1 Introduction
Herbivore grazing is an ecological and economic use of grasslands worldwide and is a primary disturbance affecting the plant community (Asner et al., 2004). Although maintenance of a desirable mixture of shrubs and herbaceous plants is beneficial to the functions and services of the grassland ecosystem (Barbosa da Silva et al., 2016), grazing management can intentionally or unintentionally change the abundance and distribution of shrubs and herbaceous plants in the community (Madany et al., 1983; Brown et al., 1999). Many studies have found that long-term grazing can reduce grass cover, and aboveground and belowground biomass, which not only increases resource availability for the recruitment of shrubs and facilitates shrub regeneration, but also reduces surface combustibles, which may reduce fire frequency (Archer, 1994; Lin et al., 2010; Lohmann et al., 2014). These effects change grassland vegetation structure from herbaceous- to shrub-dominated communities and can accelerate the process of shrub encroachment (Case and Staver, 2017; Devine et al., 2017). For example, studies in savanna grassland suggest that overgrazing reduces grass cover and shifts dominant herbaceous plants from perennial to annual species, which have been linked with reductions in competition between shrubs and grasses, ultimately facilitating shrub development (Goheen et al., 2004; Belsky and Blumenthal, 2010; Graz, 2010; February et al., 2013). Studies in the western United States have shown that increasing density of grassland shrubs is related to the introduction of cattle grazing, because livestock can be used as a carrier for shrub seed diffusion while feeding on grassland plants (Painter et al., 1993; Tews et al., 2004). In contrast, other findings in semiarid north Chinese steppes have shown that grazing is not a direct cause of encroachment (Zhang et al., 2014) and some studies have even shown that grazing delays or limits shrub encroachment in a subhumid south American grassland (Altesor et al., 2006). In the latter case, grazing by domestic animals kept the shrubs at a smaller size and a low or static density, while shrub size and density increased with removal of the grazers (Friedel, 1985; Van Auken, 2009). Similarly, grazing exclusion in the temperate steppe in northern China has led to Caragana micropylla invasion, because of preferential utilization of palatable shrubs by grazers, such that after grazing exclusion vegetation species composition shifts from more to less palatable species due to the release of herbivore feeding pressure on grasses (Zhang et al., 2014). It has also been suggested that livestock consumption of seedlings and young branches of shrubs delays shrub development, suppressing their ability to compete with the dominant herbaceous plants (de Dios et al., 2014), and even causing shrub frequency to decrease or disappear.
There are often different shrub species on the grassland that form the ‘background plant’ of the plant community, giving the vegetation structure a clear mosaic pattern (Wei et al., 2013). Although there are many shrub species in grasslands, only a few species have actually become aggressive encroachers in regional flora (Stockes et al., 2010; Eldridge et al., 2011). For example, in a Sonoran Desert grassland, the coverage of Prosopis velutina increased markedly from the 1950s to the 1990s, whereas other shrubs have maintained low coverage (Brikes, 2017). Therefore, shrub proliferation depends not only on external driving factors, but also on the characteristics of the shrub itself (Van Auken, 2009). From the perspective of community theory, population distribution and species richness are the result of a regional species pool passing through a series of abiotic environment and biotic community processes, such that species with unsuitable environmental traits are excluded from the community (Keddy, 1992). Invasive shrubs must have a strong response to at least one encroachment driver, but must also overcome all other environmental factors in order to survive. Non-invasive shrubs may not respond to this driving factor, or may be limited by other environmental factors.
In summary, the direction of succession in grassland plant communities under grazing disturbance is the result of the interaction of grazing livestock and shrub characteristics. When shrubs are consumed selectively by grazing livestock, shrubs in the plant community will be at a disadvantage, and when grazing animals mainly feed on herbaceous plants, the herbaceous plants will be at a disadvantage (Van Auken and Smeins, 2008). Therefore, the succession of grassland vegetation should be comprehensively judged according to the regional climate, dominant plant species and grazing animals.
The Stipa breviflora desert steppe is located in the transition zone of the Eurasian steppe from the typical steppe zone to the desert zone. The vegetation in this transition zone is shaped by the influence of a harsh climatic environment and scarce soil resources, and the ecosystem is sensitive to interference. More than 10 years of monitoring has shown that there are more than 60 plant species in the community, including shrubs and semi-shrubs such as Artemisia frigida ,Kochia prostrata , Caragana microphylla andCeratoides latens . Long-term research was carried out in this region to answer the following three questions. First, does shrub encroachment in Stipa breviflora desert steppe increase with the introduction of sheep grazing? Second, how do shrub species respond to grazing under different stocking rates? And third, can biotic or abiotic factors predict the direction of succession in the desert steppe under different stocking rates? Elucidating these questions can not only deepen our understanding of shrub-herbaceous plant interactions and succession patterns in desert steppe vegetation, but also provide a theoretical basis and experimental evidence to guide grazing management in desert grasslands to improve the ecological environment.