The babassu palm tree (Attalea speciosa Mart. ex Spreng.) is an endemic species of the amazonian forests. For decades, the progress of the pioneer front highlights this palm tree in the anthropogenic open areas: pastures and cultivated fields. The babassu is one of the "extractive" resources in Brazil: gathering activity followed by marketing of non-timber products. This activity involves people among the most disadvantaged in the country. Despite this fact, the knowledge of long term functioning of the palm tree is sorely lacking. Our goal is to understand its population dynamics in the pastures and to guide local populations towards a sustainable management of the species. Our modelization process is based on field data series collected between 2013 and 2015. This field research work was conducted in situ in the community of Benefica, state of Para, Brazil. First we selected transects inside pastures randomly, then we counted systematically all the individuals inside those transects. We had analyzed the babassu life cycle and we decided to modelize each one of the three major biological processes: mortality, growth and recruitment. We actually develop a multinomial hierarchical model approach which relies on aggregated data: sum of the individuals per stage and per transect. In this talk we present a Bayesian modeling framework analysis with MCMC algorithms for estimate mortality, growth and recruitment rates. Mortality and growth rates are estimated for all six biological stages. We detect and highlight a critical transition between stages 3 and 4. Biologicaly this transition corresponds to the transfer of the bud from the soil to the surface. The vulnerability of babassu thus increases enormously. Our results also underline differences between transects - important location variability.