The growing prevalence of antibiotic resistance (AR) genes has rendered important pathogens nearly or fully unresponsive to antibiotics. Pathogens are thought to acquire AR traits through horizontal gene transfer (HGT) with an undefined reservoir of non-pathogenic bacteria across multiple environments, largely via plasmid conjugation. Yet, there remain unanswered questions about how broad and prolific the AR reservoir truly is due to insufficient methods for associating extrachromosomal DNA like plasmids with their bacterial hosts. Common molecular tools such as short- and long-read sequencing of microbiomes cannot unambiguously associate plasmids with the genome of the host bacterium, and while culture-based screening can provide a window into AR host range, surveying the entire gut flora is impractical due diverse culturing conditions and variable resistance phenotypes. To address these limitations, we developed OIL PCR (One-step Isolation and Lysis PCR), a sensitive, culture-independent, single-cell fusion PCR approach that can establish plasmid-host associations in complex communities with over 99% accuracy. Using OIL PCR, we correctly associate three beta-lactamases with a strain of Klebsiella in the stool of a neutropenic cancer patient across multiple time points. We also detect transient colonization by a strain of Escherichia carrying the TEM beta-lactamase and a direct physical interaction between Klebsiella and the gut commensal Romboutsia. We next sought to apply OIL PCR to assay the unknown reservoir of important AR genes within the gut. First, we established the currently known global prevalence and taxonomic breadth of 49 clinically relevant AR gene families. We analyzed 12,489 human gut metagenomes from 33 countries and paired our analysis with the AR profiles of nearly 600,000 isolate genomes. We observe that the most concerning AR genes, such as carbapenemases and colistin resistance, remain taxonomically restricted with low global prevalence—even in cases where the AR genes have been found on mobilizable plasmids. To screen for the existence of an unknown expanded AR reservoir within the gut, we employed OIL PCR to identify the hosts of three clinically-relevant AR genes, CTX-M, QnrS, and AAC(6’)-Ib, in the stool of individuals from Honduras, India, Pakistan, and Vietnam. We found that these genes remain taxonomically restricted to known Enterobacteriaceae. These data suggest that globally prevalent and clinically relevant AR genes have not established themselves across diverse commensal gut microbiota. Lastly, we transitioned from focusing specifically on AR genes to investigating the vectors which likely facilitate their transfer between pathogens and the non-pathogenic reservoir. Specifically, we chose to examine broad host range (BHR) plasmids which are self-transmissible conjugal elements that can transfer into distantly related bacterial hosts. Understanding the host range of concerning BHR plasmid would allow improved predictions of which non-pathogenic bacteria may be a responsible for transferring AR genes into pathogens. We developed a bacterial compatible dCas9-deaminase based palindromic lineage the tracing cassette, pDrome, with the goal of tracing the spread of BHR plasmids through complex bacterial communities. The pDrome recording array is less than 200 bp in length but can generate millions of unique barcodes over the course of several days. We integrate pDrome into the BHR plasmid RP4 and find that we can accurately reconstruct a known lineage of conjugation events over three days. pDrome will be a powerful tool for assessing the host range of BHR plasmids and provide deep insight into the factors which govern HGT of AR genes in all environments.