Résumé
Background: In wild plant populations, genetic divergence within
continuous stands is common, sometimes at very short geographical scales.
While restrictions to gene flow combined with local inbreeding and genetic
drift may cause neutral differentiation among subpopulations,
microgeographical variations in environmental conditions can drive
adaptive divergence through natural selection at some targeted loci. Such
phenomena have recurrently been observed in plant populations occurring
across sharp environmental boundaries, but the interplay between selective
processes and neutral genetic divergence has seldom been studied. Methods:
We assessed the extent of within-stand neutral and environmentally-driven
divergence in the Neotropical tree Eperua falcate Aubl. (Fabaceae) through
a genome-scan approach. Populations of this species grow in dense stands
that cross the boundaries between starkly contrasting habitats.
Within-stand phenotypic and candidate-gene divergence have already been
proven, making this species a suitable model for the study of genome-wide
microgeographic divergence. Thirty trees from each of two habitats
(seasonally flooded swamps and well-drained plateaus) in two separate
populations were genotyped using thousands of AFLPs markers. To avoid
genotyping errors and increase marker reliability, each sample was
genotyped twice and submitted to a rigorous procedure for data cleaning,
which resulted in 1196 reliable and reproducible markers. Results: Despite
the short spatial distances, we detected within-populations genetic
divergence, probably caused by neutral processes, such as restrictions in
gene flow. Moreover, habitat-structured subpopulations belonging to
otherwise continuous stands also diverge in relation to environmental
variability and habitat patchiness: we detected convincing evidence of
divergent selection at the genome-wide level and for a fraction of the
analyzed loci (comprised between 0.25% and 1.6%). Simulations showed that
the levels of differentiation for these outliers are compatible with
scenarios of strong divergent selection.