Revisiting the bulge–halo conspiracy – II. Towards explaining its puzzling dependence on redshift

We carry out a systematic investigation of the total mass density profile of massive (Mstar~3e11 Msun) early-type galaxies and its dependence on redshift, specifically in the range 0<z<1. We start from a large sample of SDSS early-type galaxies with stellar masses and effective radii measured assuming two different profiles, de Vaucouleurs and S\'ersic. We assign dark matter haloes to galaxies via abundance matching relations with standard LCDM profiles and concentrations. We then compute the total, mass-weighted density slope at the effective radius gamma', and study its redshift dependence at fixed stellar mass. We find that a necessary condition to induce an increasingly flatter gamma' at higher redshifts, as suggested by current strong lensing data, is to allow the intrinsic stellar profile of massive galaxies to be S\'ersic and the input S\'ersic index n to vary with redshift approximately as n(z)~(1+z)^(-1). This conclusion holds irrespective of the input Mstar-Mhalo relation, the assumed stellar initial mass function, or even the chosen level of adiabatic contraction in the model. Secondary contributors to the observed redshift evolution of gamma' may come from an increased contribution at higher redshifts of adiabatic contraction and/or bottom-light stellar initial mass functions. The strong lensing selection effects we have simulated seem not to contribute to this effect. A steadily increasing S\'ersic index with cosmic time is supported by independent observations, though it is not yet clear whether cosmological hierarchical models (e.g., mergers) are capable of reproducing such a fast and sharp evolution.