Discussion
Trait dissimilarity effects were widely considered to explain species co-occurrence and coexistence over the past decade (Burns & Strauss, 2011; He & Biswas, 2019; Kraft & Ackerly, 2010). Consistent to the findings in He & Biswas (2019), we observed negative relationships between trait dissimilarity and pairwise spatial associations in this study across different summary statistics and spatial scales for individual and multiple functional traits (Fig. 2), except LDMC that showed non-significant effects on pairwise spatial associations. However, while necessary, this evidence is not sufficient to support that environmental filtering is the dominant mechanism. Instead of simply interpreting this negative relationship as a result of environmental filtering and absence of competition (He and Biswas 2019), we pointed out that hierarchical competition that selects species with traits conferring competitive advantages (i.e. hierarchical competition) could also be able to produce a negative relationship between trait dissimilarity and pairwise spatial associations, which is identical to the result of environmental filtering (Fig. 1d).
By comparing the relative strengths of trait dissimilarity and hierarchy on pairwise spatial associations, we demonstrated that competitive hierarchies captured by the traits of LA, WD and Hmaxbetter explained species co-occurrence than trait dissimilarity (Fig. 3 and Fig. S2-6), which supports the hypothesis that hierarchical competition contributes to the co-occurrence patterns. As for the trait of SLA and WDMC, we observed that the effects of trait dissimilarity and hierarchy were comparable on the pairwise spatial associations.
Taking together the effects of both trait dissimilarity and trait hierarchy on pairwise spatial associations, we infer that except LDMC, other traits showed either stronger (e.g. LA, WD and Hmax) or comparable (SLA and WDMC) trait hierarchy effects relative to trait dissimilarity effects on interspecific spatial associations, indicating that the effects of hierarchical competition on the co-occurrence patterns in our forest plot were greater than or comparable to the effects of environmental filtering. Other traits may show stronger effects of environmental filtering or limiting similarity in structuring the forest community, but we currently lack the trait data to capture such effects.
This study provides a novel method to disentangle the relative importance of multiple assembly mechanisms in structuring co-occurrence patterns by assessing the effects of trait hierarchy and trait dissimilarity on pairwise spatial associations. By linking the pairwise spatial associations, which reflect signatures left by different assembly mechanisms to the effects of trait dissimilarity and trait hierarchy, our study provides alternative perspectives and better understanding in the underlying mechanisms that govern the co-occurrence pattern (He & Duncan, 2000; Wiegand et al., 2007; Wiegand & Moloney, 2014).
The negative relationship between trait dissimilarity and pairwise spatial associations was typically interpreted as evidence for no signal of competition and inferred as a result of environmental filtering (He & Biswas 2019). However, this interpretation could be misleading because the negative relationship between trait dissimilarity and pairwise spatial associations could also be caused by neighborhood competition that selects species with particular trait values independent of environmental filtering (Carmona et al., 2019; HilleRisLambers et al., 2012; Mayfield & Levine, 2010). In this study, we found support for the hypothesis that hierarchical competition leads to the negative relationship between trait dissimilarity and pairwise spatial associations as well.
This study also strongly suggests that trait dissimilarity has little effects on neighborhood competition and that neighborhood competition is more likely to be driven by trait hierarchy, which is consistent to the findings in Kunstler et al. (2012; 2016) and Carmona et al. (2019). If trait dissimilarity was positively correlated to pairwise spatial associations, we would infer that trait dissimilarity affects the neighborhood competition and leads species with similar trait occupying segregated areas. However, which is not the case in this study. Now that the positive relationship between trait dissimilarity and spatial associations was absent and effects of hierarchical competition that exclude inferior competitors were found, we therefore speculate that neighborhood competition in our forest plot was more likely to be driven by trait hierarchy but not by trait dissimilarity as presumed (Carmona et al., 2019; Kunstler et al., 2012).
The two metrics summarizing spatial point patterns (gij (r ) andDij (r )) that we used in this study showed no significant differences in the effects of trait dissimilarity and trait hierarchy on spatial associations for each trait (Fig. 2 and Fig. 3). We therefore speculate that extreme heterogeneity of species distributions were not prevalent in our forest plot (Wiegand et al., 2007). Since these two summary statistics respectively characterize the mean number of individuals and the nearest neighbors of the second species around the focal species, the findings that these two summary statistics of spatial point patterns reveal similar trait effects suggest that the neighborhood interspecific competitive effects on the focal trees come from both the average neighbor density and the nearest neighbors of the other species at least within the scale of 50 m.
By linking the magnitude that trait hierarchy effects outcompete trait dissimilarity effects (MHD) on pairwise spatial associations to the abundance of focal species, we found inconsistency among the three traits LA, WD and Hmax that all showed trait hierarchy effects on population fitness (Fig. 4). This inconsistency suggests that different fitness components, e.g., growth, survival and reproduction of each species, might be unequally influenced by traits and a trait might positively affect one component but negatively affects another, e.g., growth-survival trade-off and reproduction-survival trade-off (Laughlin, Gremer, Adler, Mitchell, & Moore, 2020). These unequal effects on different components of fitness and trade-offs might therefore influence the abundance of the focal species and lead to variations in the relationships between MHD and focal species abundance for different traits.
In conclusion, we disentangled the assembly mechanisms of limiting similarity, environmental filtering and hierarchical completion in structuring our forest community by assessing and comparing the effects of trait dissimilarity and trait hierarchy on pairwise spatial associations in this study. More specifically, we found that limiting similarity was weak or absent and hierarchical competition played a more (or at least equally) important role than environmental filtering in structuring the co-occurrence patterns in our forest community. This study also reinforced the importance of trait hierarchy, rather than trait dissimilarity, in driving interspecific competition.