Where it says 1kHz, thats just there for an initial generator value. When I did the frequency sweep, it ignores what that generator is set to and sweeps the selected range. That chart sweeps 100Hz to 10kHz, so it includes a datapoint at 400Hz.
I made a minor change and re-ran the circuit, I added the proper part labels and added C60 and R100 to better match the low end response and a load of 10k to represent the input at TR16. Not much changed besides the low frequencies dropping off a little faster.
I can understand aahhhhh showing more power since it is harmonic rich, but I cannot explain a 400Hz sine wave producing more power than a 1kHz sine wave. It would be fun to run a sweeping tone into the radio and follow it to see where 1kHz drops below 400Hz. Could the mic response be to blame?