Waterborne diseases are transmitted to humans through the fecal contamination of water, where homeothermic species are the main reservoir. Fecal indicator bacteria (FIB) are often used to determine the occurrence of fecal contamination. However, FIB cannot provide the source of fecal contamination. Furthermore, as fecal inputs and contamination could originate from multiple sources (e.g., human, livestock, wildlife), multiple source tracking markers are required to identify fecal sources. From a previous study, we developed a mitochondrial DNA (mtDNA) metabarcoding approach to assess the presence of multiple homeotherms in four surface waters. Here, we have broadened our approach by sampling 86 surface water samples from the L'Assomption River and Ville-Marie watersheds (Province of Quebec, Canada). Fecal coliform levels were higher than the expected sanitary recommendations for recreational water (> 200 CFU/100 mL) in 73 % samples. The occurrence of mtDNA from human, livestock, domestic animals, wild mammals and wild birds was found in 40-88 % of the samples. Multivariate analyses showed significant covariations between homeothermic taxa and fecal coliforms, enterococci, ß-D-glucuronidase, conductivity, the human-specific Bacteroidales Hf183 genetic marker, and the human population, in the watersheds of L'Assomption River (p = 0.001) and Ville-Marie (p = 0.015) (Province of Quebec, Canada). Through the application of Bayes Theorem, it was determined that fecal coliforms co-occurred with the detection of bovine, beaver, robin and chicken mtDNA in 100 % of cases in the L'Assomption River watershed, and human mtDNA co-occurred with fecal coliforms in 93 % and 76 % of cases in L'Assomption River watershed and Ville-Marie sub-catchment, respectively. This study suggests that fecal contamination could be the result of multiple species, among which some wild animals may contribute to fecal inputs in surface waters, resulting in potential risk to human health. This reinforces the necessity of using the mtDNA metabarcoding method to monitor multi-animal species.