The botanical identification of trees species in forest inventories is complicated. The majority of analytical identification keys is based on reproductive structures which quite often cannot be seen in the forest. Therefore, trees must be identified through vegetative characteristics. This paper discussed the use of a data base to do botanical identification of trees and presented a set of characteristics which can be used in computerized tree identification. GUESS, a program developed by NCSU was used to manage a data base with information about 226 species of 34 families. Each species was described wiht data obtained from 10-20 trees. All trees are in the Ducke Forest Reserve, Experimental Station of Tropical Silviculture and Forest Management Station, of INPA, 26 and 90 km on north of Manaus. Botanical material was identified in INPA Herbarium. The use of a data base helps botanists, foresters and other users. It is fast and reliable but it is just a complementary tool to consultation to Herbaria and specialists, which are essencial to a precise identification, due to the the great floristic diversity of the tropical rain forest.
Dendrology; taxonomy; botanical identification; computation