Christian,
I don't know if it counts for much (considering the South Africal variability) but I grow 3 different clones which in my opinion are D. collinsiae. 2 of the forms have an upright habit, although they are not caulescent. Both upright forms have tetrete petioles with very small marginal hairs. The third form is prostrate, and the petioles are more elliptical in cross section. The flowers of both types are identical.
Worse yet! I have another plant, which I think is a result of uncontrolled fertilization in my collection between D. nidiformis and D. dielsiana and the flowers are very similar to D. collinsiae! I suppose it could be D. burkeana x D. nidiformis as well since both were flowering, but it could *not* have been D. burkeana x D. madagascariensis (which is the supposed parentage of D. collinsiae) since I have yet to flower D. madagascariensis! The cross looks nothing like any D. dielsiana x D. nidiformis hybrids I have seen, so what is it? 80% of me says it is a form of D. collinsiae, but how can this be possible?
South African plants are a headache! It's like tallying up some score sheet. In this case, what is there to call the plant other than D. collinsiae? Nothing seems "pure" in South African Drosera. You just have to trust your guts, lol.
In this particular case, the tetrete petiole speaks strongly to me of D. collinsiae, if you grant D. burkeana as a likely parent of D. collinsiae as many do. So, in my own mind, I am defining D. collinsiae now more by the nature of the cross section of the petiole. Unless the progenitor of D. collinisiae is extinct, then there is no other choice from South Africa for the parent except D. burkeana which is the only species to have tetrete petioles. Since whatever my hybrid is so closely mimics D. collinsiae, I feel the elliptical cross section must now exclude a plant from this determination.
Of course, I have a limited experience having done no herbarium studies, so this is probably just my silly opinion, lol. I guess this will be another topic of discussion with Robert in the near future. It may well be the variation across the range of D. collinisae includes every imaginable degree of gradiation in the petiole cross section. It wouldn't surprise me in the least.