The development and expansion of ponds within otherwise vegetated coastal marshes is a primary driver of marsh loss throughout the world. Previous studies propose that large ponds expand through a wind wave-driven positive feedback, where pond edge erosion rates increase with pond size, whereas biochemical processes control the formation and expansion of smaller ponds. However, it remains unclear which mechanisms dominate at a given scale, and thus how, and how fast, ponds increase their size. Here, we use historical photographs and field measurements in a rapidly submerging microtidal marsh to quantify pond development and identify the processes involved. We find that as small ponds emerge on the marsh platform, they quickly coalesce and merge, increasing the number of larger ponds. Pond expansion rates are maximized for intermediate size ponds and decrease for larger ponds, where the contribution of wave-driven erosion is negligible. Vegetation biomass, soil shear strength, and porewater biogeochemical indices of marsh health are higher in marshes adjacent to stable ponds than in those adjacent to unstable ponds, suggesting that pond growth rates are negatively related to the health of the surrounding marsh. We find that the model of Vinent et al. (2021) correctly predicts measured pond growth rates and size distribution, which suggest the different mechanisms driving pond growth are a result of marsh drowning due to sea level rise (SLR) and can be estimated by simplified physical models. Finally, we show that all relevant processes increasing pond size can be summarized by an empirical power-law equation for pond growth which predicts the temporal change of the maximum pond size as a lower bound for the total pond area in the system. This gives a timescale for the growth of ponds by merging and thus the critical time window for interventions to prevent the irreversible pond expansion associated with large scale pond merging.