Sequence-enabled community-based microbial source tracking in surface waters using machine learning classification: A review

The development of Microbial Source Tracking (MST) technologies was borne out of necessity. This was largely due to the: 1) inadequacies of the fecal indicator bacterial paradigm, 2) fact that many fecal bacteria can survive and often grow in the environment, 3) inability of traditional microbiological methods to attribute source, 4) lack of correspondence between numbers of fecal indicator bacteria in waterways and many human pathogens, and 5) source allocation requirements and load determinations needed for total maximum daily loads. The MST tools have changed over time, evolving from culture-dependent to culture-independent molecular analyses. More recently, MST tools based on microbial community analyses, mainly DNA sequencing-based approaches, have been developed in an attempt to overcome some of these issues. These approaches generate large data sets and require the use of sophisticated machine learning algorithms to allocate potential host sources to contaminated waterways. In this review we discuss the origins and needs for community-based MST methods, as well as elaborate on the Bayesian algorithm-based program SourceTracker, which is increasingly being used for the determination of sources of fecal contamination of waterways.