We present a Markov model of a land-use dynamic along a forest corridor of Madagas- car. These geographic spots are of primary ecological and economic importance. We have a data set of 22 years on 42 plots, each plot can take four states. The initial state is removed and we use a Markov model with three states. A initial processing of data by the maximum likelihood approach leads to a model admitting an absorbing state. We study the quasi-stationary distribution law of the model and the law of the hitting time of the absorbing state. According to experts, a transition not present in the data, must be added in the model leading to a Bayesian approach to identify the model. We obtain a regular model admitting an invariant distribution law. We study the speed of convergence to equilibrium. Finally we analyze the two dynamics that have been identified, on a realistic time scale, leading to consistent results.