In hydrodynamic simulations, prevailing subgrid chemical-evolution models often use a single, "IMF-averaged" supernova yield, ignoring variations in elemental abundance ratios (particularly [$\alpha$/Fe]) in the ejecta of higher- and lower-mass supernova progenitors within a stellar population. To understand the impact of this simplification and understand the impact of more explicit models, we run FIRE simulations of a dwarf galaxy $(M_\star($z = 0$) \sim 10^6 M_\odot)$ using nucleosynthetic yields from the NuGrid database that depend on the stellar progenitor mass and metallicity. While NuGrid exhibits lower aggregate $\alpha$-element production than default-FIRE yields, we find that its explicit mass dependence substantially widens the intrinsic scatter in the simulated [Fe/H]-[$\alpha$/Fe] -- a phenomenon potentially visible in recent observations of dwarf galaxies.

We explore the effect of including progenitor mass- and metallicity-dependent yields, supernova rates and energetics on variations in elemental abundance ratios (particularly [α/Fe]) in dwarf galaxies. To understand how the scatter and overall trends in [α/Fe] are affected by including variable metal yields from a discretely sampled initial mass function, we run FIRE simulations of a dwarf galaxy (M★(z = 0) ∼ 106 M☉) using nucleosynthetic yields from the NuGrid data base that depend on the stellar progenitor mass and metallicity. While NuGrid exhibits lower aggregate α-element production than default FIRE yields, we find that its explicit mass dependence, even when including turbulent metal diffusion, substantially widens the intrinsic scatter in the simulated [Fe/H]-[α/Fe] – a phenomenon visible in some observations of dwarf galaxies.