In the last decades, data fusion techniques proved their performances especially in the case of complex recognition system. The idea is to use those techniques in the context of tree species recognition. In this paper, we propose to fuse information extracted from barks with those extracted from leaves. The goal is to increase the power of discrimination of the proposed fusion system and to achieve better detection results than those obtained when using only the leaves. Two fusion system architectures, corresponding to two fusion strategies are compared in this paper.