Actually, it is both.
A very simple bit of physics. The speed of sound changes due to the density of the air. Which means altitude, temperature, and humidity.
The absorption of sound by air (that is, the way acoustic energy falls off with distance in addition to the geometry of inverse-square dispersion), also changes with humidity, and this is frequency-dependent.
But let's take the first. Assume you EQ'd your house until the speaker response was nice and flat. Well, when the audience walks in, you've got a problem.
What is room EQ? Two things, primarily. There's the response curve of the speakers and amps. And there's the response of the room. This latter is due largely to multipath reflection. Sound comes from a speaker. It travels across the audience, hits the back wall, reflects back over that same audience. Where the two paths meet, parts of the waveform will be in phase and combine, others will be out of phase and destructively interfere. Thus, peaks and valleys in the response -- the most prominent of which can be related directly back to the dimensions of the room itself.
So what happens if the speed of sound changes? The two paths increase or decrease in time. But they do so as a proportion, as a percentage, of their previous time. Which means the phase relationship changes. Which means the peaks and valleys are not in the same place.
Which means that 2 kHz peak you notched out of the mains is no longer there, and you are taking a 2 kHz cut in the middle of the sound. And there's a new, uncorrected peak at 2.5 kHz. Or maybe 1.5 kHz -- depending on how the speed of sound changes, and how it relates mathematically to the critical dimensions of the room.
Oh, yes. And the temperature and humidity does change. Not only does each human body in the audience pump 100 watts of heat into the air, each gives off water vapor, raising the humidity. And that's not the worst of it. The graph changes radically between 10% and 30% humidity. Which means starting with a dry room and introducing warm sweaty bodies will have a much greater effect on the speed of sound in your venue than you might expect.
And that's before we even get to considering the frequency-dependent attenuation that also takes place over that critical humidity curve. Which -- just to put icing on the cake -- is most prominent at 12.5 kHz; right in the middle of our frequencies of interest.