It is probably a Fano resonance. Check the literature, there are many papers about the topic.

Thanks Ferenc, I don't think plasmonic Fano resonance is the answer. In the extinction spectrum, there is no overlap of spectrum for plasmonic modes that can interact with each other to give interference.

Try SEM or TEM imaging and check their shapes. It is possible that they are not absolutely round?

try to check the size distribution of your nanoparticles and check the theory : how many bands should you have?