Dust-to-stellar mass fraction of lensed QGs (W21; Caliendo et al. 2021) as a function of redshift (left) and stellar mass (right), compared to stacked samples (G18 and M21) and local QGs (Lianou et al. 2016). Filled dots show the constraints derived from extended model fits, while open and filled squares correspond to point-source fluxes converted to dust masses using the S16 parameterization and M21 templates, respectively. All symbols are color-coded as a function of either stellar mass (left panel) or redshift (right panel). The right panel also shows the expected average dust fraction of QGs, from a simple model based on the evolution of their mass function (Gobat et al. 2020), assuming that either the initial dust fraction after quenching is a fixed fraction of the main sequence one (solid, dashed, and dotted lines corresponding to z = 1.5, 2, and 2.5, respectively) or constant irrespective of redshift and mass (gray envelope, for the same redshift range). In the left panel, the green curve shows the typical dust fraction of main sequence galaxies (Whitaker et al. 2014), while the gray envelope corresponds to the redshift evolution of the Gobat et al. (2020) model.