Numerical taxonomic analysis of cross-inoculation patterns of legumes and Rhizobium

Summary A survey was made of published results of tests of the capacity of Rhizobium derived from one legume genus to nodulate plants from other genera. The data were derived from more than 14,000 separate cross-inoculation trials involving species from 165 genera of legumes. Numerical taxonomic techniques were applied to 113 of the genera for which results of substantial cross-infection tests were available. The data were examined using mean character difference coefficients re-expressed as total and positive-only similarity coefficients. The resulting similarity matrices were clustered by the unweighted pair-group method using arithmetic averages. Eighteen affinity groups were defined at the 70% similarity level. With few exceptions, the physiological and cultural behavior of the rhizobia was consistent within the defined groups. Two broad categories were suggested in the numerical taxonomic analysis, and their validity is discussed in regard to the geographic, physiological and cultural characteristics of the legumes and their Rhizobium microsymbionts. The taxonomic and agronomic value of this approach and the new groupings are discussed.