Amplification with rep-PCR primers was performed
twice with each primer set. In total 21 reproducible
bands were scored and used for analysis. Products of
amplification with REP, ERIC and BOX primer sets are
shown in Fig. 4a, b and c, respectively. Individual cluster
analysis of REP, ERIC and BOX data revealed very
similar groupings (data not shown). Since a larger data
set increases the validity of groupings obtained by cluster
or principal coordinate analysis, data from all three repPCR
fingerprinting methods were combined in a single
similarity matrix. Cluster analysis of the combined repPCR
data (Fig. 5) revealed groupings similar, but not
identical, to those obtained from RAPDs. The six blood
disease isolates formed a single homogeneous group
which is closely related to the R. solanacearum clove
isolates R142 and B9043. The close relationship
between R142, B9043 and the clove pathogen P. syzygii
(isolate R001), which was suggested by the RAPD data,
can more clearly be seen from the principal coordinate
analysis of the rep-PCR data shown in Fig. 6. The first
two dimensions of the analysis shown here account for
71% of the total variation. South-east Asian moko and
bugtok isolates again grouped very closely, although the
rep-PCR primers did not appear to be as discriminatory
between these and the Central American SFR strains as
were the RAPD primers.