I have used this with iphones outside of Adapt.

What we did was create a user agent that checked for iphone device, and then if so it took our text in the box beside the video and overlaid it as captions on the html5. 

Iphone allows captions through that "English QuickTime 608 CC track" standard. So we just had that generate rather than text in a box beside. 

For other elements like graphics, obviously this is not possible, you just have to build the project so that anything outside the video is supplemental and not necessary in itself.