Microbial succession in West African secondary forests: rapid internal stabilisation without convergence toward old-growth reference states

This repository contains all data, code, and raw sequencing reads associated with the manuscript "Microbial succession in West African secondary forests: rapid internal stabilisation without convergence toward old-growth reference states". Study overview This study investigates the successional dynamics of soil bacterial and arbuscular mycorrhizal fungal (AMF) communities along post-agricultural chronosequences spanning six classified forests across three phytogeographic zones in Côte d'Ivoire (dry, semi-deciduous, and evergreen). Thirty plots were sampled across early (1-10 yr), intermediate (11-20 yr, 21-30 yr), and late (>30 yr) secondary forests, with old-growth forest reference plots at each site. Bacterial and AMF communities were characterised by high-throughput amplicon sequencing of the 16S rRNA V3-V4 region and the 18S rRNA SSU gene respectively. Alpha diversity, beta diversity, compositional turnover, distance to old-growth reference states, and indicator taxa were evaluated within a Bayesian hierarchical framework using generalised additive models and mixed-effects beta regression models. Contents code.zip: All R scripts required to reproduce the analyses and figures reported in the manuscript, including Bayesian hierarchical GAMs for alpha diversity (richness, Shannon, Simpson), NMDS ordination, PERMANOVA, variance partitioning, pairwise compositional turnover models, beta-dispersion GAMs, and distance-to-old-growth-forest models. A README file describing the repository structure, dependencies, and usage instructions is included. README.md: Standalone README describing data formats, script structure, software dependencies (R ≥ 4.3, brms ≥ 2.20, vegan, tidybayes, ggplot2, and related packages), and instructions to reproduce the full analysis pipeline. reads.zip (16S_*.fastq.gz): Raw paired-end Illumina MiSeq sequencing reads (2 × 300 bp) for bacterial 16S rRNA V3–V4 amplicons, generated from 30 soil samples collected across six classified forests in Côte d'Ivoire. Primers: 344F (5'-ACGGRAGGCAGCAG-3') / 802R (5'-TACCAGGGTATCTAATCCT-3'). Files are named with the prefix 16S_ followed by the sample identifier. reads.zip (18S_*.fastq.gz): Raw paired-end Illumina MiSeq sequencing reads (2 × 300 bp) for AM fungal 18S rRNA SSU amplicons from the same 30 soil samples. Primers: WANDA (5'-CAGCCGCGGTAATTCCAGCT-3') / AML2 (5'-GAACCCAAACACTTTGGTTTCC-3'). Taxonomic assignment was performed against the MaarjAM Virtual Taxa database (97% identity, 95% alignment length). Files are named with the prefix 18S_ followed by the sample identifier. Sequencing platform Sequencing was performed on an Illumina MiSeq platform (GeT sequencing platform, INRAE Toulouse, France) using a 2 × 300 bp paired-end approach. Soil sampling was conducted from September 2022 to December 2023 across the Badénou, Foumbou, Haut-Sassandra, Téné, Niegré, and Irobo classified forests in Côte d'Ivoire. Reproducibility All Bayesian models were fitted using the brms package (v2.23.0) interfacing with Stan. Convergence was verified using the potential scale reduction factor (R̂ < 1.01) and effective sample size diagnostics (bulk and tail ESS > 400) for all parameters. Full details of the statistical framework, prior specifications, and marginalisation procedures are provided in the supplementary materials of the associated manuscript.