Phylogenetic Relationships of Parasitic Flowering Plants

The number of times parasitism has evolved in flowering plants (angiosperms) has long been debated. The closest non-parasitic relatives for a number of lineages have been known for some time: Cassytha with Lauraceae, Lennoaceae with Boraginaceae (or Ehretiaceae), Orobanchaceae with "Scrophulariaceae" in the traditional sense, and Cuscuta with Convolvulaceae. 

For others, particularly the holoparasites in the traditionally recognized families Hydnoraceae, Balanophoraceae, and Rafflesiaceae, placement among photosynthetic angiosperms has been difficult. For this reason, traditional classifications were often conflicted among different workers and even in different treatments by the same worker. Relationships between parasitic and nonparasitic angiosperms have been greatly clarified through DNA sequencing and molecular phylogenetic analyses, although not without some "bumps in the road." One molecular evolutionary phenomenon that has made phylogenetic work on the holoparasites less than straightforward has been horizontal gene transfer, particularly of mitochondrial genes. This has generated conflicts between gene trees derived from the different subcellular genomes. A good example of this is the paper by Barkman et al. (2007) who used molecular methods but could not given an exact number for the origins of parasitism ("at least 11").  Currently, all clades of parasitic plants have been placed on the global angiosperm phylogenetic tree.  As described below, parasitism has arisen independently in angiosperms 12 times. 

In terms of trophic modes, 9 lineages (clades, families) are composed entirely of holoparasites (family names enclosed in solid line rectangles). Only two lineages (Cassytha and Krameriaceae) contain just hemiparasites.  In two families, Convolvulaceae (Cuscuta) and Orobanchaceae, both hemi- and holoparasites be found (family names enclosed in dashed line rectangles).  Only the order Santalales has more than one family of parasitic plants.

Click on the family names to go to those pages

1. Laurales. Cassytha, the sole parasitic member of the large family Lauraceae, is uniquivocally assigned to this family based on morphological and molecular data (Rohwer & Rudolph 2005).  Its superficial resemblance to Cuscuta is remarkable and an excellent example of convergent evolution.

2. Piperales. Molecular data were used to place Hydnoraceae with Aristolochiaceae s. lat. (Nickrent et al. 2002), however, the exact topology of the component families of this order (Aristolochiaceae, Hydnoraceae, Lactoridaceae, Piperaceae and Saururaceae) was not determined. The analysis by Nickrent (2005, IBC abstract) suggested Hydnoraceae was most closely related to Aristolochiaceae. This result was confirmed by more recent work (Naumann et al. 2013). See bottom of the Hydnoraceae page HERE for a discussion of why this family is being recognized here instead of lumping into Aristolochiaceae (as was done by APG III).

3. Saxifragales. Although previous classified with Balanophoraceae, the family Cynomoriaceae has been shown to be a component of Saxifragales (Nickrent et al. 2005).  For a more up-to-date and complete discussion of this finding, see the Supplement to Su et al. (2015) HERE.  This placement within Saxifralaes was confirmed by Bellot et al. (2016).

4. Zygophyllales. The APG2 classification considered Krameriaceae as an acceptable monophyletic alternative to Zygophyllaceae. In APG3  the two families were considered separate. This order is sister to the Fabid (previously Rosid I) clade.

5. Rafflesiaceae. This family, considered here in the strict sense (including Rafflesia, Rhizanthes, and Sapria, i.e., the "large-flowered clade") was placed with Malpighiales by Barkman et al. (2004) using mitochondrial matR gene sequences. This position was confirmed by Nickrent et al. (2004) using both nuclear SSU rDNA as well as mitochondrial sequence data. Placement within the order shows that Rafflesiaceae is near Euphorbiaceae (Davis et al. 2007).

6. Apodanthaceae. Traditionally placed in Rafflesiaceae, the "small-flowered clade" is composed of Apodanthes, Berlinianche, and Pilostyles. Mitochondrial matR and nuclear SSU rDNA data indicated either a relationship with Malvales or Cucurbitales (Nickrent et al. 2004). Additional sequencing and analyses indicated that this family is part of Cucurbitales. That result was confirmed by Filipowicz & Renner (2010).

7. Cytinaceae. Traditionally placed in Rafflesiaceae, the"inflorescence clade" is composed of Cytinus and Bdallophyton. Mitochondrial matR and nuclear SSU rDNA both strongly support a position for this family in Malvales (Nickrent et al. 2004).

8. Santalales. The analyses conducted by Soltis et al. (2000) resolved the sandalwood order as monophyletic, but this clade was part of a large polytomy among the core eudicots. The number of taxa involved in that polytomy was quite high, involving caryophyllids, rosids and asterids. A molecular analysis using complete chloroplast genomes from over 80 angiosperms (Moore et al. 2010) indicated that Santalales is sister to Caryophyllales and asterids (as shown in the above tree).
    The number of families recognized for Santalales on this web site (20 total) follows Nickrent et al. (2010) and Su et al. (2015). Although the families segregated from Olacaceae s. lat. and recognized by Malécot & Nickrent (2008) were for the most part accepted by APG III (2009), the segregate families of Santalaceae s. lat. were not, despite the evidence presented in Der & Nickrent (2008).  With regard to Santalaceae, Christenhusz et al. (2015) said "... in the future perhaps expand this family to include the majority of Santalales, apart from Balanophoraceae." From my perspective, this tendency towards extreme lumping is excessive, unjustified, and provides no scientific advancement. Moreover, this rash statement was made in the absence of information on the phylogenetic status of Balanophoraceae s. lat. (see below).
    The holoparasite family Balanophoraceae, previously placed in its own order Balanophorales (Takhtajan 1997), has been shown from molecular evidence (Nickrent et al. 2005) to be related to Santalales. A detailed molecular phylogenetic analysis of the entire order was reported in Su et al (2015) and in that study Balanophoraceae was not monophyletic. Three genera, Dactylanthus, Hachettea, and Mystropetalon, emerged in a separate clade and are therefore classified in a separate family Mystropetalaceae.

9. Mitrastemonaceae. Traditionally placed in Rafflesiaceae, this monogeneric family (Mitrastemon) was shown to be related to Ericales by Barkman et al. (2004) using mitochondrial matR gene sequences. This result is confirmed using nuclear SSU rDNA and mitochondrial sequence data (Nickrent et al. 2004).

11. Orobanchaceae. This family name traditionally referred only to an assemblage of holoparasitic taxa that were recognized to be related to the hemiparasites of Scrophulariaceae. Modern circumscriptions of this group (see Young et al. 1999, Olmstead et al. 2001) place all parasitic "scrophs" in a monophyletic family Orobanchaceae along with the non-parasite Lindenbergia. Morphological and molecular evidence clearly place this family in Lamiales.

10. Convolvulaceae. The sole parasitic genus of Convolvulaceae is Cuscuta that has sometimes been placed in its own family, Cuscutaceae. Analysis of sequence data from four chloroplast gene regions resulted in Cuscuta being nested within Convolvulaceae (Stefanovic et al. 2002), thus the classification of APG III (2009) is supported.

12. Lennoaceae. These holoparasites have traditionally been placed in their own family, but APG III lumped Lennoaceae with Boraginaceae, a family with which they are clearly related as shown by both morphological and molecular evidence. The genera Lennoa and Pholisma were shown to be a component of a monophyletic Ehretiaceae by Gottschling et al. (2014).  More recently, Luebert et al. (2016) provided a familial classification for Boraginales and retained the family Lennoaceae, sister to Ehretiaceae.

Literature Cited