Résumé
The concept of population is very useful but can sometimes lead to dead ends. Indeed, various questions in population genetics cannot be solved if studied at this level. It is shown that the intensity of dispersion, as far as it is genetically determined, does not respond to selection at the level of the population in its usual sens. A simple theorical model in relation with in-situ observations (on carduus), seems to show that a wider set, the matapopulation (Gill), is necessary to account for the processes concerned. From this viewpoint, instead of considering species as sets of independent populations, it is proposed to consider them as sets of metapopulations where individual populations are regulary founded by the others and then evolve under internal pressures. Are these internal pressures acting at the very population level ? It does not seem so, at least for some, and perhaps for most species, since each individual is likely to mate with a subset which is not representative of the whole. This question has led Wright to formulate the neighbourhood concept. In Thyme, the simultaneous integration of the emerging properties of the 3 levels (Metapopulation, Population and Neighbourhood) allows one to explain a phenomenon (very high proportions of females) which remained incomprehensible as long as one tried to describe it using only the population level.