To date, the selection of candidate strains for probiotic development in production animals has been largely based upon screens for desired phenotypic traits. However, increasing evidence indicates that the use of host-specific strains may be important, because coevolution with the animal host better prepares a bacterial strain to colonize and succeed in its respective host animal species. This concept was applied to Lactobacillus johnsonii in commercial poultry production because of its previous correlation with enhanced bird performance. Using 204 naturally isolated chicken- and turkey-source L. johnsonii, we demonstrate that there is a strong phylogenetic signal for coevolution with the animal host. These isolates differ phenotypically, even within host source, and these differences can be correlated with certain L. johnsonii phylogenetic clades. In commercial turkey poults, turkey-specific strains with strong in vitro phenotypes performed better early in life than strains lacking those phenotypes. A follow-up performance trial in broiler chickens demonstrated that chicken-specific strains result in better overall bird performance than nonchicken-specific strains. Collectively, this work provides evidence for the impact of host adaptation on a probiotic strain's potential. Furthermore, this top-down approach is useful for screening larger numbers of isolates for probiotic candidates.