@MissDan1995 - if you use welch algorithm without the tapering - you don’t get spikiness in the rolloff region.
But it really looks like everything outside of 4-7 Hz is not really analyzable since it appears to be in the filter region. I think the “spikes” in your filter region due to the tapering/windowing used in the multitaper psd calculation (see image below which looks similar to your ) -
But ultimately - I think you only have theta signal in your data. And anything in that “spikey” region has been filtered out. You can confirm by looking at your raw data. Below is the second epoch of your data in a 10s window - the data looks very slow and would be consistent with theta signal that you see in the PSD. I think that you have band-filtered the data at some point in your processing. I would rerun the preprocessing steps or just look at it on the raw unprocessed data.
Hope that helps.
-Jeff