If the same settings and apertures were used for both runs, this should of course not happen. If you'd like for us to investigate further, we'd need more information and data.
First, you'd need be running AIJ 5.4.0 and then start multi-aperture and create the "bad" plot you posted above. Then close Multi-Plot Main (which will close the other multi-plot windows also).
Next, reopen Multi-Plot and go to Multi-Plot Main > File > "Save all preferences to backup file". Note where you are saving the backup file.
We'd also need access to your calibrated image stack through some service like google drive, dropbox, etc.
You can send the link to your files and the backup AIJ_Prefs file to my email address at the top of the main user forum page.
Here are the backup file paths for both releases.
AIJ 5.3.3 : /home/fred/.astroimagej/AIJ_Prefs_userbackup.txt
AIJ 5.4.0 : /home/fred/.astroimagej/AIJ_Prefs_userbackup.txt
You'd soon get the swiss transfer link to the data set I've used on you email adress.
Great! If there is something you figured out that was going wrong and want to share, please do here. It might help another user out of a similar problem.