I’m not sure exactly what you mean, but I’m going to guess that you are asking how it can be random if there are only small differences among a single species of plants, but very large differences across all of the different plants. That looks like a pattern, not something random. I’ll answer that question, and if I got the question wrong, please let me know as I don’t mean to “straw man” you.
The answer to the question I have just asked myself, (which I hope(?) is the same as the question you are asking me) is that the variations in gene size are random in the individual, but naturally build up over time, so the more related two individuals are, the more likely they are to share DNA, and that includes having the same insertions or deletions. This in turn means that the more related two organisms are, the more likely they are to have the same DNA length.
So, just like you will see a nested hierarchy within different clades related to precise DNA content, you will also see a nested hierarchy (less precise, because the metric we are using is less precise) in terms of similarity of the length of DNA.
You will even be able to do some phylogeny if you want to determine when certain indels likely entered specific genomes, and this would allow you to determine why certain species have the same or different genome sizes and by what amounts.
Does this answer your question, or a part of it? Let me know.