Choice of indicator organism and library size considerations for phenotypic microbial source tracking by FAME profiling

The primary objective of this study was to investigate the effects of choice of the indicator organisms on the accuracy of classifying the fatty acid methyl ester (FAME) profiles of the known-source library isolates. First, a known-source library containing the FAME profiles of Enterococcus isolates cultured from six different possible sources of microbial pollution was developed. A total of 511 Enterococcus isolates were profiled: 120 isolates from sewage samples representing humans; 69 from dairy and cattle cow; 74 from chicken; 76 from swine; 94 from deer; and 78 from waterfowl. Classification of known-source Enterococcus isolates into their respective host categories resulted with a 66% average rate of correct classification (ARCC) in a six-way discriminant analysis (DA). The ARCC increased to 75% when the individual hosts were pooled into larger categories of human, livestock, and wildlife. The accuracy was 80% when isolates of human origin were discriminated against those of non-human origins. Recently, several studies reported the ARCCs for various classification schemes associated with total coliform (TC), fecal coliform (FC), and Escherichia coli of the known-source isolates. When the accuracy of classification of Enterococcus isolates was compared to those reported for TC, FC, and E. coli isolates, the lowest ARCCs were associated with classification of E. coli isolates, the only species level indicator organism among the four compared. It was found that the degree of discrimination increases as the indicator becomes more inclusive of bacteria from different genus. In addition, random cluster formation analysis indicates that known-source libraries with isolate numbers between 300 and 500 might be sufficient for MST by FAME.