An evaluation of government‐recommended stocking systems for sustaining pastoral businesses and ecosystems of the Alpine Meadows of the Qinghai‐Tibetan Plateau

Abstract China introduced the “Retire Livestock and Restore Grassland” policy in 2003. It was strengthened in 2011 by additional funding for on‐farm structures. On the Qinghai‐Tibetan Plateau (QTP), fences were erected, livestock excluded from degraded areas, rotational stocking introduced, nighttime shelters were built, forages grown, and seed sown. However, the effectiveness of these actions and their value to Tibetan herders has been questioned. We conducted a sheep stocking experiment for 5 years in an Alpine Meadow region of the QTP to evaluate stocking options recommended by Government. Cold and warm season stocking each at three rates (0, 8, and 16 sheep/ha) and continuous stocking at 0 and 4 sheep/ha were compared. We measured live weights of sheep, plant species richness and evenness, root biomass and carbon (C), nitrogen (N) and phosphorus (P) contents of the 0–10 cm of soil. We found that resting grassland from stocking during the warm season for later cold season stocking significantly reduced plant species richness and evenness and root biomass but not soil C, N, and P. During cold season stocking, live weights of sheep declined whether at a stocking rate of 8 or 16 per ha. In contrast, sheep continuously stocked on grassland at 4 per ha gained weight throughout both the warm and cold seasons and plant species richness and evenness were maintained. Warm season stocking at 8 and 16 sheep/ha increased plant species richness and root biomass but reduced plant species evenness. Resting these alpine grasslands from stocking in the warm season has adverse consequences for plant conservation. Fencing from stocking in the warm season is not justified by this study; all grassland should be judiciously stocked during the warm season to maintain plant species richness. Neither resting nor stocking during the cold season appears to have any adverse consequences but sheltering and in‐door feeding of sheep during the cold season may be more profitable than cold season stocking with use of open nighttime yards.

The nomadic stocking systems practiced by herders for many centuries are thought to have sustained the functionality of the grasslands (Brondizio & Le Tourneau, 2016;Miller, 1999). In contrast, the policies successively introduced by the Central Government of China since 1950 have been challenged (Harris, 2010;Qiu, 2016;Wang, Lassoie, et al., 2015) on grounds that they are not sustainable. In the nomadic systems that prevailed prior to 1950, stock were periodically moved during the warm seasons. Stock movement was determined by forage availability across the landscapes and a highly regulated social system that required the grassland to be maintained in a productive and functional state (Miller, 1999). From 1950, herders increasingly grazed stock in a collective manner on areas allocated by government using a twoseason stocking system; grazing their stock on mountain slopes in the warm season and in the valleys for the cold season.
In 1980, the Government introduced the "Grassland Household Contract System" policy and the two-season stocking system was mostly replaced by sedentary systems whereby herders were allocated defined areas of grassland. Unfortunately, the stock of each herder increased in number because personal wealth of herders is determined by the number, and not the condition, of their stock (Cao, Xiong, Sun, Xiong, & Du, 2011). The degradation of the QTP grassland of the QTP, especially of the nonmeadow grasslands, rose from 24.5% in 1980 to 34.5% in 1990 . It was commonly believed that the degradation predisposed an increase in Plateau Zokor and Pika populations (Kang, Han, Zhang, & Sun, 2007) but this has been challenged (Pang & Guo, 2017;Pech, Arthur, Zhang, & Lin, 2007).
In 2003, the Government introduced the "Retire Livestock and Restore Grassland" policy and fences were erected for resting from stocking of grassland in the warm season and for the long-term exclusion of stock from severely degraded areas, shelters for protecting stock from cold and predators were built, forages for supplementary feeding during the long cold period were grown, and grassland species were sown to renew degraded pasture. Some of these measures received Government financial support. The policy also recommended that rotational stocking, as practiced in other countries (Briske et al., 2008;Welchons et al., 2017), be scientifically evaluated on the QTP and if successful be promoted for adoption by herders (Du, Yan, Chang, Wang, & Hou, 2017;Liu, Li, Ouyang, Tam, & Chen, 2008;Wang, Wang, He, Liu, & Hodgkinson, 2010;Wang, Zhao, Long, & Yang, 2010).
The major problem for stocking businesses on the QTP grasslands is managing yearlong production systems in a grassland with a short growing season (Shang et al., 2014). The relationship between grassland and stock productivity during the warm season is well-established from grazing experiments conducted in Australia, England, and the USA and synthesized into a global model (Jones & Sandland, 1974). This model is a linear negative relationship between stocking rate and weight gain per animal from which the total stock productivity per unit of grassland is calculated and takes the form of a quadratic relationship. Later, Wilson and MacLeod (1991) reasoned that there would be loss of linearity if degradation occurred and they developed theoretical relationships from the global model of Jones and Sandland to show that the loss of linearity would reduce stock productivity and business profitability. Stocking rate studies on the QTP grasslands, conducted during the warm season, for both yak and sheep, however, have shown no loss of linearity in animal production at higher stocking rates (Dong, Zhao, Wu, & Chang, 2015;Kemp et al., 2013;Miao, Guo, Xue, Wang, & Shen, 2015;Sun, Angerer, & Hou, 2015). However, stocking rates in these experiments were on the conservative side; very high stocking rates in line with common practice were not imposed.
Loss of linearity would be preceded by change in the botanical composition of the grassland or other floristic attributes and/ or change in soil properties (Ludwig, Tongway, Freudenberger, Noble, & Hodgkinson, 1997). These are early warning signs of an approaching critical threshold beyond which loss of linearity will occur (Westoby, Walker, & Noy-Meir, 1989). Comparison of a range of stocked sites in QTP grasslands that differed in perceived degradation status have demonstrated changes in plant species richness and soil attributes (e.g., Wang, Lassoie, et al., 2015;Wen et al., 2013;You et al., 2014).
However, in these studies, the stocking histories at the sites were not taken into account. It is possible that differences in perceived degradation arose from a combination of factors such as the cooccurrence of drought and grazing rather than stocking rate per se. A study by Lu et al. (2017) of many paired grazed and un-grazed grassland sites on the QTP suggested plant species richness declined in the absence of stocking whereas soil carbon, nitrogen, and microbial biomass increased.
We conducted a sheep stocking experiment to evaluate the appropriateness of the "Retire Livestock and Restore Grassland" policy. The questions evaluated were as follows: (1) Are there any adverse consequences from exclusion of stocking during the warm season to produce "reserved" pasture for cold season stocking, (2) is continuous stocking in the warm and cold seasons an unsustainable practice, (3) is warm season only stocking (with housing of stock in the cold season) a sustainable practice, and (4) is a stocking rate for "optimal" profitability sustainable? Specifically, hypotheses and predictions for each question are (1) plant and soil attributes do not change when there is no stocking during the warm season; removal of large herbivores from natural grassland will not in the midterm change soil attributes because they are slow moving but loss of plant species by rank growth in the warm season lethally shading low growing species is likely, (2) continuous stocking is not sustainable; plant and soil attributes are unaffected by continuous stocking at low stocking rates but quality of sheep would deteriorate during the cold season, (3) warm season only stocking is not a sustainable practice; the cost of feed during the cold season housing of sheep is sufficiently low for the practice to be sustainable, and (4) the optimal stocking rate for profitability is not sustainable; plant and soil attributes are not adversely changed by the stocking rate at which stock weight gain per unit of grassland is at a maximum.

| Study site
The study site (latitude 33°42′21″N, longitude 102°07′02″E, elevation about 3,600 m a.s.l.) is located on the eastern side of the QTP in the Maqu County, Gannan Prefecture, Gansu Province, China ( Figure 1). Here the warm season is from June to September, and the cold season is from October to May. Mean annual temperature is 1.2°C, mean daily temperature is −8.9°C in January and 11.9°C in July. There are on average 270 frost days per year. Mean annual precipitation is 620 mm, falling mostly in July and August. Annual cloudfree solar radiation is about 2,580 h. Soil type is Alpine Meadow Soil, that is, primarily Mat-Cryic Cambisols (Chinese Soil Taxonomy Research Group 1995). Vegetation is alpine meadow (Ren et al., 2008)  Stocking each year began a month after the commencement of the warm season because it was not possible to obtain the young sheep any earlier from local herders and it took several weeks to prepare the animals for grazing. The stocking ended each year in December of the cold season (rather than continuing to May) because work and measurements in this extreme and remote environment were impossible to sustain for the researchers during the last 5 months of the cold seasons.
The terminology used in the paper is the internationally accepted standard terms for grazing lands (Allen et al., 2011).
There were eight sheep in each paddock at any time in the warm, cold, and warm + cold season stocking treatments (Table 1). For the warm and cold seasons, stocking rates of 8 and 16 sheep/ha were in paddock sizes of 1 and 0.5 ha, respectively. For the warm + cold season, the stocking rate of 4 sheep/ha was in a paddock size was 2 ha. The no stocking (Control) treatment was fenced areas of 25 m 2 in each paddock.
Stocking treatment paddocks were fenced in early spring of 2010. There were six replicates of the warm and cold season stocking treatments. Within each replicate, the warm season stocked paddocks were subdivided into three sub-paddocks and the cold season stocked paddocks were subdivided into two subpaddocks as shown in Figure 2. Sheep were moved between the sub-paddocks every 10 and 15 days in the warm and cold seasons, respectively. There were three replicates of the warm + cold season stocking treatment, and the paddocks were not sub-divided; the grassland was continuously stocked for 6 months of the year.
The experiment was conducted for 5 years and terminated at the end of 2014.

| Sheep management
Each year 150 castrated male Tibetan sheep, 5 to 7 months old were purchased in June from nearby herders. Of these, 120 were assigned to the study and the remaining were grazed outside the treatment paddocks and used to replace animals killed by wolves (Canis lupus) or disease. In December, the sheep were sold.
Initially, sheep were ear-tagged, vaccinated, drenched for parasite control with Albendazole (Hanzhong Tianyuan Pharmaceutical Co. Ltd, Shanxi, China) and weighed on two consecutive days. The 120 heaviest sheep were divided into 15 groups of eight sheep with each group having a similar average weight. The members of each group were labeled with specific rump markings. These markings enabled the herder to place the members of each group to assigned paddocks each day. For an "acclimatization" period of 1~2 weeks, the sheep grazed outside the treatment paddocks and had access to mixed-mineral block and fresh water.
After the "acclimatization" period, the sheep were distributed in their groups to designated paddocks where they had continuous access to mixed-mineral blocks. Each day in the late afternoon and early morning sheep were herded from the paddocks, given access to stream water and mixed-mineral blocks, then held overnight in designated compartments of the yard and protected from wolves and thieves. The herder slept in a tent next to the yard.

| Sheep weights
Each sheep was weighed at the end of each month on two consecutive days. The weight gain per sheep per season was calculated as the difference between the weights at the beginning and the end of the three monthly seasons. Weight gain per hectare was calculated from the number of sheep in each paddock times the average seasonal-weight gain.

| Shoot and root biomass of vegetation
Each month, a 0.25-m 2 quadrat was placed in the central region of each sub-paddock and the shoots of each quadrat were cut and the on-ground litter removed and bagged together. The soil in the quadrat was sampled in the center with a 10-cm diameter auger 40 cm long. Soil was removed in 10 cm layers, down to a depth of 40 cm and each layer was separately placed in 0.2 mm mesh bags.
Shoots of each species and the litter were separated, oven-dried at 65°C for 48 hr and weighed. The total shoot weight was the sum of individual species. Soil samples were air-dried for 1 month in a glasshouse, and roots were separated from them. The roots were washed free of soil, oven-dried at 115°C for 48 hr and weighed. In this study, the data from the soil and root samples taken in August of each year are presented.

| Shoot biomass of each species
From the shoot biomass of each species in each sample (quadrat), two indices were derived. We used biomass of each species because the plant density was too high for accurate counting of the number of individuals of each species.

Plant species richness (S):
This was obtained from a count of the number of plant species in the sample taken from each quadrat area.

Plant species evenness (E):
This was calculated from the biomass of each species in a quadrat using the formula proposed by Camargo (1993). E is an estimate of the distribution of abundance among species and as such is an important descriptor of a community of plants. A community in which each species is equally abundant has high evenness; a community in which the species differ widely in abundance has low evenness (Smith & Wilson, 1996). E is calculated independently of plant species richness and together with S is used to compute of plant species diversity. The Camargo evenness index is defined as follows: where E: Camargo evenness index; Pi: the proportion of species i in the sample; Pj: the proportion of species j in the sample; S: the total number of plant species in the quadrat.

| Soil carbon, nitrogen, and phosphorus
The soil samples were air-dried at room temperature in the laboratory for 30 days and sieved through a 2-mm sieve. Soil organic carbon (SOC) was measured by Walkey and Black method (Nelson & Sommers, 1982). Total Kjeldahl nitrogen (TN) and total phosphorus (TP) were analyzed using a FIAstar 5000 flow injection analyzer (Foss Tecator, Högnäs, Sweden).

| Sheep weight gain and deaths
The

| Plant species richness
When the grassland was not stocked ( These were minor components of the grassland and were of low to medium palatability except for the grass Stipa aliena which was of high palatability.

| Plant species evenness
The evenness index significantly declined (p = .03 to .0001) in all treatments (Figure 6a-c) over the 5 years except for the cold season stocking treatment at 8 sheep/ha (Figure 6b). The slopes were also similar indicating the loss of evenness was at a constant rate.

| Root biomass
In the nonstock treatment (controls) for each of the three treatments (Figure 7a-

| Soil organic carbon, total nitrogen, and total phosphorus
There were no changes in SOC content and TN content of the surface 10 cm of soil in any treatments (Table 2). There was no change in TP content of the surface 10 cm of soil in warm and warm + cold season stocking treatments (

| Adverse consequences of exclusion of stocking during the warm season
The removal of stock from this grassland during the warm season had adverse consequences for plant conservation. Plant species richness steadily declined (Figure 5a), although the soil attributes measured did not ( Table 2) The steady loss of plant species richness by resting grasslands during the warm season does not support the government policy of "Retire Livestock and Restore Grassland." Survey of stocked and nonstocked grassland in the Maqu County also indicated lower plant species richness in nonstocked compared with stocked grassland (Wu, Du, Liu, & Thirgood, 2009). Confirmation of plant species loss by removing stock during the warm season has implications for both restoration and stock production. To restore degraded grassland, stocking appears to be required but the rate of stocking which fosters restoration needs to be established by a long-term stocking rate study involving a range of rates from low to very high rates on nearby dysfunctional and functional landscapes. To maintain stock production, the stocking system should include stocking during the warm seasons. Not to do so would predispose loss of animal productivity as animal production declines with lowering of plant species richness in the diet of sheep Wang, Zhao, et al., 2010).

| Sustainability of continuous stocking
The cessation of stocking of sheep or yak on the QTP by exclusion fencing has been linked to increases in soil carbon, phosphorus, and nitrogen (Dong et al., 2012;Wu, Liu, Zhang, Chen, & Hu, 2010) but TA B L E 2 Organic carbon, total nitrogen, and total phosphorus percentages in the surface 10 cm of soil in warm, cold, and warm + cold season grazing treatments measured in 2010 and 2014. Stocking rates were 0, 8, and 16 sheep/ha (warm, cold season grazing) and 0 and 4 sheep/ha (warm + cold season grazing). Each percentage is the mean (±SE) of six samples. Significance of differences between each percentage value in 2010 and 2014 is * = significant at 0.05 level, ** = significant at 0.01 level. Where significance is not signified there is no significance during the warm season to produce "reserved" pasture for cold season stocking in our study, the steady loss of plant species remains a concern.
The common practice of continuous stocking in both warm and cold seasons is considered to be unsustainable by some researchers (e.g., Miao et al., 2015). However, in our continuous stocking

| Sustainability of warm season only stocking and cold season housing
An alternative to continuous stocking at a modest stocking rate is to only stock the grassland during the 4 months of the warm season and then house and hand feed stock for the next 8 months. Our data for warm season only stocking indicate that up to 16 sheep/ ha can be sustainably stocked. This is almost certainly at the peak of animal productivity/ha and profitability for the grassland at the study site (Sun et al., 2015). In a comparison of the economics of grazing versus hand feeding of housed animals  in the northern edge of the QTP, hand-fed and housed sheep in the The management of grazing and supplementary feeding during the cold season is a critical part of the sheep production systems on the QTP. Sheep cannot survive the intense cold and chilling winds at night on the grasslands, and wolf predation is a constant threat unless they are in a yard and watched. Nighttime sheltering of sheep in the cold season is essential for sustainable production.
In this study, the sheep were returned to open traditional sheep yards at night, but not fed, after daytime grazing. These sheep did not gain weight and sometimes lost weight during the cold season ( Figure 3b), and stocking rate did not affect their weight change.
Nighttime housing in enclosed sheds would raise weight gains and reduce deaths. Supplementary feeding during nighttime would further raise weight gains but the economic benefits need to be assessed against the cost of the feed (Xu et al., 2017;Yang et al., 2011). In a study by Yang et al. (2011), cold season feeding options were modeled; stocking sheep on grassland without supplementary feed at night was found to be the most profitable option for the cold season.

| Sustainability of stocking for "optimal" profitability
The stocking rate in the warm season will determine both the productivity of the sheep and the profitability, as supplementary feeding and other management costs will be fixed and linearly related to the number of sheep in the pastoral business. If it is assumed that individual sheep production declines linearly with increase in stocking rate, that housing for sheep is available and that feeding and management costs are fixed, then the stocking rate during the warm season in this environment should be about 16 sheep/ha to maintain the grassland resource and to maximize the profitability from the stocked land (see Sun et al., 2015). Raising the sheep stocking rate in the warm season beyond 16 sheep/ha would increase the probability of the pastoral business crossing a critical threshold beyond which economic recovery and restoration of the grassland becomes problematic (Ash & McIvor, 2005).

| Implications for Tibetan herders
The stocking experiment and its results pertain to the settlement model for herders. This model has the advantage of creating a large enough community to attract education facilities, small businesses, and other services. There are, however, other successful and possible models (Shang et al., 2014) such as the pure nomadic model and semi-settlement models in between. Resolving the conflict between forage and livestock production in the context of sustainability, socio-economic systems, off-farm employments, markets, etc. is extremely complex. Shang et al. (2014) identified and reviewed 18 strategies currently practiced on the QTP to achieve sustainable livelihoods for its pastoral people. No single model will suffice and national support by policy and investment, local and regional commitment to capacity building, and the involvement of herders need to be continued and strengthened. Sadanandan Nambiar, CSIRO, commented on and edited a draft manuscript and comments of two anonymous referees further improved the manuscript.

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

AUTH O R CO NTR I B UTI O N
Fujiang Hou designed and supervised the project. Yingxin Wang, Zhaofeng Wang, and Shenghua Chang conducted the field work and collected the data. Yingxin Wang, Ken Hodgkinson, and Fujiang Hou wrote the manuscript with critical input from all the authors.