The impact of Pleistocene climatic oscillations on the biodiversity of African tropical rain forests remains poorly understood, and the Congo Basin is particularly understudied. We aim to elucidate how Pleistocene climatic oscillations shaped lowland tropical rain forests by investigating the intraspecific diversity and evolutionary history of a widespread tree species, Staudtia kamerunensis Warb.

We sequenced 88 individuals of Staudtia kamerunensis and 1 of Staudtia pterocarpa using a genome skimming approach. We used maximum likelihood and Bayesian inference to infer the plastid phylogeny. We estimated the time of speciation and differentiation, genetic diversity, and we employed a continuous phylogeographic approach to infer the dispersal history of its plastid lineages.

We identified five plastid lineages that diverged during the Early or Middle Pleistocene and are parapatric, suggesting past population fragmentation. Four lineages are endemic to Lower Guinea, and one spans the Congo Basin. We found contrasting patterns of expansion in the two regions, with a rapid and recent range expansion of the Congolian lineage in the last 200,000 years, while the spread of the Lower Guinean lineages was substantially slower.

The contrasting demographic histories between eastern and western lineages, associated with contrasted levels of plant species richness and rates of endemism, suggest that forest cover was more stable in Lower Guinea during the Late Pleistocene than in Congolia, where the biodiversity might have been eroded before the forest re-expanded in the Congo basin. This study illustrates how a continuous phylogeographic inference approach, mostly applied so far for inferring the spread of fast-evolving pathogens over months or years, can provide new insights to reconstruct the dispersal history of tropical tree species over thousands or millions of years.

Sequencing was performed on an Illumina NovaSeq sequencer at the GIGA platform in Liége, Belgium. Demultiplexing was done by the GIGA platform. 

Files are paired-end. Certain files have a small amount of reads, or are better treated as single-end data.