In this work we present work focused on within farm dynamics of Mycobacterium avium subsp paratuberculosis (MAP) infection on commercial US dairies. We focused on using deterministic mathematical models informed by molecular epidemiology and analysis of longitudinal data to evaluate current assumptions of MAP transmission. In vitro analysis of bovine-specific MAP strains revealed a difference in survival of bovine-specific MAP strains but no difference in monocyte-derived macrophages from shedding and non-shedding exposed animals. Molecular analysis of strains of MAP in dam-daughter pairs in a seven-year longitudinal dataset showed that vertical transmission plays a small but identifiable role in MAP transmission. In a metaanalysis of experimental infections of cows with MAP, most calves shed MAP within 12 months of infection. Duration of early shedding and time to late shedding were both dependent on age at exposure, while dose delivered played a role in duration of early shedding only. We integrated early shedding into on-farm models and it played an important role in MAP persistence in herds undergoing intervention attempts. Adding age- and dose-dependent early shedding created a forward feedback loop in which MAP can persist in populations which would not support introduction.