It's an interesting approach, and the 'plaintext'/plant matching looks convincing. Many of the words she derives require omitting letters or adding letters to the anagram text. There go standard frequencies. In some cases one letter stands for several (x=>ver). Add to that the likelihood that the author was poor & uneducated, throw in bad spelling but great knowledge of flora.
It'd be awesome if the solution was going under so many heads for so long.