Numerical analysis of 211 strains of rapidly growing, nonphotochromogenic mycobacteria was carried out by using 116 characters. New species reported after 1981 and not yet compared with known species by numerical classification were included. At the level of 90% of the matching coefficient, the following species could be differentiated from each other and were shown as distinct clusters : Mycobacterium pulveris, M. moriokaense, M. chitae, M. diernhoferi, M. porcinum, M. fortuitum, M. chelonae subspecies chelonae, M. chelonae subspecies abscessus, M. smegmatis, M. agri, M. fallax, M. parafortuitum, M. fortuitum subspecies acetamidolyticum. Four taxa, M. fortuitum, M. porcinum, M. chelonae subsp. chelonae, and M. chelonae subsp. abscessus, were regarded as distinct. However, these were combined into one large cluster at a level of 90%) of the matching coefficient and, therefore, were regarded as forming one series or complex. M. fortuitum subsp. acetamidolyticum was reported as a subspecies of M. fortuitum, but this was differentiated clearly from the other species including M. fortuitum.Numerical analysis of rapidly growing mycobacteria has been carried out by Kubica et al (6), Saito et al (14), Tsukamura (22), and Tsukamura et al (27,28). Since 1981, seven new species (or subspecies) of rapidly growing, nonphotochromogenic mycobacteria have been reported : Mycobacterium agri (22) (32), and M. fortuitum subsp. acetamidolyticum (31). It was desirable, therefore, for a new numerical analysis including these new species to be carried out to clarify their taxonomic positions. The present study was undertaken for this purpose.
MATERIALS AND METHODSA total of 211 strains of rapidly growing, nonphotochromogenic mycobacteria were studied. Mycobacterium parafortuitum strains are not nonphotochromogenic, however, these were included in the study, as they are nonpigmented when incubated in the dark. In addition, three hypothetical median organism patterns (9) of M. fortuitum, M. pulveris, and M. porcinum were included. Furthermore, 11 strains were tested in duplicate to estimate experimental errors. The strains used are shown in Table 1. 863