Fidelity of bacterial source tracking:: Escherichia coli vs Enterococcus spp and minimizing assignment of isolates from nonlibrary sources

The goal of the study was to improve the fidelity of library-dependent bacterial source tracking efforts in determining sources of faecal pollution. The first objective was to compare the fidelity of source assignments using Escherichia coli vs Enterococcus spp. The second objective was to determine the efficacy of using thresholds during source assignments to reduce the rate of misassignments when nonlibrary isolates (i.e. isolates from animals not used in building the identification library) are present. E. coli and Enterococcus isolates from 784 human, cow, deer, dog, chicken, and gull faecal samples were fingerprinted using BOX-PCR. Jack-knife analysis of the fingerprints showed that the overall rate of correct assignment (ORCA) of 867 E. coli isolates was 67% compared with 82% for 1020 Enterococcus isolates. In a separate blind test using similarity value and quality factor thresholds, the ORCA of 130 E. coli and 131 Enterococcus isolates were 70% and 98%, respectively. The use of these thresholds reduced misassignment of 262 nonlibrary enterococcal isolates from horses, goats, pigs, bats, squirrels, ducks, geese, and migratory song birds. Misassignment was reduced from 100% when thresholds were not used, to 47% using similarity threshold alone, and to 12% when both thresholds were used. The use of enterococci provides higher rates of correct source assignment compared with E. coli. The use of similarity thresholds to decide whether to accept source assignments made by computer programmes reduces the rate of misassignment of nonlibrary isolates. Although both E. coli and Enterococcus spp. are still used in microbial source tracking, the use of enterococci should be preferred over the use of E. coli in DNA fingerprint-based efforts. In addition, because environmental isolates are not limited to those from animals used to build source tracking libraries, similarity thresholds should be used during source assignments to reduce the rate of misassignments.