The plant community, as a subject of study and modeling, is a complex, poorly formalized system. In the report various independent approaches to modeling such systems are analyzed in order to select the best one. The results of modeling inter-annual dynamics of the plant community phytomass are presented. For this end we used the data-based approaches: the regression analysis and the modeling method based on artificial neural networks. The knowledge-based approach providing fuzzy opportunity models was also applied. The conclusion was made about the advantages and the limitations of the considered approaches.