3.3 Trait phylogenetic conservatism
All K values of investigated traits were less than 1, ranging from 0.15 to 0.53 (Table 1), which indicated that all traits exhibited low to intermediate levels of conservatism. In addition, the p values of five traits (VH, LNC, LCC, LDMC, LT) were less than 0.05.
4 DISCUSSION
Community phylogenetic approach reveals the relationship of evolutionary processes, regional environments, and interactions between species, which is important to understand the mechanisms of species co-occurrence and community assembly (Cavender-Bares et al., 2009). The present findings demonstrated that the restoration processes of shrun-encroached grassland were affected by fencing duration and shrub cover..
The increase of fencing duration increased the species richness and phylogenetic diversity, suggesting that longer fencing duration was conducive to the restoration of community diversity in shrub-encroached grasslands (Yoshihara et al., 2010; Soliveres & Eldridge, 2014; Koyama et al., 2015). Generally, species richness usually increases first and then decreases in secondary succession of grazing exclusion (Chen et al., 2014). In this study, species richness did not decline even in the communities that had been fenced since 1979. The result of NMDS confirmed the differences in species composition between fencing durations. Meanwhile, there were only 19 species in species pool in this study area, far lower than the number 27-58 in the typical grassland and degraded grassland of this region (Bai et al., 2000). These results indicated that the community was still in recovering stage, but the restoration rate was slow. We also found that the communities with higher fencing duration were more stable. However, the restoration processes of shrub-encroached grassland were inconsistent with non-shrub-encroached grassland which could return to the climax state after about 11-year fencing (Li et al., 1993). Previous studies have suggested that C. microphylla had higher water capturing capability than grasses, it will lead to irreversible stage once C. microphylla dominated the landscape (Peng et al., 2013), mainly due to that there was a threshold point for transitions between states (Xiong et al., 2004; D’Odorico et al., 2012). In addition, the existence of shrub canopy would ameliorate abiotic environment for the growth of neighbor spceis, enhancing plant richness and influencing community composition (Maestre et al., 2009; Koyama et al., 2015). This also explains why the species richness and community composition of high shrub cover communities were significantly different from that of low and medium shrub cover community.
Phylogenetic diversity reflects evolutionary history of coexisting species and clarifies their capacity to adapt to environmental changes (Alberti et al. 2017). In addition, it can represent independent aspects of community structure (Barber et al., 2019; Jones et al., 2019). Previous studies found that phylogenetic diversity was positively correlated with species richness but differed in their temporal or successional patterns (Purschke et al., 2013). In the present study, the phylogenetic diversity positively correlated with species richness overall. However, phylogenetic diversity didn’t increased obviously with species richness in different succession gradient of shrub cover. This result indicated that most plant lineages present in the high shrub cover communities also occurred in the low and medium shrub cover and that they are generally less speciose in the last two shrub cover communities (Blaimer et al., 2015; Bares et al., 2019). This also reflected that the impact of shrub cover on species diversity was not consistent with fencing duration.
In this study, we found that the majority of traits were phylogenetically convergent, with six out of seven phylogenetic signals were lower than 0.40. Pipenbaher et al. (2013) found traits were convergent for most of traits when they tested phylogenetic signal in the North Adriatic Karst grassland. However, most studies in tropical and temperate forests demonstraed that traits were phylogenetically conservatism (Muscarella et al., 2015; Döbert et al., 2017). This difference may occur, because compared with woody plants, herbaceous plants tend to have shorter life cycles, more rapid reproduction and more adaptable to the environmental changes (Donoghue, 2008). The evolutionary relatedness of ecological traits have an important influence on the interpretation of community assembly (Liu et al., 2018). Cavender-Bares et al. (2004) emphasized the importance of detecting phylogenetic conservatism, and suggested that the results of testing phylogenetic conservatism should be combined with the observed phylogenetic dispersion pattern in order to make a strong inference on the ecological and functional similarity of coexisting species.
The phylogenetic pattern varied widely among different fencing duration and shrub cover communities. The low shrub cover communities under fencing since 1979 and high shrub cover communities under fencing since 1983 and 1979 showed evidence of phylogenetic clustering (SES < -1.96), with co-occurring species significantly more closely related than would be expected by chance. In contrast, phylogenetic random dispersion (-1.96 < SES < 1.96) dominated in low shrub cover communities under fencing since 2003 and 1983, medium shrub cover under fencing since 2003, 1983 and 1979 and high shrub cover communities under fencing since 2003 (based on MPD). All of communities showed a random assembly regarding MNTD, but it showed a trend of phylogenetic clustering with fencing duration or shrub cover increasing. When ecological traits evolve convergently, the interpretation of the community assembly becomes complicated, and its predictive ability declines (Kraft et al. 2007). Convergent traits can result in clustering patterns even though competitive exclusion is the dominant process of community assembly (Webb et al. 2002). With regard to the present results, clustering in low shrub cover communities under fencing since 1979 and high shrub cover communities under fencing since 1983 and 1979 could be linked to competitive exclusion rather than habitat filtering. The combination of phylogenetic clustering and convergent trait evolution seen in our study has also been found in other studies (Pipenbaher et al., 2013; Purschke et al., 2013; Liu et al., 2018). Research results from non-shrub-encroached grassland indicate that competitive exclusion would play more important role in driving the community assembly with the fencing duration due to the improvement of abiotic conditions (Wellstein et al., 2014; Conradi et al., 2017; Barber et al., 2019). Our results indicated that the communities assembly during the fencing restoration in shrub-encroached grassland was consistent with that of non-shrub-encroached grassland. In addition, we aslo found shrub cover significantly affected the community assembly process. On one hand, previous studies have shown that higher shrub cover could increase soil fertility (Maestre et al., 2009; Pipenbaher et al., 2013) and ameliorate abiotic environment(Molina-Montenegro et al., 2016; Thompson et al., 2017). The decrease in abiotic stress and the increase in species richness may intensify resource competition among herbaceous species (Chapin et al.,1994; Purschke et al., 2013). On the other hand, there may be direct competition between shrubs and herbaceous species, and higher shrubs have a higher competitive advantage in light and water resources than herbaceous species (Peng et al., 2013; Pierce et al., 2019), accelerating competitive ability for resources. Thus, niche overlap in open space should become significant and competition should consequently intensify.
More recently, the importance of random colonization and stochasticity in early community has become evident (Ulrich et al. 2016; Marteinsdóttir et al., 2018; Liu et al., 2018). Our results found a random assembly process (-1.96 < SES < 1.96) in six out of nine communities (Figure 5). Generally, random community assembly is considered to be strongest in communities that are dominated by colonization. In arid and semi-arid communities, colonization was often limited by unfavorable local environmental conditions (Conradi et al., 2017; Marteinsdóttir et al. 2018), seed availability (Knappová et al., 2017) or lack of facilitating species (Callaway, 1995). At sites like Inner Mongolia steppe with low soil nutrient status, limited, seed rain and low seedling survival (Koyama et al., 2015), a successful colonization requires a sequence of favorable events. This also proved that these communities have lower stability than those with longer fencing duration and higher shrub cover (Figure S2; Table 1).