How does the spatial heterogeneity of landscapes interact with the adaptive evolution of populations to influence their spreading speed? This question arises in agricultural contexts where a pathogen population spreads in a landscape composed of several types of crops, as well as in epidemiological settings where a virus spreads among individuals with distinct immune profiles. To address it, we introduce an analytical method based on reaction-diffusion models. We focus on spatially periodic environments with two distinct patches, where the dispersing population consists of two specialized morphs, each potentially mutating to the other. We present new formulas for the speed together with criteria for persistence, accounting for both rapidly and slowly varying environments, as well as small and large mutation rates. Altogether, our analytical and numerical results yield a comprehensive understanding of persistence and spreading dynamics. In particular, compared to a situation without mutations or to a single morph spreading in a heterogeneous landscape, the introduction of mutations to a second morph with reverse specialization, while consistently impeding persistence, can significantly increase speed, even if the mutation rate between the two morphs is very small. Additionally, we find that the amplitude of the spatial fragmentation effect is significantly increased in this case. This has implications for agroecology, emphasizing the higher importance of landscape structure in influencing adaptation-driven population dynamics.