What matters in the end is the relative gain i.e. comparing antennas under the same conditions. MMANA-GAL includes +3 dB ground gain for perfect ground, so just use the same reference for all antennas. That said, I would argue that comparing a ground plane to a free-space dipole (2.15dBi) is not valid, since GPs only exist with ground. The analysis only makes sense when both antennas are modelled with the same environment.