Liana community assembly on fine spatial scale in a Neotropical seasonal dry forest

Background/Question/Methods

Like other growth-forms, the assembly of liana communities is believed to be submitted to the constraints imposed by dispersal, environment and community internal dynamics. Our aim was to investigate how much each of these constraints contributes to liana community assembly on a fine spatial scale in a tropical semideciduous forest. The internal dynamics were expressed through tree wood density, crown illumination index, median and maximum height, and deciduousness, which together with eco-unity type, were used to characterize each plot. The environmental variables we considered plot altitude and soil pH, texture and nutrients (topo-edaphic variables). Applying PCNM (principal coordinates of neighbor matrices) to the coordinates of each plot, we admitted that liana dispersal limitation was indicated by a positive spatial autocorrelation explained only by space. We calculated the variation partitioning among topo-edaphic, forest tree dynamics, and spatial variables (PCNM) to explain the spatial distribution of liana species per plot in a well-conserved forest fragment in SE Brazil. Finally, we performed a Redundancy Analysis (RDA) with all selected variables, checked which variables most contributed to the ordination axes, and assessed which species were best modeled by the first three ordination axes.

Results/Conclusions

We found a large amount (82.05%) of unexplained variation data. Among the isolated explaining variables, space (representing dispersal limitation) contributed 9.88%, internal dynamics 1.16%, and topo-edaphic variables 0.88%. The remaining 6.03% was due to interaction among these three sets of variables, the most being redundant with space. Applying RDA to the species with the greatest scores, we observed that some liana species occurred most frequently in shaded plots with tall trees, denoting variation of liana composition among eco-units, whereas some species were more associated with tree-fall gaps and others were more frequent in biostatic eco-units. Moreover, liana species distribution responded to the position on slope. Dispersal limitation was the main constraint assembling liana community on fine scale, whereas constraints associated with tree community internal dynamics and topo-edaphic variables played a reduced role. We concluded that dispersal limitation is the main constraint which controls the assembly of liana communities at fine scale in a seasonal semideciduous forest. Slope position and some forest variables (tree height and crown openness) also drive few differences in liana assembly.