Bdw. Jarvis et al., IDENTIFICATION OF AGROBACTERIUM AND RIZOBIUM SPECIES BASED ON CELLULAR FATTY-ACID COMPOSITION, Plant and soil, 184(1), 1996, pp. 143-158
The increasing number of phylogenetically defined species in the gener
a Agrobacterium, Rhizobium and Sinorhizobium suggests a need for a rap
id identification method which will distinguish between these species.
We have examined 65 strains of Agrobacterium representing: A. tumefac
iens, (34); A. rhizogenes, (16) A. vitis, (10) Al rubi (2) and some un
classified strains, and 150 strains of Rhizobium and Sinorhizobium rep
resenting: R. etli (21); R. galegae (20); R. huakuii (17); R, legumino
sarum (20); R. loti (16); R. topici (18); S. fredii( 19); and S. melil
oti (20). Fatty acid methyl esters (FAME) were obtained from each stra
in, as previously described, and analysed by gas-chromatography using
the MIDI Hewlett-Packard Microbial Identification System (MIS). Fatty
acid profiles were recorded, characteristic fatty acids noted and the
overall similarities between fatty acid profiles for each species calc
ulated. Relationships between species were also derived from the fatty
acid data by principal component analysis. This showed overlapping cl
usters for strains of R, leguminosarum and R, etli, R. topici and A. r
hizogenes and S. fredii and S. meliloti within one supercluster. Strai
ns of A. tumefaciens, A I nbi, A. vitis and R. galegae formed a second
supercluster while R. loti and R. huakuii strains formed a third clus
ter well separated from all the other strains. The fatty acid profiles
were used to correctly identify at least 94% of the strains represent
ing each species in the collection except R. etli. R. etli strains (23
.8%) were misidentified as R. leguminosarum. This was attributed to th
e high similarity (44.7%) between R. etli and R. leguminosarum. It is
concluded that whole cell fatty acid analysis should form part of the
polyphasic description of new species of root nodule bacteria, with th
e proviso that growth conditions and analytical methods be carefully s
tandardized. It is suggested that FAME-MIS system and the database we
have compiled provide a basis for future development.