Matching the number counts and redshift distribution of submillimeter
galaxies (SMGs) without invoking modifications to the initial mass function
(IMF) has proved challenging for semi-analytic models (SAMs) of galaxy
formation. We adopt a previously developed SAM that is constrained to match the
$z = 0$ galaxy stellar mass function and implement three simple prescriptions
to predict the submillimeter flux densities of the model galaxies; two depend
solely on star formation rate, whereas the other also depends on the dust mass.
By comparing the predictions of the models, we find that taking into account
the dust mass, which affects the dust temperature and thus influences the
far-infrared spectral energy distribution, is crucial for matching the number
counts and redshift distribution of SMGs. Moreover, despite using a standard
IMF, our model can match the observed SMG number counts and redshift
distribution reasonably well, which contradicts the conclusions of some
previous studies that a top-heavy IMF is needed to match these observations.
Our results suggest that differences in the treatments of processes such as
stellar feedback may enable some SAMs to reproduce the properties of the SMG
population without IMF variation.