Until the advent of molecular phylogeny, plant taxonomists had no clue where Cephalotus fit into the tree of life. The most common guesses were that it was related to Sarracenia and Nepenthes (as if they are related!) or to the Saxifragales where Drosera was once thought to reside.
Thanks to DNA sequencing we now know Cephalotus is in the order Oxalidales (oxalis) and not related to any other carnivorous plant. It ties in at the root of the Oxalidales and has as sisters 5 families with over 2200 species in 63 genera, most of which you have never heard of. The family Cephalotaceae has one genus with one species. For what it is worth, the closest outgroup orders are the Cucurbitales (cucumber, begonia) and Fagales (beech, walnut, and she-oak). Does a family tree like that fit the definition of orphan?
Genetic relationships of the plant families in the order Oxalidales based on chloroplast ribulose bisphosphate carboxylase large subunit (rbcL) sequences in the NCBI taxonomy database. See NCBI Blast Tree in a popup window.
The Elaeocarpaceae, Cunoniaceae, and Connaraceae are generally evergreen shrubs found in the southern hemisphere. The Brunelliaceae which was thought to be closest to Cephalotus is a group of evergreen shrubs in central America. The Oxalidaceae consists mostly of herbaceous plants in the genus Oxalis but also contains some shrubs and small trees in other genera. In the rbcL tree none of these families are any closer to Cephalotus than any other. Of these "relatives", the species in the Oxalidaceae and Connaraceae have compound leaves. The species in the other families in the Oxalidales tend to have simple leaves. The same mixed situation exists in the outgroup orders. I doubt very much any additional DNA analysis will help us understand the phylogeny of Cephalotus unless we find an actual close relative. This means any discussion about this species needs to revert to good old fashioned comparative plant development.
There has been a long discussion in the literature about how the pitchers of Cephalotus are constructed. Plant anatomists have been staring at and slicing and dicing the pitchers since the middle of the 19th century. On a general inspection the pitchers are as elaborate as any Nepenthes pitcher. But more detailed inspection showed that the pitchers are not your folded simple leaf like the other pitcher plants. Lloyd (1942) and DeGreef (1990 (
)) summarize the studies demonstrating this and discuss the results under the assumption Cephalotus is a member of the Saxifragales and proto-Cephalotus had a peltate saxifrage-type leaf.
Evidence that Cephalotus is derived from a peltate leaf is the arrangement of vascular bundles in the petiole. In a typical simple or flat leaf the bundles are arranged in an arc. In Cephalotus they are arranged in a circle exactly like Oxalis leaves. Curiously you can also take an Oxalis leaf and easily fold it into a Cephalotus-like pitcher. The lid would be one leaflet. The leaf ridges that act as prey guides are at the seams between the other leaflets and sub-leaflets. This does not prove Cephalotus evolved from Oxalis. It demonstrates how Oxalis and Cephalotus have very similar leaf morphology.
Probably the biggest impediment to understanding Cephalotus pitchers is the simple leaves the plant produces seasonally. DeGreef points out these leaves are unifacial leaves rather than normal bifacial leaves. They can not be representative of ancestral leaves. The simple leaves are produced during the winter when presumably there are not enough prey available to make production of pitchers advantageous. The occasional aberrant leaves we see more than likely happen when the development pathway to produce pitchers kicks in after the leaf has started to grow and it is too late for that leaf to make proper pitchers. The structure of the aberrant leaves is totally consistent with the peltate nature of the leaf and an ancestor with oxalis-like leaves.
-- John Brittnacher
Aberrant Cephalotus leaf. Photo by RL7836.
Aberrant unifacial winter leaf.