Fig. 4: The ecological drivers of population dynamics in Malagasy baobabs.

a, A PSMC model inferring the effective population sizes of Malagasy baobabs. A generation time of 13 years was used to translate substitution rates to ages. b, A coupled logistic model supporting the asynchronous population size decreases from around 1 Ma apparent in the PSMC model (Methods). c, Variation in ecological valence of six Malagasy baobabs to different environmental factors. The red arrows highlight the widest ecological range observed for A. za. d, Niche equivalency and similarity statistical tests and patterns of niche overlaps between baobabs. Coloured patches represent ecological niche differentiation of species pairs (pale blue and brown with areas of highest density being most intense; x axis, PC1; y axis, PC2). Areas of niche overlap are shown in blue. e, Geological events impacting the distribution of Malagasy baobabs. Changing sea levels over time (top, GMSL, RSL)39,48 was inferred by different models (Methods). The dynamics of the land area of Madagascar (middle) were derived from a calculation using the GMSL (orange patch) and RSL (dash line). The timescale of reported geological events31,48,49 is indicated by coloured strips. The predicted suitable areas of habitat for baobabs were mapped onto the simulated territories of the island under different sea-level scenarios (bottom). The acronyms used for species are as in Fig. 1a.