Volume 37, no. 3, p. 581-590, 1999. In our article, we suggested that non-O1 strains of identical or closely similar electrophoretic type (ET) collected on different continents represent clonal lineages (Tables 3 and 5 in the original article). Reconsideration of our data, however, indicates that for many of these sets of strains an equally plausible if not more likely explanation of multilocus genotypic similarity is the independent recombinational assembly of common alleles. This is the case for ET 196 and ET 128 (Table 3), each of which has a genotype consisting of alleles that occur in high or moderate frequency in populations, as well as for many of the ETs listed in Table 5, which pertains to sets of strains differing at single loci.

However, for a number of ETs of non-O1 strains, a clonal relationship is probable because their genotypes include unique or rare alleles. Thus, for example, the genotype of ET 256, which was represented by a strain collected in India in 1979 and five strains recovered from patients in Mexico and Guatemala in the early 1990s, includes a unique allele of leucine aminopeptidase, an extremely rare indophenol oxidase allele, and uncommon alleles at four additional loci. Similarly, a clonal relationship is indicated for a pair of serotype O44 strains from India (1973) and Mexico (1991) that share a unique allele of nucleoside phosphorylase and a rare allele of phosphoglucomutase.

Multilocus enzyme electrophoresis is of limited use in identifying clonal lineages because, as we have noted, electromorphs cannot be equated with isoalleles and convergence in electrophoretic mobility of an enzyme is not infrequent. To determine clonal relationships among strains with a high degree of confidence, sequence data for multiple housekeeping genes will be required. This reinterpretation of the likely status of certain groups of non-O1 strains does not affect any other aspect of the work reported in our study.