend in the local climate, which might be a very influential factor in the succession of plant communities (Bai et al., 2004).

Our study showed that long-term non-grazed exclosures remarkably increased the aboveground biomass and coverage of vegetation, but decreased the density and biodiversity of vegetation (Fig. 2; Tables 1 and 2). Similar results have also been reported in other grassland types (Haugland & Froud-Williams, 1999; Wu et al., 2009; Wang et al., 2012). Non-grazed exclosures increased the coverage and aboveground biomass of gramineous (GG) and sedge (SG) plants in alpine meadow communities by excluding intake of herbivores, which had good palatability for domestic animals (Figs. 3 and 4; Tables 1 and 2). Studies have shown that forage with good palatability is more competitive than those forage with poor palatability, and non-grazed exclosures significantly increased the biomass of gramineous and sedge species which have good palatability (Gallego et al., 2004; Wu et al., 2009). These results are consistent with previous reports (Gallego et al., 2004; Wu et al., 2009; Shang et al., 2013), supporting that non-grazed exclosures benefits for the improvement of biomass and coverage of four plant functional types (GG, SG, LG and FG) and inhibition in the development of noxious species type (NG). Livestock grazing accelerated the loss of plant roots and leaf biomass, promoted the recycling of nutrients (Semmartin, Garibaldi & Chaneton, 2008), and decreased the vegetation biomass of grazed alpine meadows. However, non-grazed exclosures showed a negative effect on the density and biodiversity of plant functional types. In high biomass grasslands, the loss of vegetation diversity might be due to greater competition of canopy resources (i.e., light and air) (Huston, 1994). Some of the less competitive species had limited availability of light or nutrient (Grime, 1998; Van Der Wal et al., 2004), resulted in their decrease density or species disappearance. On the contrary, grazed treatment decreased the biomass of dominant functional types (GG and SG) (Table 1), allowing other functional types in the community to have more development opportunities and promoting balanced development of the community.

The alpine meadows 11–13 years non-grazed exclosures reflected changes in functional types from smaller to larger, and in density from higher to lower (Table 1). In non-grazed alpine meadows, the biodiversity of five plant functional types (GG, SG, LG, FG and NG) decreased (Table 1). The density changes of five plant functional types (GG, SG, LG, FG and NG) in the non-grazed area means that the number of species in the non-grazed alpine meadow was less, total aboveground biomass was higher and number of newly appeared species is fewer (Figs. 2 and 3; Table 1). This indicated that the non-grazed exclosures has affected the concealment of the habitat and led to the loss of species diversity. The concealment of the habitat determines the supplement of local plant seedlings (Oba, Vetaas & Stenseth, 2001) and disruption of some unusual plant species (Inderjit, 2005). The species density and diversity in the fenced alpine meadow is significantly higher than grazing alpine meadow. Grazed treatment inhibits the development of dominant community (GG and SG), increases spatial heterogeneity, and makes other functional types (LG, FG and NG) in the community to grow to achieve a balanced development (Figs. 3 and 4), which is consistent with the previous reports (Begon, Harper & Townsend, 1990; Sheppard et al., 2002; Schippers & Joenje, 2002; Holdo et al., 2007; Wu et al., 2009).

Grazed treatment is regarded as a key factor leading to grassland degradation; meanwhile it is also a main driving force for grassland succession (Holdo et al., 2007). Plant diversity is mainly dependent on grazing intensity. Overgrazing may lead to the grassland degradation and biodiversity loss and light grazing may lead to grassland succession to woodland and the loss of grassland habitats. Not only is grazed intensity important, but the time of grazing and the type of grazed livestock are also important (Hulme et al., 1999). It is necessary to do more research about the effects of non-grazed and grazed on alpine meadows, especially in terms of global climate change (Watkinson & Ormerod, 2001).

Vegetation nutritional values response to non-grazed treatment and year variations

It has been reported that grazing usually increased the forage nutritional value (Bai et al., 2012; Schönbach et al., 2012), which will in turn affects the performance of livestock (Mysterud et al., 2001; Mysterud et al., 2011; Lin et al., 2011; Müller et al., 2014). Our research showed that grazed treatment significantly increased the nutritional value (CP and IVTD) of four edible functional types (GG, SG, LG and FG), which is consistent with other studies (Schönbach et al., 2009; Fanselow et al., 2011; Ren et al., 2016). Firstly, a large amount of nitrogen was stored in the stems and leaves, because grazed forage had a higher relative absorbance of nitrogen and it was moved into the young tissue of shoots and leaves when herbage was taken (Lambers et al., 2009; Fanselow et al., 2011). Secondly, nitrogen originating from the dung and urine of grazed livestock had a positive effect on nitrogen concentrations of forage and herbivore excretions usually accelerated the rate of mineralization of the soil surface by senescent plant litter. As the soil mineral nitrogen increased, the nitrogen content in plants also increased (Semmartin, Garibaldi & Chaneton, 2008; Wang et al., 2009; Jiang et al., 2012; Miao et al., 2015). Finally, grazed treatment could affect the nutritional value of plants by using young and protein-rich parts and regenerating it, instead of the aged parts of plants (Mysterud et al., 2001; Schönbach et al., 2009; Ren et al., 2016). Due to the regeneration of new tissues under grazing pressure, the maturation and lignification of the species are delayed, and the CP content increased; our findings support these results (Milchunas et al., 1995; Garcia et al., 2003).

The CP and IVTD of four edible functional types (GG, SG, LG and FG) showed a gradually decreasing trend with the increase of non-grazed time, while NDF gradually increased. The CP and IVTD of functional types in grazed alpine meadows showed a decreasing trend at first and then increasing trend with time was observed. Contrary to the changing trend of CP and IVTD, NDF content showed at first rising trend and then decreasing trend. This might be due to rainfall in the study area. As reported previously that variation in precipitation rate between years affects the herbage nutritional values (Miao et al., 2015). The herbage nutritional value depends on the amount of precipitation, which increases with increasing rainfall, consistent with past research (Schönbach et al., 2009; Müller et al., 2014; Miao et al., 2015). Our results also indicated that nutritional value of four edible functional types in relatively wet years (2015 and 2017) was higher than in dry year (2016) (Fig. 1). Plant growth was limited by the amount of precipitation and water availability (Miao et al., 2015). In relatively wet years, abundant precipitation accelerated soil water utilization rate and soil mineralization, and promoted the ability of plants to absorb nitrogen (N), thus promoting biomass production (Austin et al., 2004; Xu & Zhou, 2005). Meanwhile, drought can caused severe water stress, resulted in rapid ripening of plants, thereby reducing the concentration of N in forage. Therefore, water stress in drought years increases forage fibrosis and reduces forages digestibility, while in wet years due to high rainfall maturation process delays, forage fibrosis reduces and in turn forage digestibility was improved. However, in the case of non-grazed alpine meadow, the CP and IVTD of four edible functional types (GG, SG, LG, and FG) did not show any increasing trend with increase in rainfall. This may be due to the fact that as the non-grazed time was increased, biomass accumulated with the passage of every year. Mature and aged tissues inhibited the germination and growth of young tissues from the seedlings in the growing season, causing the CP and IVTD to decrease and NDF to increase every year.

Livestock grazing can significantly change the structure and nutritional value of vegetation and their trampling behavior and excrement can also affect the community structure and soil properties of the ground (Gibson et al., 2000). Therefore, vegetation succession, functional types characteristics and nutritional values are closely related to livestock grazed. For the succession of grasslands vegetation and utilization of grassland resources, regular grazed and non-grazed treatment are beneficial to grasslands management.

Soil properties response to non-grazed treatment and year variations

As the non-grazed time increased, the soil TN, TP, AN, AP and AK went significantly up, showing that the soil nutrients of degraded alpine meadow were being restored by fencing approach (Jing et al., 2014), indicating that natural succession of degraded soils in alpine meadow areas of the QTP could improve soil fertility. The improvement of soil properties in alpine meadows with increased of non-grazed time had two explanations: first, the productivity of vegetation has a direct impact on the accumulation of litter. With the accumulation of litter and in the presence of soil moisture, litter decomposition rate enhanced and soil nutrients showed an increasing trend (Wu et al., 2009). Secondly, higher soil nutrients might be due to higher community coverage. Previous studies have found that vegetation coverage has an obvious impact on the quality of soil nutrients (Zhang et al., 2011), our results further supports these findings.

The interaction of soil and plant is a complex process (Lambers et al., 2009). The movement of energy and nutrients in soil can directly and indirectly reflect the species composition, productivity and nutritional value of vegetation (Venterink, 2011). In non-grazed grasslands, firstly, the plant community of non-grazed grassland locked-in nutrients (Harris et al., 2007) in their tissues, reduced the outflow of energy and nutrients from soil-plant system to the consumer (grazed livestock), especially gramineous (GG) and sedge (SG) functional types had high productivity and good quality (Moretto & Distel, 1997). Vegetation resources (coverage and productivity) were significantly improved with the increase in non-grazed time. These resources could go back to the soil by the decomposition of the litter layer (Bardgett & Wardle, 2003; Wu et al., 2009). Secondly, non-grazed removed the trampling effect of grazing livestock, improved soil characteristics, increased water interception and improved vegetation status (Li et al., 2007). Along with the development of aboveground vegetation, the better vegetation conditions reduced the wind erosion and some nutrients richer particles and dust was captured in the soil (Liu et al., 2007). Thirdly, the improvement of soil nutrients had a positive regeneration effect on the aboveground biomass and structure of plant functional types, because the utilization of higher nutrient levels is beneficial for the competition of gramineous (GG) and sedge (SG) functional types to other species (Van Der Wal et al., 2004). Finally, decrease in the quantity of rodents (Myospalax fontanierii and Microtus leucurus) has a positive effect on soil biological communities and soil processes by altering the input of soil resources (Bardgett & Wardle, 2003). However, for grazed grasslands, some energy and nutrients flow from soil-plant system to livestock by grazing, which at first altered soil properties, reduced the litter and root biomass that were fed back into the soil after decomposition (Gao et al., 2008); Secondly, the edible functional types grazed by livestock had higher litter decomposition rate and efficient soil nitrogen than inedible functional group forage (Moretto & Distel, 2002). Finally, long-term trampling by livestock transforms soil composition, infiltration rates, bulk density, soil porosity, limiting soil respiration and reducing soil microbial activity (Holt, 1997).

Therefore, soil nutrient has a positive effect on the aboveground biomass, composition and nutritional value of plant functional types. However, this study only analyzed the soil chemical characteristic. In future, the relationship between plant functional types and soil physics or biology should also be considered.