Model it as a dielectric loss. Mechanically, what's happening is myriad streamers discharge the strong electric field from the electrode into surrounding space, effectively growing the diameter of the electrode until field strength is below breakdown and further sparking/corona emission ceases. Repeat ad nauseum at very fine time scales (breakdown occurs in fractions of a ns) and take the average over human time scales (ms, say) and you have the brushlike shape typical of RF corona discharge.
Each individual discharge might be some ohms, but it's only discharging some ~pC of nearby air, the reactance of which dominates. But the discharge only lasts for a tiny instant (give or take displacement currents keeping it hot and ionized; this is more likely, close to the electrode, than far off at the periphery, where the discharges are constantly breaking out, recombining, and re-breaking), so although its instantaneous resistance can be small, the time-averaged resistance is much larger.
It's a similar effect as in a switched-capacitor circuit, or a mixer or analog switch circuit run at low duty cycle, where the switch resistance becomes divided by the duty cycle of the on-time.
I'm not familiar with typical figures, or how much goes into dissipating it, but my rough guess from Tesla coil stuff is the loss tangent is around 10 or 20%.
This must be accounted for in parallel with the natural capacitance of the secondary. If the discharge dominates, it might well be the overall loaded Q is comparable, i.e. 10 ish, but more likely it's less than half the total, i.e. loaded Q of 20 to 60 is a reasonable starting point.
Speaking of Tesla coils, this means the unloaded Q factor of a coil, and the precision of tuning, isn't so important, because as soon as some breakdown gets going, the Q tanks, and bandwidth widens. In an SSTC where you're pumping energy into the system during a tone burst, you tend to see the amplitude grow or decline more like linearly rather than exponentially.
This also means the resonant frequency shifts downward, which is like the discharge making itself into a conductive shell and therefore adding as much capacitance as such a shell would provide.
Tim