Chemical bonding to oxide surfaces is often dominated by surface defects, but their nature remains elusive. Calorimetric measurements of Ca and Li adsorption energies on MgO(100) and ion-damaged MgO(100), when combined with density functional theory (DFT) calculations and kinetic modeling, are shown to be a powerful way to assess the nature of the defect sites on oxide surfaces and their lateral distributions. While ion sputtering causes a strong increase in the initial adsorption energy for Li on MgO(100) at 300 K, the initial adsorption energy for Ca is independent of the extent of sputtering. This result and the measured coverage dependence of the adsorption energies of Ca and Li on MgO(100) surfaces with similar to 5, 12, and 25% defects were simulated with a kinetic model based on DFT input regarding site binding energies and adatom migration barriers. Reproducing the experimental results required models with distinct probability distributions of local defect concentrations for the differing extents of ion damage. A key difference between Li and Ca revealed by DFT and necessary to reproduce their differing adsorption energy versus ion damage measurements is the much greater tendency for a diffusing Li adatom to remain locked in place once it reaches a terrace site neighboring an occupied step or kink site, thus nucleating a 2D island on a terrace. In contrast, Ca adatoms thermally diffuse from such sites quickly, to seek out the remaining defect sites. The model also reproduces the measured Li and Ca film morphology seen by ion scattering spectroscopy.