2025MNRAS.544.1530C

Some massive, quiescent galaxies at $z>3$ appear to contain considerable numbers of old stars (forming at $z>7$). Works inferring the star formation histories of at least one such galaxy observed with James Webb Space Telescope have suggested that the assembly of so much stellar mass so early may challenge the well-established cold dark matter (CDM) cosmological framework, or else indicate extraordinarily high past star formation efficiencies. However, these studies implicitly assume that all the stars in place at the epoch of observation assembled in situ, i.e. in a single galaxy. In hierarchical models like CDM, massive galaxies assemble following successive mergers of smaller galaxies. Thus, inferences of the growth of stellar mass using the ages of stars within a descendant massive galaxy will be biased. In this paper, I use the TNG100 simulation to quantify this bias across a range of descendant galaxy masses and redshifts, for inferences made for different past epochs. I demonstrate that the assumption of in situ stellar mass assembly can lead to significant biases in inferred stellar mass histories, with historic (i.e. looking back from the epoch of observation) stellar masses overestimated by over an order of magnitude in some cases. The bias increases with increasing halo mass, for inferences made further from the epoch of observation, and with decreasing descendant galaxy redshift. I derive corrections that can be applied to inferred stellar mass histories for more robust comparisons with dark matter halo mass functions at high redshift.