Résumé
Spatial genetic structure and diversity within forest tree populations. An isolation by distance model is presented, which is better fitted to forest tree populations than those available to date. This model was developed in order to understand which factors might determine genetic diversity and spatial structure within populations of these species, for in situ management of their genetic resources. It allows a nonuniform spatial distribution of individuals, as well as some selfing. This model was studied by means of simulations, using a complete factorial design with 32 highly contrasted cases, which resulted from the combination of different values of the following input variables: global density, seed and pollen dispersal distances, generation overlap and overdominant selection. Our results show that much of the variation in spatial genetic structure and inbreeding at the individual level can be explained by variations in clumping and outcrossing rate. Inbreeding at the population level is mainly determined by the degree of generation overlap. As a general conclusion, all the input variables studied here do affect the genetic characteristics of simulated populations, and most interaction effects between these variables are highly significant. These results suggest that the explanation of differences in the level and/or structure of within-population genetic diversity between tree species could presumably be improved by using additional classification factors such as density, the level of clumping, generation overlapping, and/or selection, as well as by taking into account the interaction effects between classification factors. (C) Inra/Elsevier, Paris