Factors that drive species richness over space and time are still poorly understood and are often context specific. Identifying these drivers for ant diversity has become particularly relevant within the context of contemporary global change events. We report on a long-term bi-annual (wet and dry seasons), standardized sampling of epigeal ants over a five year period on the mesic and arid aspects of an inselberg (Soutpansberg Mountain Range) in the tropics of Africa. We detail seasonal, annual and long-term trends of species density, test the relative contribution of geometric constraints, energy, available area, climate, local environmental variables, time, and space in explaining ant species density patterns through Generalized Linear Mixed Models (GLMM) where replicates were included as random factors to account for temporal pseudo-replication. Seasonal patterns were very variable and we found evidence of decreased seasonal variation in species density with increased elevation. The extent and significance of a decrease in species density with increased elevation varied with season. Annual patterns point to an increase in ant diversity over time. Ant density patterns were positively correlated with mean monthly temperature but geometric constraints dominated model performance while soil characteristics were minor correlates. These drivers and correlates accounted for all the spatio-temporal variability in the database. Ant diversity was therefore mainly determined by geometric constraints and temperature while soil characteristics (clay and carbon content) accounted for smaller but significant amounts of variation. This study documents the role of season, elevation and their interaction in affecting ant species densities while highlighting the importance of neutral processes and temperature in driving these patterns.