Phylogenetic patterns of Atlantic forest restoration communities are mainly driven by stochastic, dispersal related factors

• We pioneered the use of phylogenetic ecology in Atlantic forests restoration.
• We did not find a clear trajectory of restoration sites resembling reference sites with age.
• Most sites showed clustering tendencies, significant in one restoration and one reference site.
• Differences in species composition suggest dispersal drives phylogenetic patterns.
• Also identity of planted species affects observed phylogenetic patterns.

Phylogenetic ecology complements trait-based analysis on community assembly by considering that species are not independent units but are related to each other by their evolutionary history. Phylogenetic patterns clustered when there are more close relatives than expected by chance or overdispersed with less close relatives than expected. Patterns among species in a community indicate underlying biotic and abiotic processes acting on species functional traits. However, phylogenetic ecology has seldom been applied to forest restoration. We used floristic and abundance data from six forest restoration sites of different ages and four old-growth reference forests in the Brazilian Atlantic forest to evaluate similarities in phylogenetic patterns between restoration and reference forests as a measure of restoration success. The presence of an initial tree canopy in restoration forests conducted by planting species increases seed dispersal. Nevertheless, we expected random phylogenetic patterns early in restoration due to dispersal limitation in a highly fragmented landscape. As time since planting increases and in reference forests, we expected less of an effect of dispersal on community composition and more of an effect of negative biotic interactions among close relatives to lead to overdispersed patterns. We did not find a clear trajectory showing that restoration sites would resemble the phylogenetic patterns of reference sites with age since planting. We found significant clustering patterns in two sites, the oldest restoration site and one reference forest. The other reference forests showed, non-significant yet clustering tendencies. The functional traits studied were less conserved than expected by chance, therefore, we cannot relate clustering to be solely the result of environmental filters leading to the presence of close relatives with similar habitat requirements. The presence of closely related species in the Meliaceae family in reference forests and in the oldest restoration site, which was next to a forest remnant, points toward dispersal as the main factor driving phylogenetic patterns in the sites studied. Despite the use of a high number of planted species, differences in the composition of planted species among sites also affected the observed phylogenetic structure. We believe that phylogenetic ecology complements floristic studies by providing information on trait conservatism and shedding light on community assembly processes that affect the successional trajectory of restoration forest.