Lineage-specific phylogenetic structure of boreal habitats suggests different assembly processes across phylogenetic and spatial scales

The phylogenetic distance among species in a community (community phylogenetic structure) has been used to infer deterministic and stochastic assembly processes, albeit with criticisms. The effect of phylogenetic scale (old versus young lineages) and spatial scale on measures of community phylogenetic structure are rarely tested simultaneously especially in the boreal biome, yet are essential to unravel different assembly processes that might operate in a community. We examined lineage-specific phylogenetic structure for six vascular plant communities defined at the habitat scale (arctic-alpine barren, bog, fen, kalmia barren, limestone barren, and serpentine barren) on the island of Newfoundland, Canada, and the phylogenetic structure of plant communities defined at a plot scale (72 plots x 1m²). Contrary to the expectation under the stress-dominance hypothesis of phylogenetic clustering in challenging boreal environments, the majority of clades across the six boreal habitats had random phylogenetic structure. However, we observed a shift from phylogenetic clustering at the deepest nodes of the angiosperms to no phylogenetic structure at shallower nodes (<110 mya), suggesting changes in assembly processes with phylogenetic scale within a habitat, and the potential role for deterministic processes at deep nodes. The random phylogenetic structure of 1m² plots and our modeling effort to test the effect of an environmental stress gradient on community composition suggest that a complex set of stochastic and deterministic factors is responsible for species assembly at this fine spatial scale, not just abiotic filtering in hostile environments like the serpentine as predicted by the stress-dominance hypothesis. The interpretation of phylogenetic structure metrics did not change when considering species abundances or when polytomies were resolved. Taken together, inference of assembly processes must be lineage-, habitat- and spatial scale-specific, supplemented with knowledge on trait role and evolution for which we outline future research hypotheses.