Abstract
We used six rhizobium-legume systems to test the reliability of the most-probable-number (MPN) technique for enumerating rhizobia introduced into 14 sites representing four soil orders. The range-of-transition values (the number of dilution steps between the first not-entirely-positive and the last not-entirely-negative growth units) were compared for each species and for each soil. The probability that the observed data were significantly different from theoretical values varied with the species. The acceptability of MPN codes (P > 0.99) was the highest (97 to 99%) with Vicia sativa, Trifolium repens, and Glycine max and lowest (72%) with Leucaena leucocephala. Medicago sativa and Macroptilium atropurpureum yielded 87 and 75% acceptable MPN codes, respectively. The acceptability of the MPN data obtained for a host species was related to rooting habit and time to nodulation. Comparison of data for each soil indicated that, despite large differences in characteristics, the soil was not a major source of variability in the MPN counts. There was no significant interaction of the range of transition of rhizobium-legume plant infection count data between species and site.