Very compelling images Mongo, the 200-500 is getting harder and harder to ignore as my "something long" for 2016.
It would definitely have to be one of the main contenders for the “something long” category. In Mongo’s review he goes through the tale of misery he had with this lens from purchase until recently. It had been back to Nikon twice and Mongo started to seriously dislike it and Nikon. Mongo has not tried it properly since its last return but can tell from just some quick shots that it is noticeably better and should now be fine (ie give slightly better results than those posted here).
Have tried the Tamron 150-600mm and found it to be quite reasonably good. However, Mongo would now choose the Nikon 200-500mm (provided you get a good copy).
Bought new 200-400mm f4 VR I and had it for 3 years. Had it back to Nikon 4 times. Poor with converters and poor over distance. Never happy with it and sold it. Tried a friend’s copy of the 200-400mm f4 and found it much better and what Mongo would have expected the lens to perform like. His works better with converters than Mongo’s copy of the 200-400mm f4. However, still doubtful about distance shots. It is about 1 kilogram heavier than the 200-500mm, a little larger and about 4 to 5 times the price.
The 200-500mm on the other hand, seems to handle distance shots better and works with converters. Works with 1.4 EII on D800E and D4s. Will AF with 1.7EII on D4s but not D800E. However, for practical reasons, it may be too slow at stops beyond f8 for other than virtually stationary subjects.
Mongo now only uses the 600 f4 in specific and convenient circumstances.
Knowing what Mongo knows now, he would pick a good copy 200-500 over a 200-400 taking most things into consideration.
...... The first birds are sharp with nice bokeh (although different on images #5 and #7, different stops ?)
Mongo checked the data on 5 and 7 Rosko and it is identical to most of the others. However, the distance to subject is different as is the distance from subject to its background. Mongo was quite close and even small differences at that range may exaggerate differences in the end product. These are probably the variables giving rise to the differences you have noticed.