Interesting Stacking Issue with Sulfer and Oxygen - Any Thoughts?

I collected 90 second frames of R, G, B and 300 second frames of Ha, OIII, and SII. I then stacked them using the default setting of WBPP in PixInsight as I always do. The R, G, B, and Ha came out looking really well, however, the SII and OIII to a lesser extent had a weird stripe from top to bottom. I did some experiments and the results are leaving me a bit confused.

Experiment:

I restacked the SII with no darks, flats, and bias and there was no stripe. I then stacked them with the darks and the stripe was back. Eureka - the darks are somehow the problem! To be sure I restacked them again with only the flats and there was no stripe.

The confusing part is the same darks were also used for the Ha. If the darks were the issue, the Ha should also be affected.

Results:

Figure 1 - SII stacked with darks, flats, and bias 

Figure 2 - SII stacked with no support frames

Figure 3 - SII stacked with flats only 

Figure 4 - SII stacked with darks only 

Figure 5 - OIII stacked with darks, flats, and bias 

Figure 6 - Ha stacked with darks, flats, and bias 

Anyone have any ideas?

Equipment:

Camera: ZWO ASI1600MM-Pro
Telescope: Astro-Tech AT115EDT 115mm Refractor Telescope
Focal Length: 805mm (644mm w/ FR)
f/7
Focal Reducer: 0.8x AstroTech Field Flatterner/Focal Reducer
Mount: Orion Sirius
Filter Wheel: ZWO
EFW 8 x 1.25"
Filter: Antlia Ha, OIII, SII; ZWO L, R, G, B
Focuser: ZWO EAF
Autoguiding: ASI120 Mini attached to an Agena 50mm Guide Scope/ZWO 60mm Guidescope
Exposure: Ha 77 x 300, OIII 68 x 300, SII 71 x 300, R 78 x 90, G 48 x 90, B 49 x 90
Gain: 139
Sensor Temp: 0 C