There are also mathematical models of why the number of editors is decreasing that have nothing to do with a decreasing opinion of Wikipedia or a migration elsewhere.  In the beginning, Wikipedia was tabula rasa, and editors could add sections on anything.  Addition, multiplication, trigonometry, basic set theory, etc. all needed _something_, even if it was a very trivial entry.  The level of expertise required to contribute was very low.  Over time, the articles have indeed improved, and the information content has clearly grown.  Those areas with little barrier to entry have been filled out quite extensively.

So, I expect that the data presented in the question is not something to be overly alarmed about.  There is much of the trend that should be expected.  As the requirements for constructively contributing increase, fewer individuals will be able to add useful information or correct improper or awkwardly worded entries.

I think this is true throughout Wikipedia.  The type of editing that is easy these days are more focused on current events.  Other data will natural decrease to some convergence and only be edited as cultural changes make different expositions better, a very slow process.  Different data would be needed to separate this phenomenon from the other causes for edit rate changes.