Application of expert systems in vegetation classification

The aim of this study was to evaluate the efficiency of two expert systems in automatic classification of pre-defined yew (Taxus baccata L.) communities in the central and eastern of the Hyrcanian forests. Therefore, initial plant communities were recognized using Two-Way Indicator Species Analysis (TWINSPAN) based on floristic data of 408 relevés with 400 m2 areas each. To determine representative species, the integrative results of Phi-coefficient index, constancy ratio, and the ratio of average cover were used. Five plant communities were introduced after transferring the initial communities to the Braun-Blanquet synthesis table. The assignment of relevés into the fifth yew associations were done by functional and sociological species groups as two different automatic algorithms of expert system for vegetation classification. The results showed that all relevés in the expert system based on functional species groups were correctly assigned to the target groups. This method was in 100% agreement with the existing yew syntaxa. However, 87 relevés were not assigned to any of the five pre-defined yew associations, or they were assigned to more than one unit in the Cocktail method, which showed 78.7% agreement with the previously yew syntaxa. Finally, our results illustrate that application of expert system based on functional species groups is preferred to sociological species groups in automatically plant community classification due to the importance of consistency and flexibility of vegetation classifications and the correct assignment of relevés to the target units.