Top: long-term time evolution of the midplane temperature deviation from the background power law for the dynamical model MSIG2000. After about t = 700 orbits, we see that the first and the second bump merge together. We can also spot an additional bump at ≈20 au. The dashed green line corresponds to the final state of the MFID model in Fig. 6, for comparison. Bottom: the evolution of the radial temperature gradient in time similar to Fig. 6. Here, we see that the innermost bump is a local maximum, producing a temperature gradient inversion, while the other two shadows produce slope changes in the profile.