He is open to and appreciative of guidance. One thing I've personally done is taken on the role harmony breakdown guy, and I make mp3s with harmonies broken out. Make one for him that repeats just his line and one with the other lines without his, so he can practice separately (to hopefully get some muscle memory and pitch persistence) and in context his line missing.
An example that I did specifically for him
" His voice is naturally baritone and he says he can't sing falsetto at all "Nothing comes out!" which is limiting to say the least for rock and pop backing vocals...all those great baritone parts."
When I read the above, I already had an idea of what was happening, and then when I listened to your GarageBand practice guide (great work by the way), I was confirmed.
If the guy is a natural baritone, your "low harmony" in the GB YouTube, is
way too high. You have middle C and D , It would be natural for him to sing if it was a
full octave lower. Baritone voices crack at middle C.
That sound creates a big space at the bottom between the 3 voices, so then...
Take "high harmony" and bring it down one full octave (G3 and F3) to give to the baritone, then have the "low harmony" be the new "high harmony", for that tenor singer (D4 and C4). The total will have an open voiced chord, a Drop 2 voicing of your current arrangement.
This problem seems to be to be a part writing / arrangement problem at core.
Here is a big issue.
On albums, often the same person, lead singer, recorded the harmonies to themselves, and so with the same voice have tight harmonies with close position chords in all the same vocal range. But with different singers, baritone, tenor, alto.. you need bigger spacing between the parts.
Your vocal parts aren't good for your group! They are too tight and close together, which only works if all the singers are naturally the same vocal type (ie tenor), which according to your OP is not the case.
Think I it this way.
If an alto saxophonist recorded a song, and did the lead and two close position lower harmonies all in the high alto range, and then he plays live with a tenor and baritone saxophonist, it would be silly to expect the baritone to sound good playing those alto harmonies at the pitch recorded. It has to be re-arranged to fit the ensemble.
I suggest making Drop 2 voicing of your above arrangement and finding out how that goes. Then continue with drop 2 as I suspect it will work better.
instead of
low to high : D G Bb to C F A; you change it to. G D Bb to F C A. Close position triads in 2nd inversion become Drop 2 voiced triads in root position.
His notes should mainly be in the range of A2 to A3 (110Hz to 220Hz).
That will help a lot. IMO, your arrangements are setting him up to fail.
BTW, it doesn't matter what the original recording harmony voicings are, arrange the song so your group sounds as amazing as possible.