Ok time to cut through a little of the confusion:
Technically there should be no difference in exposure when using a smaller aperture in one of the automatic modes as the camera should vary the other parameters to compensate. You might see a difference in the extreme corners due to vignetting at wide apertures, but this would make the small aperture shot brighter.
In this case shutter speed decreased in the small aperture shot, meaning in this case metering is not [entirely] to blame. There was a 1/3 stop discrepancy between shots however.
Lets dig into the details a little...
The recorded parameters for the two shots were
ISO was 100 for both. Exposure is proportional to shutter time over f-number squared. This gives a 1/3 stop disparity in the camera settings (this is probably a result of the camera's meter, however I'd be willing to forgive this minor variation as the composition did change a bit).
However if you look at the images, taking a reading from the highlighted side of the building bottom centre gives:
RGB = [134, 138, 135]
RGB = [74, 73, 72]
assuming a roughly linear tonecurve (unless you shot RAW we can't assume otherwise) this gives a 0.9 stop difference. So clearly there is something else going on here.
f/18 isn't exactly a third of a stop down from f/16, it's just rounded to 18 for convenience. In any case we don't know how accurately the camera can control the aperture when it's that small. Likewise you would expect some variation in shutter speed. As the shutter gets worn the curtains can get out of sync, giving you considerable variation, especially at high speeds.
If we say the actual settings used were
Then suddenly we get a 0.9 stop difference in the expected exposures. There other factors that could be involved
Normally these things cancel out, but if they all line up one way you can get sizeable swings in exposure.
The lesson to bring from this is that it's probably normal, if it happens consistently it may well be something out of calibration, but I wouldn't worry about it, the camera's meter is not infallible, a better approach  with digital is just to check the histogram after each shot.