# Relationship between conjugate poles and resonance peak in frequency response

I have two root locus of two closed loop transfer fuctions, which differ just by a different value of the gain. I have that the one with a larger gain has a reasonance peak bigger in the frequency responce and the one with the smaller gain as almost no overshoot. I also plotted the root locus to see this:

for the transfer function with a larger gain I have

while for the one with smaller gain:

I can see that the upper one has two poles complex conjugate, so I guess this is the reason for the reasonance peak.

My question is: Why the transfer fuction with smaller gain has only poles on the real axis, while the one with the larger gain has two conjugate poles?

I have also tried tuning values of the gain, and what I get is that as I increase the gain the distance between the conjugate poles increases.

Can somebody please explain this behaviour? And what is the relationship between complex conjugate poles and resonance peak?