The second cooperative numerical taxonomic analysis of Runyon group IV rapidly growing mycobacteria was facilitated by combining the results of 12 individual participants, each performing one or more of the selected tests by his or her own choice of method. The 177 characters were coded for analysis and the matching matrix was generated. Eight of nine clusters which were defined in this investigation were of previously named species. The recognized clusters were as follows: Mycobacterium flavescens, M . vaccae, M . chitae, M . thermoresistibile, and M . smegmatis. One large cluster, designated "M. parafortuitum complex" was noted which contained strains representative of taxa earlier identified as M . parafortuitum, M . diernhoferi, M . aurum, M . neoaurum, and a cluster of strains of unnamed species. It is suggested that segregation of species or subspecies within this complex not be attempted at this time.Since the earlier report of a cooperative numerical analysis of rapidly growing mycobacteria (51, several new taxa within this group of acid-fast bacilli have been reported. To determine the distinctiveness and precise definition of these new rapidly growing species, a second cooperative study by the International Working Group on Mycobacterial Taxonomy (IWGMT) was carried out by 12 different laboratories. As in previous cooperative studies (5, 17), the same "permissive" philosophy (17) was used; i.e., each participant was permitted to perform any tests that he or she chose. The only stipulations were that all investigators had to submit their raw data to the study coordinator for collation and numerical analyses, and none was allowed to remove his data once the final numerical analysis was completed. This report presents the results of this study.
MATERIALS AND METHODSFifty-nine strains attributed to eight taxa of rapidly growing mycobacteria were examined. These strains are listed in Table 1 Yactae." Since our study began, one strain, ATCC 25795, at first called M . aurum, has become the type strain of M. neoaurum (12). In most instances, study strains were deposited directly with the American Type Culture Collection (ATCC) by the laboratory of origin. In a few cases, cultures were submitted through the study coordinator to the ATCC. The deposited strains were given code numbers at ATCC, freeze-dried, and sent to 19 members in 10 different countries.Each investigator performed his series of selected tests on the set of cultures received. In the final analysis, data were submitted to the study coordinator by only 12 of these 19 participants. Data from nine participants were of non-immunoserological nature and were used for numerical analysis, whereas those of the remaining three participants were either of serological character or represented other data not easily coded for computer analysis.
75