Soil factors rather than stand age drive tree phylogenetic diversity along Amazon Forest succession

Understanding the phylogenetic diversity and structure of woody communities can explain how deterministic or stochastic processes drive a forest community assembly. This study assessed the tree community's phylogenetic diversity and structure during secondary succession after shifting cultivation (SC) in the Northern Amazon Forest. We hypothesized that soil-mediated filtering, rather than habitat filtering based on stand age and neutrality-based stochastic processes, determines phylogenetic diversity and structure of woody community along Amazon Forest succession after SC. We used forest inventory data from 45 plots across three sites representing second-growth forests (SG) having stand ages after shifting cultivation (e.g., 5, 10, 15, and 20 years old) and an old-growth reference forest (> 100 years old, OG). We tested different linear mixed-effects models to determine the main effects of soil-mediated filtering (i.e., nutrients and soil texture) and stand age on phylogenetic metrics. Phylogenetic diversity (PD) showed a significant difference between SG and OG: the highest PD was found in OG, whereas the lowest PD was found during the initial successional stage. We found a trend of phylogenetic structure promoted by soil attributes; the variability of soil texture mainly explained most of the variation of phylogenetic diversity and structure. Stand age did not demonstrate a significant influence on phylogenetic metrics across any of the tested models. Higher soil fertility may favor the growth of species from multiple distant clades, increasing phylogenetic diversity and reducing phylogenetic clustering. However, SC may affect the fertility content in silt soils of OG and generate soils with a high proportion of sand and low fertility in SG. Thus, our study demonstrates that soil-mediated abiotic filtering shapes the phylogenetic structure and diversity of tree communities along Amazon forest succession due to deterministic processes rather than stand age and neutrality-based stochastic processes.