Obvious signs of fatigue cracking, which looks to have initiated at about 7 o'clock in the picture on the outside of the shaft. The progression and origin is consistent with primarily bending load, as is the final failure plane (perpendicular to shaft).
What I'm less sure of is what caused that loading - imbalance is the most likely source of bending load to me, but the only cyclic change is in magnitude and that only with change in compressor shaft speed, which means relatively few cycles of fairly low magnitude - for that to cause failure it would probably have to be significant imbalance and/or significantly higher than design speed. That being said, I am not sure what compressor surging looks like in term of shaft load - I would expect vibration and torsional loading, but I'm not sure if you get enough pressure imbalance across the wheel to cause significant bending load, and with rapid enough change to cause cyclic loading. Also, I would be surprised to see enough load to cause shaft fatigue without also having visible compressor blade damage.