Writing a caption is not unlike constructing a table. Some of the same principles apply. The captions and graphics carry on a succinct, relevant conversation outside of the main text. As such, captions should be reasonably quick for the reader to comprehend. If you end up having too much explanation, you should just put it in the text. Or take the graphic and the discussion or tangent and make it into a sidebar (if that is one of your options).
But you shouldn’t be too brief either. There should be at least one scrap of real information in the caption to justify its existence. Let’s just say “Fruit trees originated in the wild ages ago” with a generic picture of evergreens in the sunlight is not an effective picture–caption pairing. To begin with, ages ago is so general as to be totally useless. But then the reader is forced to stare at the picture to see if there are any fruit trees beneath the hemlocks. (There aren’t any.) There is no conversation. There is no useful gem of information. There is confusion and delay. (I’ve been hanging out with three small Thomas the Tank Engine aficionados.)
A caption should be like a story you tell in a job interview: brief, informative, and appropriate.