This depends on quantitative analys of variance by using Iris Fisher's approach. In brief morphology is measured by shape analysis as in http://aob.oxfordjournals.org/content/93/2/149.full.pdf and stats described in https://en.wikipedia.org/wiki/Iris_flower_data_set applied to venation or leaf shape parameters
I think that the classification of these species via classical methods such as venation may be an accurate, because some time the stress physiology can change the morphology of the plant.