Characterization and source identification of antibiotic resistance genes in the sediments of an interconnected river-lake system

•Characteristics and source-sink relationship of ARGs in the river-lake system was studied.•Abundant and diverse ARGs were detected in the sediments of the river-lake system.•Higher potential resistiome dissemination risk was observed in Lake Baiyang than other lakes.•Sewage pollution impacted the prevalence of ARGs in the river-lake system.•Fuhe River was the dominant contributor of ARGs in the sediments of Lake Baiyang. Antimicrobial resistance has been a global public health concern. The river-lake systems are one of the tightly connected terrestrial ecosystems and, appear to be reservoirs of antibiotic resistant genes (ARGs) and dispersal routes of resistant pathogens because they are easily impacted by human activities. Currently, systematic knowledge on the prevalence, transfer risk and source of ARGs in river-lake systems is largely lacking. In this study, we focused on the high-throughput profiling and source-sink relationship disentangling of ARGs in the sediments of an interconnected river-lake system (Fuhe River and its receiving Lake Baiyang in northern China). To this end, 40 surface sediments were collected for metagenomic shotgun sequencing. The profile and co-occurrence of ARGs in the sediments of the river-lake system were comprehensively characterized, as well as the mobile genetic elements (MGEs) carrying ARGs and their potential resistome dissemination risk. CrAssphage, a recently-discovered bacteriophage, was used to track human fecal pollution on the prevalence of ARGs. Meanwhile, a novel fast expectation-maximization microbial source tracking (FEAST) method was combined with linear discriminant analysis effect size method (LEfSe) for quantitatively apportioning the contribution of river sediment to the presence of ARGs in the receiving lake. Results showed abundant and diverse ARGs (24 types consisting of 510 subtypes) were detected in the sediments of the river-lake system, including some emerging ARGs such as mcr-1, tetX and carbapenemases types. Network analysis suggested non-random co-occurrence patterns of ARGs within the same type and among different types. Importantly, a number of MGE-carrying contigs were identified with jointly containing one or more ARGs, resulting in higher resistome risk potential in Lake Baiyang than many worldwide lakes. Source tracking indicated the prevalence of ARGs in the sediments of the river-lake system might be largely explained by the extent of human fecal contamination, and apportionment estim