Microbes exist in virtually all environments on Earth’s surface. As asexually reproducingorganisms, one strategy microbes employ to adapt to these environments is horizontal gene transfer, the movement of genetic material from one cell to another rather than parent to offspring. Plasmids, small circular DNA molecules that often carry beneficial traits, can facilitate this transfer of genetic material. In this dissertation, I discuss the identification of plasmid sequences, and many different aspects of their subsequent characterization. In the first half of this thesis, I describe the development of a machine-learning based model that I used to predict 68,350 plasmid sequences from human gut metagenomes. Downstream characterization of the genetic content of these plasmids reveals evolutionary patterns called ’plasmid systems’ resulting from plasmid recombination. Plasmid systems are comprised of backbone genes, encoding basic functions for plasmid replication, and cargo genes, encod- ing fitness determining genes. I then present an example where the environmental variable of chloramphenicol usage correlates with the acquisition of of chloramphenicol resistance as cargo genes in plasmid systems. In the second half of this thesis, I focus on a par- ticularly prevalent and abundant plasmid called pBI143 that is present in up to 92% of individuals across 4,513 metagenomes. I show that the host range of pBI143 is broad, span- ning Bacteroides, Parabacteroides and Phocaeicola, and that pBI143 can transfer between these genera. pBI143 is specific humans, appearing in only human- and sewage-associated metagenomes, and lacking in any other environmental samples. pBI143 only encodes 2 genes, repA for plasmid replication, and mobA for plasmid transfer, and exists in 3 predominant versions that differ in their repA sequences. Across our metagenomes, I show that pBI143 is under strong purifying selection, and that it is monoclonal in most individuals. pBI143 is transferred from mothers to infants, and I suggest that the monoclonal nature of pBI143 may be due to priority effects of the first pBI143 version to inhabit the gut after birth. To address how pBI143 impacts the bacterial hosts, I construct isogenic strain sets of cells with and without the naive version of pBI143 and compete these strains in gnotobiotic mice, which shows no clear fitness benefit or detriment to host cells carrying this plasmid. How- ever, pBI143 is able to take up additional cargo genes in nature, suggesting it acts as a "discretionary parasite", where it transitions between a state of benefiting or parasitizing the host cell. The plastic nature of this plasmid makes it a good candidate for gene de- livery to human gut microbiomes. Similar to other mobile genetic elements, I showed that pBI143 increases its copy number in vitro when the host cell is stressed, and tested this same phenomenon in naturally occurring stressful environments to demonstrate that pBI143 also increase its copy number during inflammatory bowel disease and has future potential as a diagnostic biomarker. Finally, given the widespread, abundant, and human specific nature of pBI143, we showed that this plasmid can be used as an amplifiable biomarker of human fecal contamination in water samples.