I've had this paper for some time:
The CoQH2/CoQ Ratio Serves as a Sensor of Respiratory Chain Efficiency
It really grabbed me as it was one of the earlier references to supercomplex assembly of electron transport chain components, failure of C57Bl/6 mice to manage this correctly due to a truncated supercomplex assembly protein and the deconstruction of excess complex one when electrons were fed in to the ETC via electron transporting flavoprotein dehydrogenase. So it's a great paper of enormous scope.
But there is more. Dave Speijer has a great discussion of why mitochondrial preparations are so hard to interpret because they are so far away from the in vivo situation and preparation artefacts are massively influential of results. It's in here on page 4110 if you'd like to browse.
Back to the Guarás paper. They got out of the problems of using isolated mitochondria by using intact fibroblast, treating them with a cell permeant dye which becomes fluorescent on exposure to ROS, followed by flow cytometry to assess ROS production. Then they could treat the cells with glucose or fatty acids +/- very low dose rotenone, which blocks RET without having the unacceptable off target effects of higher doses.
This is what they got. Glucose was 5mmol/l and FFAs, mixed types, were supplied bound to albumin at 1000micromol/l. Both quite physiological. Here is what they got (Fig 6, section H), we can ignore the galactose results:
So mixed FFAs produce roughly twice the ROS produced by physiological glucose. The effect is markedly reduced by inhibiting RET through complex I.
Sadly no one has done the experiment to compare saturated fatty acids, MUFA or PUFA on the generation of ROS. Still less to look at the effects of background glucose elevation to represent the immediate post prandial period, with or without insulin. But the basic proof of concept is there.