At least in healthy individuals, delivery of O2 via muscle blood flow is very tightly linked to muscle O2 demand - so tightly, in fact, that it is still not entirely clear which limits the rate of increase in muscle O2 uptake at the onset of exercise (or when intensity varies). Given this, you would expect that muscle oxygenation would mostly reflect recruitment, and in fact it has been shown that NIRS data and EMG data track each very closely during cycling exercise.

As an example of this, consider the workout that I did this morning, which consisted of 6 x 5 min intervals @ 90-95% of estimated VO2max. As you can see in the upper chart, muscle O2 content (in volume %) was higher in my left quad at lower power outputs, but lower at higher power outputs. I pedal asymmetrically, however, so once you take this fact (and the lag in changes in muscle O2 content in response to changes in power) into consideration, i.e., by plotting muscle O2 content for each leg against the GPR of that leg, the difference between my two legs almost completely disappears (bottom chart).