The 40kHz is from the axis. The fundamental appears at about 0.4 on the scale, which is per 105 Hz, so that is 0.4 x 105 Hz, or 40 kHz.
As for the rest, it's not at all clear to me. They don't seem to mention the large 210 kHz signal which is the same size as the fundamental. Maybe this is an alias? The symmetry makes me think this.
The second and third harmonics appear in b,c & d, but as to what causes them there's no indication.
Edit: I haven't read the paper, though they appear to be claiming the harmonics are due to non-linear effects in the liquid. I would want to check carefully that it's not non-linear effects in the driver.