Just a super chaotic and unstable environment inside a tornado vortex, especially as it traverses varying terrain/surface features, ingests and expels huge amounts of debris, etc. Multiple vortices are a function of a tornado's swirl ratio (basically the balance between the air flowing into the tornado and the updraft), so anything that affects that balance will affect the overall structure. To give you an idea of the typical timeframes, most papers classify long-lived subvortices as any that last longer than 15 seconds. Obviously most are even more brief, like a few seconds or so.
I have seen papers that mention especially persistent vortices lasting maybe a minute or two, but I don't recall much beyond that. Which isn't necessarily to say that it's impossible, of course. My minimally-informed guess is that anything longer than that is probably super rare, if it happens at all. And re: fluctuations in size, I'd assume so? I don't really know for sure but I don't see why that wouldn't be the case, at least with those vortices that last long enough. Pretty much nothing about a tornado is stable usually.