Reliability of phenotypic tests for identification of Acinetobacter species

A numerical approach was used for identification of 198 Acinetobacter strains assigned to DNA groups according to the classification of Tjernberg and Ursing (I. Tjernberg and J. Ursing, APMIS 97:595-605, 1989). The matrix used was constructed from data published by Bouvet and Grimont (P.J.M. Bouvet and P.A.D. Grimont, Int. J. Syst. Bacteriol. 36:228-240, 1986) and Bouvet and Jeanjean (P.J.M. Bouvet and S. Jeanjean, Res. Microbiol. 140:291-299, 1989). The tests chosen were those of the simplified identification scheme for Acinetobacter species devised by Bouvet and Grimont (P.J.M. Bouvet and P.A.D. Grimont, Ann. Inst. Pasteur/Microbiol. 138:569-578, 1987), namely, growth at 37, 41, and 44 degrees C, oxidation of glucose, gelatin hydrolysis, and assimilation of 14 carbon sources. Of the strains tested, 181 represented 12 DNA groups in the matrix; at a probability level of greater than or equal to 0.95, 78% of them were correctly identified, 2.2% were misidentified, and 19.8% were not identified. Seventeen strains represented two DNA groups not included in the matrix; nine of them were incorrectly assigned to a DNA group by these phenotypic tests. Because of problems of separating strains belonging to DNA groups 1, 2, 3, and 13 by using the phenotypic tests proposed by Bouvet and Grimont (Ann. Inst. Pasteur/Microbiol.), we suggest that these groups should be referred to as the Acinetobacter calcoaceticus-A. baumannii complex.