Salt marshes provide a range of ecosystem services and yet are subjected to anthropogenic impacts that alter the biogeochemical processes underlying these services. In particular, human activities may modify salt marsh greenhouse gas (carbon dioxide, methane, nitrous oxide) emissions by changing plant and microbial communities, hydrological regime, and sediment chemistry. Quantifying the effects of human impacts on greenhouse gas emissions is important for complete carbon budgets, and for effective management of salt marshes and the ecosystem services they provide. In Chapters 1 and 2, I investigate the effects of hydrology and plant invasion on greenhouse gas emissions. First, I show how the restriction and restoration history of four salt marshes influence methane flux in unpredictable ways. Despite comparable salinity, methane emissions from one partially restored marsh were 25 times higher than unimpacted reference sites 13+ years after restoration, but emissions from other restored sites were equal or lower. Next, I show that greenhouse gas emissions associated with invasive Phragmites australis are not different from those associated with native Spartina alterniflora. These Chapters demonstrate the de-coupling of greenhouse gas emissions, and carbon sequestration more generally, from ecosystem degradation and restoration. In Chapters 3 and 4, I quantify greenhouse gas fluxes and microbial community structure under precipitation changes that may occur with global climate change. In a field experiment, doubled rainfall and drought had significant transient impacts on porewater salinity following storms, and on the community structure of plants (doubled rainfall) or microbes (drought), yet greenhouse gas fluxes and other biogeochemical processes were not affected. The absence of biogeochemical change indicates functional redundancy and resistance or resilience exist in the microbial community, suggesting marshes may continue providing services as precipitation changes. In a lab experiment, rewetting intact cores to simulate tidal inundation or rainstorms produced a nitrous oxide pulse 10-20x the baseline flux rates, without changing the microbial community. A model of rewetting event frequency suggests that pulsed emissions may be responsible for the majority of marsh nitrous oxide emission. Precipitation change may increase coastal nitrous oxide emission if it causes more or stronger storms, and thus more rewetting events.