Voltages only vary a bit under uniform illumination as brightness and temperature vary.
But if a shadow covers a single PV cell, even though Voc will be virtually unchanged, Vmp will drop according to one diode-bypassed section of PV panel (typically 1/3 or 1/2 of panel voltage).
When current is drawn, that section of panel puts out nothing so only Vmp and Imp of remaining sections of panels contribute.
If 1/2 the area of one cell is shaded, then you get 1/2 Imp at Vmp of the string. MPPT has to be smart enough to pull voltage lower and get higher current of 1.0 Imp, bypassing that section.
Similar deal if two strings in parallel and some panels of one string shaded. There may be two peaks in the watts/volts curve and MPPT ideally finds the higher one, but that depends on firmware.
Very close /above vmp I assume? Not under
Definitely can be under Vmp. Vmp was determined with laboratory conditions, flashed with 1 "sun" but panel at 25 degrees C panel temperature.
Real world conditions are typically warmer (could be colder) and typically less light (could be more).
So voltage delivering most power could be higher or lower than Vmp. You ought to design your system to work above the minimum MPPT voltage range of your controller; otherwise, it will sit at that minimum and get less than maximum power from panel.
136Vmp for 4 panels, 120V minimum mppt.
I think that will usually work, look up the temperature coefficient of Vmp in data sheet of panel.
It is less clearly documented how Vmp changes with illumination. Data sheet may plot curves, but if shows only as I/V not W/V, difficult to tell.
If a shadow ever falls on a panel, power output may drop to zero. I've done that, where I reconfigured some 12V panels for 16s2p instead of 24s1p. Didn't take much shade to stop producing.
You'd be better off with 5 panels in series.