The scales, graticules, rhumb lines, and compasses all conflict with each other or otherwise just don't make sense. The one thing that does work is that the large and small scale maps do seem to have corresponding scales. The middle map though doesn't match up. You also have a compass rose and a set of rhumb lines on the small scale map, and then several other rosettes of rhumb lines in other parts of the map, that don't line up with the first set. The large and small scale maps also have rhumb lines and a compass rose in the case of the small scale map, which all means they are bearing preserving, the middle map seems to be using a graticule from an equatorial aspect of an azimuthal projection or a derived projection like Aitoff or Hammer, which are decidedly not bearing preserving. Notice that the meridians aren't parallel lines, that means that north and south are varying. (A note, even if they are parallel lines, that just means north-south is preserved, some projections preserve that while still distorting directions like "northeast").
Even ignoring the mismatch between the maps, the graticule doesn't make sense. As I said, I'm not entirely sure what projection it is, but it's not a projection that would be used for that map. It's hard to say where exactly the map is on the globe or what its true extent or orientation are given the inconsistencies, but it would probably make most sense to use a conic projection, or, if you were going to use an azimuthal projection, one which is actually centred on the map.
Also, if the small scale map really is bearing preserving over such a large extent, then it has to be in Normal Mercator projection, which significantly distorts scale, and so shouldn't have a scale bar.
You seem to be doing curved text using Envelope Deformation. This gives the text a sort of squashed or smeared look. "Text on Path" is a better way to do this. It will shift and rotate each glyph to fit on the curve without distorting it.
Otherwise very pretty, and I like that the different maps show a degree of error between them rather than computer precise replication.