Plasmids are autonomous genetic elements that can be exchanged between microorganisms via horizontal gene transfer (HGT). Despite the central role they play in antibiotic resistance and modern biotechnology, our understanding of plasmids’ natural ecology is limited. Recent experiments have shown that plasmids can spread even when they are a burden to the cell, suggesting that natural plasmids may exist as parasites. Here, we use mathematical modeling to explore the ecology of such parasitic plasmids. We first develop models of single plasmids and find that a plasmid’s population dynamics and optimal infection strategy are strongly determined by the plasmid’s HGT mechanism. We then analyze models of co-infecting plasmids and show that parasitic plasmids are prone to a “tragedy of the commons” in which runaway plasmid invasion severely reduces host fitness. We propose that this tragedy of the commons is averted by selection between competing populations and demonstrate this effect in a metapopulation model. We derive predicted distributions of unique plasmid types in genomes—comparison to the distribution of plasmids in a collection of 17,725 genomes supports a model of parasitic plasmids with positive plasmid–plasmid interactions that ameliorate plasmid fitness costs or promote the invasion of new plasmids.

质粒是自主遗传元件，可以通过水平基因转移 (HGT) 在微生物之间进行交换。尽管它们在抗生素耐药性和现代生物技术中发挥着核心作用，但我们对质粒自然生态学的了解是有限的。最近的实验表明，即使质粒对细胞造成负担，它们也可以传播，这表明天然质粒可能作为寄生虫存在。在这里，我们使用数学模型来探索这种寄生质粒的生态学。我们首先开发了单个质粒的模型，发现质粒的种群动态和最佳感染策略很大程度上取决于质粒的 HGT 机制。然后，我们分析了共感染质粒的模型，并表明寄生质粒容易发生“公地悲剧”，其中失控的质粒入侵严重降低了宿主适应性。我们建议通过竞争种群之间的选择来避免这种公地悲剧，并在集合种群模型中证明这种效应。我们推导出基因组中独特质粒类型的预测分布——与 17,725 个基因组集合中质粒的分布进行比较，支持具有正质粒-质粒相互作用的寄生质粒模型，可改善质粒适应性成本或促进新质粒的入侵。