EVALUATION OF TESTS EMPLOYED IN THE NUMERICAL TAXONOMY OF RIVER BACTERIA

Comparisons of native bacterial populations from different locations to determine possible environmental influences can be accomplished using numerical taxonomy. This study is centered on the data base used for the numerical analysis. Numerical taxonomy tests selected to study aerobic, heterotrophic river bacteria were analysed for cluster formation ability. Character usefulness was assessed by using a unique computer program, DINDEX, based on a test ranking system whereby the ind" of cluster formation ability was calculated by linking the formula to the numerical taxonomy program, an analytical device not previously reported. The 220 tests were evaluated on two data sets comprising 636 strains with a variety of genera and species of river bacteria. Of the 220 tests, 31 had a cluster formation index of 0.0 which warranted eliminating them. Of these, 19 had a positive frequency of less than 20% and 12 had a positive frequency of greater than 80%. This implied that the selection of the data base from Bergey's Manual was a sound decision overall. To illustrate some of the disparities which were found in certain clusters, a small cluster of Pseudomonas fluorescens which appeared within a much larger cluster of Pseudomonas caryophylli was used as an "ample of a typical cluster and was analysed for phenotypic concordance with the description in Bergey's Manual of Systematic Bacteriology. Forty-five tests were not in agreement with the Manual. There was a high degree of compliance with the classical descriptions in spite of the variety of environmental influences on the bacteria in the river.