Introduction
How and why biodiversity varies along environmental gradient is one of the fundamental issues in ecology and biogeography (Patino et al.2017; Brown 2014; MacArthur & Wilson 2001). Proper answering this question is meaningful and essential, not only for understanding biodiversity per se , but also for promoting sustainable biodiversity conservation and relevant government-lead policy making. The observed pattern of biodiversity and underlying mechanism have been constantly debated for centuries (Hutchinson 1959; Darwin 1859). Earlier ecologists have paid more attention to the variation of taxonomic composition (i.e., species richness, evenness and variety) (Allenet al. 2002; Whittaker et al. 2001) and proposed many empirical explanatory hypotheses related to contemporary and/or historical environmental variance (i.e., habitat heterogeneity, energy, water-energy balance, area and geometric constrain) (McCain 2007; Evanset al. 2005; Boyce et al. 2003; Hawkins et al.2003; Colwell & Lees 2000). Definitely, these traditional taxonomic approaches have provided a comprehensive understanding on community assembly. However, subsequent ecologists have realized that taxonomic measurements alone are insufficient to identify mechanistic processes without regard to interspecific ecological and evolutionary information (e.g., differentiations in ecological niche and evolution) (Guittaret al. 2019; Blonder 2018; Martiny et al. 2015; Liuet al. 2013; Cavender-Bares et al. 2009; Swenson & Enquist 2009; Wiens & Graham 2005).
    Recently raised phylogenetic and trait-based approaches provide a more reasonable perspective in mechanistic understanding of species assembly (Swenson 2013). Based on classical niche theory (Hutchinson 1959) and phylogenetic niche conservatism hypothesis (PNC) (Wiens & Graham 2005; Pagel 1999), Webb (2000) offered an approach to calculate community structure based on phylogeny. As a follow-up, Webb et al. (2002) later raised a framework of phylogenetic community structure with which species assembly process could be roughly estimated by measuring phylogenetic structure and evaluating trait conservatism. In the last decade, more and more studies have applied the information of phylogeny and functional trait into understanding community assembly process along environmental gradient (Chun & Lee 2018; Feng et al. 2014; Cianciaruso et al. 2012) at various spatial and temporal scales (Maherali & Klironomos 2007; Cavender-Bares et al. 2006; Swensonet al. 2006). Meanwhile, the framework of community construction has been constantly modified (Kraft & Ackerly 2010; Kraft et al.2007; Cavender-Bares et al. 2004). The effect of stochastic process, environmental complexity and the effect of negative density-dependence have been taken into account in later frameworks, which enabled it more practical in interpreting species assembly in real and complex ecosystems (Kraft & Ackerly 2010; Kraft et al.2007). By examining trait similarity on phylogeny and correlation between phylogenetic and trait dispersion, Cavender-Bares et al.(2004) and Losos et al. (2003) have reported that long history of competitive interactions could produce trait convergence on phylogeny. This means that PNC can interpret trait evolution in some taxa, but does not so in any single case (Losos 2008). In other words, ecologists should carefully examine the magnitude of PNC rather than subjectively presuppose its existence in functional attributes (Losos 2008). This has enhanced the worldwide application of phylogenetic and trait-based approaches to study on the potential process driving community assembly.
    Although ecologists have made a mass of efforts to open the ‘Pandora’s Box’ of community construction, the mechanism underlying community assembly are still poorly understood. Under prior frameworks of community structure, if relevant functional attribute is phylogenetically conserved on phylogeny, phylogenetic dispersion is expected to be concordant with the trait dispersion of concern (Kraft & Ackerly 2010; Kraft et al. 2007; Cavender-Bares et al.2004; Webb et al. 2002). But discordant or opposing patterns between phylogenetic relatedness and phylogenetically conserved trait were also reported in flora (Yang et al. 2014; Swenson & Enquist 2009) and fauna (Du et al. 2017). Generally, most of empirical studies believed that these contradictory patterns were resulting from the low magnitude of phylogenetic signal (Du et al. 2017; Yanget al. 2014). An alternative explanation points to the potential erroneous inferences provided by phylogenetic signal metrics (Swenson & Enquist 2009).
    Kraft and Ackerly (2010) have ever mentioned that some kinds of traits are more sensitive responding to ecological process than the others. However, it appears extremely challenge to quantify how much one functional attribute relates to a certain ecological function. This is because, on one hand, ecological process usually organizes community by acting on multiple phenotypes and their interactions (Miner et al. 2005; Norberg et al. 2001). Under a certain scenario, the functional roles of ecological attributes usually differs from each other and exhibits distinct dispersion. For instance, plant traits related to productivity should be functional clustering in local plant communities, whereas traits related to disturbance and regeneration are expected to be locally over-dispersed (Swenson & Enquist 2009; Grime 2006; Thompson et al. 1996). On the other hand, nonrandom patterns of functional structure along environmental gradient have implied the functional role of a certain phenotype likely varies across assemblages (Du et al. 2017). It means that the trait dominating species assembly within one habitat might act an assistant role in another habitat, vice versa . These evidences above mentioned have implied that disentangling distinct functional contribution of traits is essential for revealing the truth underlying community assembly.