Utilization of molecular phylogenetic information over the past decade has resulted in clarification of the position of most angiosperm orders, as demonstrated by the classification produced by the Angiosperm Phylogeny Group. A group of 11 families was listed at the beginning of the APG classification because "they belong neither in any of the phylogenetically basal orders at the beginning nor in the monocots or eudicots." The phylogenetic positions of most of these families (e.g. Amborellaceae, Nymphaeaceae, Winteraceae, etc.) have since been clarified, however, the positions of two holoparasitic families, Hydnoraceae and Rafflesiaceae, have remained enigmatic. To address the question of phylogenetic position of Hydnoraceae, nuclear SSU and LSU rDNA and mitochondrial atp1 and matR sequences were obtained for Hydnora and Prosopanche. These sequences were used in a combined analysis that included the above four genes as well as chloroplast rbcL and atp (these plastid genes are mising in Hydnoraceae and were hence coded as missing). Three data sets were analyzed using maximum parsimony: 1) 3 genes/461 taxa; 2) 5 genes/77 taxa; and 3) 6 genes/38 taxa. All of these analyses support the monophyly of Hydnoraceae and the association of that clade with the "paleoherb" families Aristolochiaceae and Lactoridaceae. These results suggests that either Aristolochiaceae are paraphyletic or that Hydnoraceae should be included (along with Lactoridaceae) in a more broadly defined family Aristolochiaceae. In contrast to most traditional classifications, molecular phylogenetic analyses do not suggest a close relationship between Hydnoraceae and Rafflesiaceae s. lat.

Key words: Aristolochiaceae, Hydnora, molecular phylogeny, paleoherb, Prosopanche