Introduction
The “entangled bank” metaphor of Darwin has inspired generations of community ecologists to explore the rules governing species coexistence and co-occurrence (Chesson, 2000; Gause, 1934; MacArthur, 1958; Ricklefs & Schluter, 1993; Tilman, 1982). Studies on species co-occurrence in species-rich communities over the past decades have reinforced the importance of the relationship between trait-mediated species differences and spatial distribution patterns among species in understanding the processes underlying coexistence and patterns of biodiversity (Chesson, 2000, 2013; He & Biswas, 2019; HilleRisLambers, Adler, Harpole, Levine, & Mayfield, 2012; Kraft, Godoy, & Levine, 2015; Laughlin, 2014; Li et al., 2018).
Species differences quantified by trait dissimilarity are frequently used as a proxy for the niche differences among species that are believed to drive species co-occurrence by influencing species’ response to environmental conditions and neighborhood interaction (Burns & Strauss, 2011; Cadotte, Carboni, Si, & Tatsumi, 2019; Cavender-Bares, Kozak, Fine, & Kembel, 2009; Kraft & Ackerly, 2010). With such approaches, the environment is often assumed to act as a filter that selects for species possessing similar traits, leading to aggregated interspecific spatial associations between species with similar traits, while the pairwise spatial repulsion between species with similar traits are thought to result from limiting similarity via competition (Cavender-Bares & Wilczek, 2003; He & Biswas, 2019). However, the assumed link between species differences and co-occurrence only holds when the measured traits dissimilarity actually reflect niche differences and influence neighborhood competition (Cadotte, Davies, & Peres-Neto, 2017). When these assumptions do not hold, for example, neighborhood competition is not driven by trait dissimilarity but by competitive advantage associated with particular trait values (i.e. trait hierarchy) (Kunstler et al., 2016, 2012), the pattern that species with similar functional traits co-occur together could also be the result of competitive exclusion of inferior competitors (hereafter hierarchical competition) in the absence of niche segregation when trait similar species have similarly high fitness -e.g. trees with resource conservative traits (Chesson, 2000; Lasky, Uriarte, Boukili, & Chazdon, 2014; Mayfield & Levine, 2010). Therefore, the relationship between interspecific spatial associations and species differences characterized by trait dissimilarity and trait hierarchy is key for disentangling the relative importance of multiple assembly mechanisms, especially those leading to identical co-occurrence patterns, e.g. environmental filtering and hierarchical competition .
Trait dissimilarity and trait hierarchy can be characterized respectively as absolute and hierarchical (i.e. directional) interspecific trait differences to represent species niche and fitness differences (Kunstler et al., 2012). Bivariate spatial point pattern analysis is a primary tool for estimating the degree of pairwise species co-occurrence patterns departing from independence (Fig. 1a and Fig. 1b), and understanding the underlying processes that cause the departure from independence (He & Duncan, 2000; Wiegand et al., 2007; Wiegand & Moloney, 2014). Associations with trait dissimilarity and trait hierarchy provide the bivariate analysis with a basis for detecting the relative importance of multiple assembly processes (Carmona, de Bello, Azcárate, Mason, & Peco, 2019; Kunstler et al., 2016, 2012; Lasky et al., 2014; Shen, Wiegand, Mi, & He, 2013; Wiegand et al., 2007, 2017). It is notable that the relative importance of different assembly mechanisms and their signatures on spatial associations is highly scale-dependent (Gianuca et al., 2016; Smith, Sandel, Kraft, & Carey, 2013).
In this study, we predict that limiting similarity should result in functionally similar species occupying segregated areas (i.e. functionally similar-spatial repulsive pattern), leading to a positive relationship between the absolute functional trait distance (trait dissimilarity) and pairwise spatial associations (Fig. 1c). While, at the same time, functionally similar species would co-occur together more likely (i.e. forming functionally similar-spatial aggregated pattern) resulting from either environmental filtering or hierarchical competition (HilleRisLambers et al., 2012; Mayfield & Levine, 2010), leading to a negative relationship between the absolute functional trait distance and spatial associations Fig. 1d). To disentangle which of these two processes in the case of Fig. 1d , environmental filtering and hierarchical competition, is responsible for the functionally similar-spatial aggregated pattern, it is necessary to simultaneously test and compare the relative strengths of trait dissimilarity and trait hierarchy on pairwise spatial associations. If environmental filtering prevails, we would expect that the strength of trait dissimilarity should be greater than that of trait hierarchy (Fig. 1e), and if hierarchical competition drives the community assemblage, the effects of trait hierarchy are expected to be stronger than that of trait dissimilarity (Fig. 1f).
To link forest assembly mechanisms to trait difference- spatial association relationships and test the three hypotheses above (Fig. 1c, d, e and f), we addressed the following questions about spatial associations: (1) How are pairwise spatial associations correlated with trait dissimilarity and trait hierarchy? (2) Do the spatial pattern-trait difference relationships remain consistent across different spatial scales? Moreover, little attention has been paid to variation in abundances among the focal tree species and relative importance of different assembly mechanisms. In this study, we explored how the magnitude that the hierarchical competition effects outcompete the environmental filtering effects was related to the abundance of each focal species when trait hierarchy effects were detected.
To address these questions, we analyzed the bivariate spatial associations of 80 common species in a fully mapped 50 ha (1000× 500 m) plot in Heishiding nature reserve in southern China using spatial point pattern analysis. To reveal how trait dissimilarity and hierarchy determine species co-occurrence patterns in the study forest, we evaluated the support for the three hypotheses by assessing the spatial pattern-trait difference relationships across different spatial scales. We then related the relative strengths of trait hierarchy and trait dissimilarity on spatial associations to the abundance of each focal species to better understand the underlying mechanisms.