Why isn't there a dip at driver resonant frequecy (Fo) on frequency response graphs?

bg17_8_fs.gif

Here is the frequency response graph of the Visaton's BG 17 driver, whose Fo = 120 Hz. The response rolls off from 200 Hz to 120 Hz which is understandable because of the rise in impedance.

However, shouldn't the response rises from 120 Hz to 75 Hz since impedance decreases from 22 Ohm to 10 Ohm?
 
Driver resonance is defined as a frequency and Q value. If the driver Q is .5 then the output at resonance should theoretically be .5 of the value at midband frequencies. The graph response can also be influenced by acoustic variables and other response anomalies so you need to know the exact conditions used in the measurements.
If you have Winisd or other box sim software you can play with closed box sims and see the output change at the box resonant frequency depending on Qt.