From the chemodynamical properties of tidal debris in the Milky Way (MW), it has been inferred that disrupted dwarf satellites had different chemical abundances at their time of accretion compared with similar-mass dwarf satellites which survive at present day. Specifically, disrupted satellites appear to have had lower [Fe/H] and higher [Mg/Fe] at fixed stellar mass than the surviving ones. In a recent study (Grimozzi, Font & De Rossi 2024), we have used the ARTEMIS simulations to investigate this problem, and determine the evolution of chemical abundances (e.g., the stellar mass-metallicity relation, MZR) with redshift. We have found a strong correlation between the scatter in the MZR of the disrupted dwarfs and their accretion redshift (zacc), as well as with their cold gas fractions at accretion. The slopes of the MZRs of disrupted dwarf satellites are fairly similar at different accretion redshifts, and are comparable with the MZR slope of surviving satellites in the MW today (≈ 0.32). This findings constrain some of the physical processes that regulate the chemical enrichment of dwarf galaxies (for example, the stellar feedback). The simulations also predict strong correlations between averaged properties of the disrupted dwarf populations, such as between «zacc», «[Fe/H]» and «[Mg/Fe]»), which suggests that the chemical abundances of the entire disrupted dwarf population can be used to constrain the merger history of its host. More specifically for the MW, the ARTEMIS simulations predict that the bulk of the disrupted population was accreted at «zacc» ≈ 2, to match the averaged «[Fe/H]» and «[Mg/Fe]». More broadly, our results suggest that one can gain an insight into the formation histories of other MW ‘analogues’, such as M31 or other massive galaxies nearby, provided that chemical abundances ([Fe/H] and [alpha/Fe]) of their debris from disrupted satellites become available.