8

Molecular Phylogenetic and Evolutionary Studies of Parasitic Plants Daniel L. Nickrent, R. Joel Duff, Alison E. Colwell, Andrea D. Wolfe, Nelson D. Young, Kim E. Steiner, and Claude W. dePamphilis

The parasitic nutritional mode is a frequently 1995). For the vast majority of parasitic plants, evolved adaptation in animals (Price, 1980), as negative effects upon the host are difficult to de• well as in flowering plants (Kuijt, 1969). Het• tect, yet others (e.g., Striga, Orobanche) are se• erotrophic angiosperms can be classified as ei• rious weeds of economically important crops ther mycotrophs or as haustorial parasites. The (Kuijt, 1969; Musselman, 1980; Eplee, 1981; former derive nutrients via a symbiotic relation• Stewart and Press, 1990; Press and Graves, ship with mycorrhizal fungi. Haustorial para• 1995). sites, in contrast, directly penetrate host tissues The degree of nutritional dependence on the via a modified root called a haustorium and host varies among haustorial parasites. Hemi• thereby obtain water and nutrients. Although parasites are photosynthetic during at least one such categories are often a matter of semantics, phase of their life cycle and derive mainly water we use the term parasite in a strict sense to refer and dissolved minerals from their hosts. Oblig• to haustorial parasites. Angiosperm parasites are ate hemiparasites require a host plant to com• restricted to the dicot subclasses Magnoliidae, plete their life cycles whereas facultative hemi• Rosidae, and Asteridae; have evolved approxi• parasites do not. Hemiparasites can be found in mately 11 times; and represent approximately Laurales (Cassytha), Polygalales (Krameria), 22 families, 265 genera, and 4,000 species, that and all families of Santalales. In Solanales is, about 1% of all angiosperms (Fig. 8.1). Ow• (Cuscuta) and Scrophulariales, some species ing to their unique adaptations, parasitic plants are chlorophyllous hemiparasites whereas other have long been the focus of anatomical, mor• species are achlorophyllous holoparasites. Holo• phological, biochemical, systematic, and eco• parasites represent the most extreme manifesta• logical research (Kuijt, 1969; Press and Graves, tion of the parasitic mode because they lack

This work was supported by grants from the National Science Foundation (DEB 94-07984 to DLN, DEB 91-20258 to CWO, and BIR 93-03630 to ADW), the Special Research Program of the Office of Research Development Administration, SIUC and the University Research Council of Vanderbilt University. Thanks go to C. Augspurger, W. Barthlott, I. Beaman, D. E. Bran, S. Carlquist, W. Forstreuter, I. Leebens-Mack, A. Markey, C. Marticorena, D. McCauley, S. Medbury, M. Melampy, Willem Meijer, B. Molloy, L. Musselman, R. N(lfayana, M. Nees, I. Paxton, S. Sargent, B. Swalla, W. Takeuchi, and M. Wetherwax for helpful discussions and/or for contributing plant material. The manuscript was improved by the criti• cal comments of M. Bowe and an anonymous reviewer. 212 MOLECULAR SYSTEMATICS OF PLANTS

/""/"""""~, ~ Sanlalales ~ Solanales , Ol"""c,,, (271190) , , IoIls.,.,.,ndr",,". (118) ~ ~ lor.nlh.ce•• (741cI. 700), , Opillacuc (9129) , , Sanlilleo•• (371480) ~ Boraginales ~ (Incl. Elcmoloptcl.ee.e) , l,nnoocial (315) ~ Vlsca.,.l. (6-7/350) /' ';J;;""""",

Cynomoriales CynomOlloull (111·2) Scrophulariates Scrophul.... I""... s. I. >-oI;-.,...... I-~ (Incl. Orobancllae ..,) Q) ---- (ell. 78/1938) ~ Balanophorales Qj Silanophoflcen (3117) ~ Dac'Yllnlhle••• (314) 1-+-T3~ III D• Sircophylace." (213) O lIlrlCOphlllee .. (6/9) '0 lophophylilceae (318) J: "C ;: !!! Cy1lnales :G Cy1In.ce.e (117) c:: o Z Raffieslales Rlllllllae.la .... (3119) Apodanlhac.ae (2119) IoIlIr.51.monlc ••" (112) Bdallophyron (III"')

Hydnorales Hydnor.ceae (2114-18)

Figure 8.1. The distribution of haustorial parasitism among angiosperms. This generalized diagram incorporates information from global molecular phylogenetic studies using rbcL (Chase, Soltis, Olmstead et al. 1993) and nuclear 18S rONA (Soltis, Soltis, Nickrent et aI., 1997). No attempt was made to show all taxa, only to indicate groups that were supported by both studies. Hemiparasitic angiosperms are enclosed within dashed borders and holoparasites by black borders. Both trophic modes occur in Scrophulariaceae s. I. and Cuscuta. Arrows that touch a group indicate that strong evidence exists for the placement of that parasitic taxon within the group. Uncertain affinities are indicated by arrows with question marks. The familial classification of the nonasterid holoparasites is modified from Tahktajan (1987); however the placement of these orders is not concordant with his superordinal classification. The number of genera and species is indicated in parentheses following each family name. For Scrophulariaceae s. I., only the parasitic members are tabulated.

photosynthesis and must rely upon the host for (1988) by excluding Balanophoraceae, Medu• both water and inorganic and organic nutrients. sandraceae, and Dipentodontaceae. Morpholog• Six groups (orders or families-Fig. 8.1) are ical, cytological, and molecular evidence all represented entirely by holoparasites: Bal• point toward the separation of Cynomoriaceae anophorales, Cynomoriaceae, Cytinaceae, Hyd• from Balanophoraceae and of Cytinaceae from noraceae, Lennoaceae, and Rafflesiales. Rafflesiaceae (Takhtajan, 1987; Nickrent and The relationships shown in Figure 8.1 are Duff, 1996; Pazy et aI., 1996). Traditional clas• based upon those of Takhtajan (1987) as well as sifications have often allied these holoparasites results of recent molecular analyses. For this pa• with Santalales; however, considerable variation per, Santalales are considered, in a strict sense, can be seen in alternate classifications and such to include Olacaceae, Misodendraceae, Loran• an affinity is not apparent from molecular inves• thaceae, Opiliaceae, Santalaceae, and Viscaceae. tigations (see below). Given this, the term This composition differs from that of Cronquist nonasterid holoparasites will be used to distin- PARASITIC PLANTS 213 guish these plants from holoparasites in Asteri• sites. Our goals are to demonstrate the utility of dae such as those found in Scrophulariales, Bor• these molecular markers in documenting phylo• aginales, and Solanales. The nonasterid holo• genetic relationships (at the genus level and parasites are Balanophorales, Cynomoriales, above) and to show how parasitic plants repre• Cytinales, Hydnorales, and Rafflesiales (Fig. sent unique models that can be used to study 8.1). Results of molecular phylogenetic studies molecular evolutionary and genetic processes indicate that the nonasterid holoparasites are not such as the structure, function, and evolution of closely related to each other (Nickrent and Duff, plant genomes. For example, the continuum of 1996), hence the term is applied for reference trophic modes in Scrophulariaceae s. 1. from purposes only. nonparasitic to hemiparasitic to holoparasitic Additional parasitic plants can be found in makes this group ideal for investigating ques• Scrophulariaceae, a large family (over 250 gen• tions concerning the evolution of parasitism and era) that is broadly and at present inexactly de• the molecular changes that accompany adapta• fined within Scrophulariales/Lamiales (Burtt, tion to a heterotrophic existence. Holoparasitic 1965; Cronquist, 1981; Thorne, 1992; Olmstead plants that show increased rates of molecular and Reeves, 1995). Although most species are evolution pose particular problems for phyloge• completely autotrophic, the members of two netic analysis but at the same time provide tribes (Buchnereae, Pediculareae; [Pennell, intriguing subjects for studying genome reorga• 1935]) display a wide range of parasitic modes nizations that accompany the loss of photosyn• from fully photosynthetic, facultative hemipara• thesis. sites, to nonphotosynthetic holoparasites. Orobanchaceae are a group of nonphotosyn• Problems with the Classification thetic holoparasites closely related to the of Parasitic Plants holoparasitic Scrophulariaceae. A continuum of morphological and physiological traits unites The placement of many parasitic plants within the two, as do several "transitional genera" (Har• the global angiosperm phylogeny is not dis• veya, Hyobanche, and Lathraea) that have been puted. For example, despite questions about fa• classified alternatively in one family or the other milial boundaries and interfamilial relationships (Boeshore, 1920; Kuijt, 1969; Minkin and Esh• within Scrophulariales, it is clear that Scrophu• baugh, 1989). Orobanchaceae are alternatively lariaceae are allied with other sympetalous di• included within Scrophu1ariaceae (Takhtajan, cots of Asteridae s. 1. Such is not the case, how• 1987; Thorne, 1992) or recognized, by tradition, ever, for the nonasterid holoparasites whose at the family level (Cronquist, 1981). Regardless higher-level classification still remains prob• of rank, most workers are in agreement that lematic. Two processes that occur during the Orobanchaceae are derived from within the par• evolution of advanced parasitism are reduction asitic Scrophulariaceae. Numerous lines of evi• (and/or extreme modification) of morphological dence support this conclusion, including the features and convergence. The first may involve shared presence of several morphological char• loss of leaves, chlorophyll, perianth parts, or acters (Boeshore, 1920; Weber, 1980), pollen even ovular integuments. Loss of features con• features, and a derived chloroplast DNA restric• founds phylogenetic analysis unless clear trans• tion site loss present in both groups (C. dePam• formation series are apparent. Convergences in philis, unpub1.). For these reasons, this chapter parasitic plants are rampant because similar fea• considers Scrophulariaceae s. 1. to include tures have evolved in unrelated groups. For ex• Orobanchaceae (Fig. 8.1). ample, the squamate habit (i.e., with scale In this chapter we discuss the results of leaves) has evolved independently in aerial par• macromolecular studies of the groups shown in asites of Viscaceae, Santalaceae, and Misoden• Fig. 8.1. Most attention is, by necessity, directed draceae. One of the most striking examples of toward four groups for which DNA sequence convergence involves Cuscuta (Asteridae) and data are available, Scrophulariaceae s. 1., Cus• Cassytha (Magnoliidae), two twining, yellow to cuta, Santalales, and the nonasterid holopara- orange parasites that are frequently confused 214 MOLECULAR SYSTEMAnCS OF PLANTS based on a superficial examination of morphol• approximately 25%) encode proteins for photo• ogy. With reference to the internal haustorial tis• synthetic carbon fixation and electron transport, sues (endophytes) of Rafflesiaceae and Vis• about half (60,54%) encode proteins or RNAs caceae, Cronquist (1981, p. 698) states " ... involved in gene expression (transcription, parallelism in a number of features is in itself translation, and related processes). An addi• some indication of relationship, to be considered tionalll genes (10%; the ndh genes) encode pu• along with other evidence." Because parallelism tative "chlororespiratory" proteins, based on involves similarities due to analogy and homol• their sequence similarity to mitochondrial ndh ogy, and given that the common ancestry of the genes (Shimada and Sugiura, 1991). Ten or two families has not been established, this situa• more large ORFs (and many smaller ones) also tion should best be described as convergence. A represent potentially functional protein genes, more serious problem is that this statement im• but their functions are largely unknown. Among plies that all characters, whether they evolved the many hundreds of species whose cpDNAs via parallelism or convergence, should receive have been characterized using restriction en• equal consideration when constructing a phy• donuclease mapping (Downie and Palmer, logeny. 1992), most retain a very similar gene content The classifications of parasitic plants are of• and gene order. Molecular systematic research ten plagued by past misconceptions or overem• on photosynthetic plants has extensively used phasis on a few conspicuous characters. The phylogenetically informative variation in the parasitic habit itself, known to have had multiple presence and absence of restriction endonucle• origins, was used by Cronquist (1981) to link ase sites, as well as a very large and rapidly Balanophoraceae, Hydnoraceae, and Rafflesi• growing comparative database of DNA se• aceae to Santalales. It is not our contention that quences for the plastid genes rbcL, matK, and molecular data are more immune to the effects ndhF, and smaller databases for other plastid of convergence, parallelism, and reversal (i.e., genes (see Chapter 1). Rare structural mutations homoplasy) than morphological characters, only such as inversions and insertions or deletions of that homologous DNA sequences provide addi• genes or introns provide additional characters of tional, independent genetic characters that can special phylogenetic significance (reviewed by be rigorously analyzed to test phylogenetic hy• Downie and Palmer, 1992; see also Chapter 1). potheses. Furthermore, molecular data can be Because many of the parasitic plants dis• used to explore the dynamics of molecular evo• cussed herein are nonphotosynthetic, the more lution in parasitic plants by examining the ge• general term plastid will be used instead of netic structure and biochemical processes un• chloroplast and ptDNA (plastid DNA or plas• derlying evolutionary change. tome) will be used instead of cpDNA (chloro• plast DNA). Given that a large fraction of the THE PLASTOME OF PARASITIC PLANTS plastome is devoted to photosynthetic function and to the expression of photosynthetic and The chloroplast genomes (cpDNA) of photosyn• other genes, parasitic plants provide a unique thetic plants are circular molecules ranging in opportunity to determine the extent to which the size from about 120 to 217 kilobase pairs (kb) evolutionary conservation of ptDNA is related (Downie and Palmer, 1992). These genomes to photosynthetic ability. In this regard, parasitic typically have a large (about 25 kb) inverted re• plants may be considered to be natural genetic peat that contains rRNA-encoding and other mutants for the dissection of ptDNA function genes, and separates the large and small single• and evolutionary processes (dePamphilis, 1995, copy regions of the molecule. A typical chloro• and references therein). At the same time, devel• plast genome, such as the completely sequenced opment of a phylogenetic framework to interpret cpDNA of Nicotiana (Shimada and Sugiura, significant variation among parasitic and non• 1991), consists of about 112 genes and poten• parasitic lineages is complicated by the absence tially functional open reading frames (ORFs). of many plastid genes from at least some para• Although a substantial number of the genes (29, sitic lineages and by the acceleration of evolu- PARASITIC PLANTS 215 tionary rate for many other plastid and nuclear just 42 intact genes. Surprisingly, although the genes (see Rate Variation among Genomes and Epifagus ptDNA had sustained a large number Lineages). of deletions (and some small insertions as well) relative to another asterid, Nicotiana, the two Epifagus genomes are almost entirely colinear (dePam• philis and Palmer, 1990). Complete sequencing Epifagus virginiana (beechdrops, Scrophulari• revealed only one small inversion in trnL of the aceae s. 1.) is a holoparasite native to eastern small single-copy region of Epifagus relative to North America whose sole host is Fagus grandi• Nicotiana. Furthermore, the parasite retains a folia (American beech). Although entirely lack• nearly full-sized inverted repeat of 22,735 bp ing chlorophyll and photosynthetic ability, that separates the greatly reduced large and Epifagus retains plastids (Walsh et aI., 1980) small single-copy regions of the genome. and ptDNA (dePamphilis and Palmer, 1989). Comparison of the ptDNA gene content of Detailed mapping (dePamphilis and Palmer, Epifagus with that of Nicotiana revealed a 1990) and complete sequencing of the Epifagus highly selective pattern of gene deletion in the plastome (Wolfe et aI., 1992c) revealed a greatly holoparasitic plant (Table 8.l). All plastid• reduced genome of only 70,028 bp, containing encoded photosynthetic genes have been deleted

Table 8.1. Plastid genes in Epifagus (42 totala) compared to Nicotiana (112 total).

Gene Present Deleted or Pseudogene (1jI) in Epifagus in Epifagus Photosynthesis Photosystem I psaA, B, C, I, J Photosystem II IjIpsbA, IjIB, C, D, E, F, psbH, 1, J, K, L, M Cytochrome blf petA, B, D, G ATP synthase ljIatpA, IjIB, E, F, H, I Calvin cycle IjIrbcL

Chlororespiration ndhA, IjIB, C, D, E, F, G, ndhH, 1, J, K

Gene Expression rRNA 16S, 23S, 4.5S, 5S Ribosomal protein rps2, 3, 4, 7, 8, 11, rpsI5, 16, 1jI14, rps14, 18, 19, rp122, 1j123, 32 rp12, 16, 20, 33, 36

Transfer RNA trnD GUC' EuuC' F GAA' IjItmA UGC' IjICGCA, GGCc' tmHGUG' IcAu' LCAA' tmGucc' IjI/GAU' Kuuu' tm4AG' MCAU' NGUU' tm4AA' IjIRucu' IjISGGA' tmPUGG' QUUG' RACG' tmTGGU' TUGU' VGAC' tmSGCu' SUGA' WCCA' tmVuAC tmYGUA' tmjMcAu RNA polymerase IjIrpoA, B, CI, C2 Maturase rootK Initiation factor infA

Other protein genes clpP, aceD, ORF1738b, ORF29, 31, 34, 62,168, ORF22l6b ORF184, 229, 313

"Data based on gene mapping (dePamphilis and Palmer, 1990) and complete sequence (Wolfe et al., 1992c). All genes listed are present in Nicotiana chloroplast DNA except in/A, which is a pseudogene in that plant (Wolfe et aI., 1992c). 'ORF1738 and ORF2216 in Epifagus are homologs of Nicotiana ORF1901 and ORF2280, respectively (Wolfe et aI., 1992c). 216 MOLECULAR SYSTEMATICS OF PLANTS from Epifagus or remain as small fragments or (Mayfield et aI., 1995). Expression of plastid pseudogenes. Similarly, the 11 ndh genes have genes in the absence of plastid-encoded RNA been lost. A majority of genes of unknown func• polymerase would imply that another RNA tion (ORFs 168, 184,229, and 313, plus a host polymerase, presumably of nuclear origin, is in• of smaller ones) has been completely lost from volved in the transcription of beechdrops Epifagus. In contrast, 38 (or over 90%) of the re• ptDNA (Morden et al., 1991; Ems et aI., 1995). tained genes represent components of the plastid It is interesting to note also that virtually all of machinery for gene expression. All of the four the proposed eubacterial-type -35 and -10 ribosomal RNA genes are retained, as are intact promoter regions have either diverged or been copies of 13 of 19 plastid-encoded ribosomal deleted in Epifagus (Wolfe et aI., 1992a, 1992c), protein genes and 17 of 30 plastid tRNA genes. consistent with the possibility that these genes The implications of these results are that ptDNA now use different promoters, possibly ones ca• can evolve very rapidly under certain pre• pable of interacting with an RNA polymerase of dictable conditions, and that DNA deletions, nonplastid origin. Similarly, the loss of 13 of the both large and small, are a dominant mode of 30 normally plastid-encoded tRNA genes would structural evolution in that molecule. This is a suggest that plastid translation, to the extent that striking example of the role natural selection it occurs, must also require the import of some plays in shaping genome structure and suggests tRNAs from outside the plastid (Wolfe et aI., that rapid structural alteration is possible when 1992c). Finally, the retention of just four pro• photosynthetic constraints are altered. tein-coding genes not involved in gene expres• These results also have important implica• sion (clpP, aceD, ORF1738, and ORF2216) in tions for our understanding of the function of the plastome of Epifagus suggests that one or ptDNA. RNA polymerase subunit genes (rpoA, more of these genes encodes a protein required B, Cl, and C2), which encode the central en• for a nonbioenergetic (i.e., not involved in pho• zymes of transcription, are present in tobacco tosynthetic or chlororespiratory) function in the but absent in Epifagus. If plastid-encoded RNA Epifagus plastid and may be the raison d' etre for polymerase is solely responsible for ptDNA the retention of a functional ptDNA (Wolfe et transcription, we would expect the beechdrops al., 1992c; dePamphilis, 1995). ptDNA to be unexpressed. To the contrary, de• A related study investigated the phylogenetic spite the loss of all RNA polymerase genes, the distribution of one of the lost tRNA genes, plastome of Epifagus is clearly transcribed and tRNA-cys, in 11 parasitic and nonparasitic is probably also translated (Wolfe et aI., 1992c). plants related to Epifagus (Taylor et aI., 1991). The evidence comes from several sources, The polymerase chain reaction was used to am• including direct observation of transcripts of plify fragments that would contain the tRNA• plastid rRNA genes (dePamphilis and Palmer, cys locus from each plant. The fragments were 1990), ribosomal protein genes (Ems et then cloned and sequenced to reveal that all of aI., 1995), and additional plastid ORFs (Ems et the photosynthetic plants had a normal tRNA• aI., 1995). Epifagus plastid RNAs are subject to cys gene, whereas all of the nonphotosynthetic appropriate intron splicing and even experience plants had lost the gene. This result suggests that RNA editing, as observed in other plastid RNAs the tRNA-cys gene was lost at about the same (Ems et aI., 1995). Furthermore, the highly spe• time as photosynthesis was lost and that the two cific pattern of gene retention and bias of syn• events may be causally related. onymous over nonsynonymous base substitu• tions, particularly in large genes subject to Conopholis significant levels of sequence divergence (Wolfe et aI., 1992b, 1992c; dePamphilis et aI., 1997) is Concurrent with the early work on Epifagus, the clearly indicative of genes that have evolved un• plastome of another orobanchaceous holopara• der functional constraint. site, Conopholis americana (squawroot), was That gene expression in plastids is linked to being examined at the molecular level. Using nuclear-encoded genes is well documented heterologous probes, Wimpee et ai. (1991) doc- PARASITIC PLANTS 217 umented the modification or absence of many plastome. The complete loss of ribosomal pro• photosynthetic genes. Later, the rONA operon tein genes such as rp12 invites questions about was cloned and sequenced, thereby demonstrat• how the translational apparatus of the plastid ing that the 16S/23S spacer lacked tRNA genes functions and its relationship to the nuclear and and contained substantial deletions (Wimpee et mitochondrial genomes. The plastid genome of al., 1992a). Experiments using the polymerase the holoparasite C. europaea has undergone ad• chain reaction and hybridizations failed to detect ditional deletions in comparison to C. reflexa, the lost tRNA genes elsewhere in the squawroot such as the loss of the cis-spliced intron of rps12 genome (Wimpee et al., 1992b). Colwell (1994) (Freyer et al., 1995). Despite this alteration, cor• conducted restriction site mapping of the plas• rectly processed transcripts and ribosomal pro• tome of squawroot, documenting its size as 43 teins are produced, thus providing evidence that kb, making it the smallest ptDNA molecule then the translational apparatus of the plastid may re• observed in plants. This genome is only slightly main functional following significant structural smaller than that of Epifagus if the length of one alterations. inverted repeat is removed from the latter genus (70 kb - 22.7 kb = 47.3 kb). Do the Nonasterid Holoparasites Retain a Plastome1 Cuscuta As described in the foregoing discussion, the Other molecular genetic investigations have fo• plastomes of Epifagus, Conopholis, and Cus• cused on Cuscuta, a genus sometimes placed in cuta have experienced extensive losses of pho• its own family (Cuscutaceae) but widely recog• tosynthetic genes, yet all have retained intact nized to share a common ancestor with nonpara• and functional ribosomal cistrons. It was there• sitic Convolvulaceae. Cuscuta, like Scrophular• fore reasoned that if any portion of the plastome iaceae, includes hemiparasitic species (e.g., c. remained in nonasterid holoparasites, it would reflexa) as well as holoparasitic species (e.g., C. likely be the ribosomal cistron. The first evi• europaea) that lack thylakoids, chlorophyll, dence that rONA was present in these holopara• RUBIS CO (ribulose bisphosphate carboxy• sites was obtained when plastid 16S rONA from lase/oxygenase) and light-dependent CO2 fixa• Cytinus (Cytinaceae) was amplified and se• tion, but (strangely) retain rbcL (Machado and quenced (Nickrent et aI., 1995; Nickrent and Zetsche, 1990). Although the plastome of Cus• Duff, 1996). In pairwise comparisons to other cuta is yet to be sequenced completely, signifi• angiosperms, this 16S sequence contained ap• cant progress is being made. A 6-kb portion of proximately three times more base substitutions, ptDNA from C. reflexa that includes petG, tm V, yet this rRNA retained all major secondary tmM, atpE, atpB, and rbcL was sequenced structural features (because most changes were (Haberhausen et al., 1992; Haberhausen and compensatory), thus suggesting functionality. Zetsche, 1992). Further work on this species Since these initial studies, 16S rONA sequences (Bommer et al., 1993) resulted in sequences for have been obtained from all nonasterid holopar• 16S rDNA,psbA, tmH, 0RF740, 0RF77, tmL, asite lineages (Nickrent et al., 1997a). These and ORF55. Later, a 9-kb portion of ptDNA represent the most diverse 16S rRNA sequences from this same species that included a large por• documented among angiosperms and contain tion of inverted repeat A spanning a segment several structural features unknown among land from trnA to tmH was cloned and sequenced plants. (Haberhausen and Zetsche, 1994) Although Southern blots using both homologous and some short sequences were identical to those of heterologous probes developed for five plastid Nicotiana (e.g., tml), many deletions were ob• genes (16S rONA and four ribosomal protein served. For example, rp12 and rp123 were both genes) resulted in positive hybridizations to di• deleted, and ORF2280 was reduced to only 740 gested total genomic DNA from Cytinus (Nick• bp. These results show that, like Epifagus, C. re• rent et al., 1997b). These preliminary data sug• flexa has experienced major deletions in the gest the Cytinus genome is approximately 20 kb 218 MOLECULAR SYSTEMATICS OF PLANTS in size, thus the smallest yet documented for obanche, and Lathraea, have been alternatively angiosperms. Hybridizations using more de• placed in either family (Kuijt, 1969). Does the rived holoparasites such as Rafflesia and Bal• evolutionary series nonparasite· hemiparasite' anophora were negative, thus suggesting their holoparasite explain the pathway to all holopar• plastome is either absent or even more exten• asites? Are the transitional genera and Oroban• sively modified than that of Cytinus. Recently, chaceae monophyletic or are they independent sequences of PCR-amplified products derived lineages that have converged on a nonphotosyn• from the plastid 16S-23S spacer have been ob• thetic lifestyle? If the latter is true, how many tained from Cytinus and Cynomorium (D. Nick• times has photosynthesis been lost? The avail• rent and R. Duff, unpubl.), further suggesting ability of numerous extant species displaying all the retention of a plastome in these plants. modes of parasitism makes these questions tractable via systematic analyses. The molecular phylogeny of parasitic Scroph• MOLECULAR PHYLOGENETIC STUDIES ulariaceae has been examined using sequences OF SCROPHULARIACEAE S. L. from three plastid genes: rps2, marK, and rbcL. Systematic and Phylogenetic Problems The first two are relatively rapidly evolving plastid genes that encode nonphotosynthetic Until recently, most opinions on the circum• proteins (see Chapter 1) and that we have found scription of Scrophulariaceae could be traced to are present in all parasitic Scrophulariaceae re• concepts from the nineteenth century, such as gardless of photosynthetic ability. Although the the recognition of three subfamilies (Pseudo• photosynthetic gene rbcL is absent from some solaneae, Antirrhinoideae, and Rhinanthoideae) holoparasites (see Phylogeny of Scrophulari• and a distinct Orobanchaceae. Rhinanthoideae aceae s.1. using rbcL), it is present in most taxa. were traditionally considered to include the par• Phylogenetic analyses have revealed much asitic tribes Buchnereae and Euphrasieae, as about the molecular evolution of this key plastid well as the nonparasitic tribes Veronicae and gene in relation to photosynthesis. Digitaleae. Anatomical evidence was used to re• fute a presumed relationship between Rhinan• Phylogeny of Scrophulariaceae s. I. thoideae and the slightly zygomorphic Pseu• Using rps2 dosolaneae and Solanaceae (Thieret, 1967). Other family-level issues concern the presumed Phylogenetic analysis of plastid ribosomal pro• "edges" of the family, including whether to rec• tein gene rps2 was conducted for 55 species of ognize the Selagineae and/or the Globulariaceae Scrophulariaceae and related families (Fig. 8.2), as formal taxa within Scrophulariaceae. A gen• including 32 species of parasites (dePamphilis et eral trend has been the recognition of various al., 1997). Nonparasites were selected to reflect small tribes (including Angeloneae, Russelieae, the patterns shown following analysis of a larger Collinsieae, Melospermeae, and many others) data set (58 genera of nonparasitic Scrophulari• and especially the splitting of three large and aceae and relevant outgroup taxa; C. dePam• heterogeneous tribes, the Digitaleae (Hong, philis unpubl.). Parsimony analysis of rps2 iden• 1984), Gratioleae, and Cheloneae (Thieret, tifies a single clade (bootstrap value of 53%) 1967) into smaller, presumably monophyletic containing a broad sampling of parasitic Scroph• tribes. Attempts to analyze cladistically morpho• ulariaceae s. 1. With the exception of Lathraea, logical variation across the entire family have the "transitional" genera Harveya and Hyo• met with little success, leading An-Ming (1990) banche are present on the rps2 consensus tree as to speculate about the possible paraphyly, or a clade (88% bootstrap support) ofhemiparasitic even polyphyly, of the family as presently de• Scrophulariaceae; without these genera, the re• fined. maining orobanchaceous taxa are weakly sup• Although a close relationship between Oro• ported as monophyletic. These data indicate that banchaceae and parasitic Scrophulariaceae hemiparasites gave rise to holoparasites a mini• seems assured, genera such as Harveya, Hy- mum of five separate times (dePamphilis et al., PARASITIC PLANTS 219

Aga/lnls .". B CatlIMja IInsrlfoliB E rps2 89 Seymerls pectlnats B PedIcuItlriB IoIIosa E PedlcultlrlslltOlleus E A/ec:Ira orobancholdes B 97 AIet:tra ....1I1f1ora B Melasma scabrum B Buch".,. fIorIdIInB B StrIgs asiatica B StrIgalJfl8l'lfll'lo/ B Parasitism ....-- Cycnlum rtJCeIfJ08um B ---- SopubiB cans B Harveya t:apflMis B * Harveya purpurea B * "53 Hyobtmche .."gulnea B * p------~ism~ns E Euphreais spectabilis E PIll'8lltut:ellia vlsCOSlJ E Tozzilf a/pIfJII E Lllthraea clandestlna E * I....:~...- Afelampyrum II"..,. E Rhlnanthus crlstarus E 77 Soschnls/cls hooker/ Boschnis/cla strobilacea Cooopholis americana Eplfagus virglniana Orobanche cemua Orobanche tamosa Orobanche col)'mbosa Orobanche fasciculata Orobanche uniflata ....------Lindenberg'" sp. K"/{/6lia pinnata (BIG) I------+_ Schlegelis parvlt/ors Verbena bonarlensis (VER) 1-______Mlmulus aunmtlacus 1------Paulownia tomentosa (?) HB/1er#tJ luelda 1------eelsis arcturus Scrophu"'rIs ca/lftlm/ca Verbasc:um biBtterfa 1 -----'-':.::::: Belago thunbergll ] t Hem/merls sabulosa - ....------Myoporum parvifolium (MYO) 1..------Collinsissp. r----- ehe/one obllqus (I ~ ______~ AnUnhmwnme~s ] Q DW~Ispurpu,... ~ Veronica arvensls ::r Callitrlche hermaphroditica (ell) = ------G~is~1osa 1..------Koh/eria dlgital/flata (GSN) 1------Ligustrumjaponicum(OlE) 1..------Nicotiana tabacum(SOl) Figure 8.2. Phylogenetic reconstruction of Scrophulariaceae and outgroups based on maximum parsimony analysis of plastid rps2 gene sequences. The tree was constructed using 174 phylogenetically informative base substitutions. Strict consensus of 78 trees at 660 steps (Consistency Index [CI] without autapomorphies = 0.492; Retention Index [RI] = 0.657). Bootstrap values shown above branches; nodes without values were supported in less than 50% of the replications. The arrow indicates a parasite clade including genera traditionally assigned to Orobanchaceae. Nonphotosynthetic holoparasites are indicated by thick lines. Taxa in bold font are traditionally placed in Scrophulariaceae. The traditional parasitic tribes Buchnereae and Euphrasieae are indicated by B and E, respectively. Taxa followed by an asterisk (*) are "transitional" genera classified alternatively in Scrophulariaceae or Orobanchaceae. Scroph I and II refers to clades identified in Olmstead and Reeves (1995). Abbreviations for outgroups and families in Scrophulariales: BIG = Bignoniaceae; CLL = Callitrichaceae, GSN = Gesneriaceae; MYO = Myoporaceae; OLE = Oleaceae; SOL = Solanaceae; VRB = Verbenaceae. The "T' for Paulownia indicates its higher-level classification is uncertain. 220 MOLECULAR SYSTEMATICS OF PLANTS

1997). Seven of the nine genera classified as reconstructions. The rbcL locus from 30 addi• members of Euphrasieae occur in a single clade; tional taxa representing 20 genera of Scrophu• the remaining two genera (Castilleja and Pedic• lariaceae has been sequenced (Wolfe and ularis) occur in a clade with two genera of Buch• dePamphilis 1995, 1997). Phylogenetic recon• nereae (Agalinis and Seymeria). struction based on the new sequences and those In contrast to the apparent monophyly of the from Olmstead and Reeves (1995) reveals that parasitic taxa, the nonparasitic Scrophulariaceae Scrophulariaceae, as traditionally circumscribed, are only weakly resolved with rps2 and appear are not monophyletic (Fig. 8.3). Although Olm• as a complex paraphyletic assemblage with rep• stead and Reeves (1995), arrived at the same resentatives of other families in Scrophulariales conclusion based on a combined analysis of such as Bignoniaceae, Verbenaceae, and My• rbcL and ndhF sequences, specific relationships oporaceae. Low bootstrap values suggest that between major clades differ between the two the power of this short gene to resolve phyloge• studies. The "scroph I" and "scroph II" clades netic relationships in the nonparasitic lineages is were distinct in Olmstead and Reeves (1995) but low. Yet, two of the nonparasitic lineages recov• in the present analysis they occur in one clade ered correspond to the scroph I and scroph II with representatives of other Scrophulariales. clades identified by Olmstead and Reeves One of the major results of the phylogenetic (1995) in a phylogenetic analysis using rbcL and analysis based on both rps2 (Fig. 8.2) and rbcL ndhF sequences. Other genera not sampled by (Fig. 8.3) sequences is that there is a single ori• Olmstead and Reeves (Gratiola, Collinsia, Hal• gin of the parasitic habit in Scrophulariaceae. Ie ria, Mimulus, and Lindenbergia plus the entire Taxa traditionally placed in Orobanchaceae are parasitic lineage) are not components of these intercalated with hemi- and holoparasitic genera clades. These results suggest that at least three of the Scrophulariaceae, hence Orobanchaceae and possibly additional major lineages may rep• (as traditionally defined) are paraphyletic. The resent what has now clearly been identified as a genus Orobanche is also not monophyletic ac• polyphyletic Scrophulariaceae (Olmstead and cording to the rbcL strict consensus tree (Fig. Reeves, 1995). Neither Veronica nor Digitalis, 8.3). The rps2 results suggested at least five in• genera usually included within the rhinanthoid dependent losses of photosynthesis. A strict in• subfamily, are close relatives of the parasitic terpretation of the rbcL tree indicates seven plants. In contrast, Lindenbergia, a nonparasitic losses of photosynthesis, assuming that it is un• genus originally classified in the Gratioleae likely that the nonphotosynthetic ancestors of (Wettstein, 1897), occurs as sister to all other Orobanche, Boschniakia, and others gave rise to parasitic Scrophulariaceae. photosynthetic hemiparasites. Some of these re• lationships, such as lack of monophyly for Phylogeny of Scrophulariaceae s. I. Orobanche or a distant relationship between Using rbcL Harveya and Hyobanche, are not in accord with rps2 or 18S rDNAdata (see next section); hence Studies employing rbcL sequencing for phylo• further examination of the phylogenetic utility genetic reconstruction in Scrophulariaceae have of rbcL in these plants is in order. been, until recently, restricted to nonparasitic In those holoparasites where rbcL no longer plants (Olmstead and Reeves, 1995; Wolfe and produces a functional product, it is considered a dePamphilis, 1995, 1997; dePamphilis et aI., pseudogene. In most cases of pseudogene forma• 1997). Recent studies by Wolfe and dePamphilis tion, a functional gene is retained in addition to a (1995, 1997) have revealed accelerated synony• degenerated duplicated copy (e.g., l3-globin genes mous and nonsynonymous substitution rates for [Li, 1983]); however, in Scrophulariaceae no du• rbcL in several lineages of derived members of plication has occurred, and only the primary rbcL Scrophulariaceae. In comparison to nonparasitic pseudogene remains. These data indicate that pri• plants of Scrophulariales, the evolutionary rate mary pseudogene formation occurred indepen• of rbcL in those parasites that retain this gene is dently three times (Fig. 8.3). The most surprising faster, thus enabling more resolved phylogenetic discovery from this rbcL phylogenetic analysis is PARASITIC PLANTS 221

Aleetra orobanchoides Alectra sessl/lflora rbel Hyobanche atropurpurea 'II * Hyobanche sangulnes '11* Buchnera florldane Cycnium racemosum Strip asiatica Striga gesnerloides Strip hermonthlca Strlga passargei 99 Harveya capensis * Harveya purpurea * ~ 100 Orobanche cernua 'II Orobanche tamosa 'II i Boschnlakla hookerl 5 'II ::T 68 Boschnlakia strobilaces 'II I» 100 O~hecorymbosa J~ Orobanche fascicu/ata 8c 54 Castilleja linarHolia UJ Pedicularis foliosa Seymer/a pectinata Me/ampyrum lineare Lathraea c/andestina * Tozzia alpina Paulownia tomentosa (?) Schlegella parvfflora Myoporum paNifolium (MYO) Torenia foumlerl Catalpa sp. (BIG) Martinella obovata (BIG) Tabebuia heterophylla (BIG) Acanthus montanus (ACA) Hypoestes forsskaolii (ACA) Justicia americana (ACA) Lepidagathis villosa (ACA) Ruellia graezicans (ACA) Thunbergia usamberica (ACA) Sesamum indicum (PED) AlonsOil uni/abiats -, Antirrhinum maJus Digitalis purpurea Veronica catenats UJ Plantago lanceolata (PTG) n.... Callitriche hermaphroditica (CLL) 0 ~ Che/one obliqua ::T Collinsia grandiflora Celsia arcturus AD Scrophu/ar/a sp. = Verbascum thapsus Se/ago thunbergii Gratio/a pilosa Utricularia bit/ora (LNT) Haller/a/uclda Mlmulus aurant/acus ..oJ Nematanthus hirsutus (GSN) Streptocarpus holstii (GSN) Nicotiana tabacum (SOL)

Figure 8.3. Phylogenetic reconstruction of Scrophulariaceae based on maximum parsimony analysis of rbcL nucleotide sequences. Taxa with primary pseudogenes are indicated as I/J (see Fig. 8.6 and text). Strict consensus of 12 most• parsimonious trees with 1,482 steps (CI = 0.550, RI = 0.521). The arrow indicates a parasite clade including genera traditionally assigned to Orobanchaceae. Nonphotosynthetic holoparasites are indicated by thick lines. Taxa in bold font are traditionally placed in Scrophulariaceae. The clade identified as scroph I and II refers to taxa described in Olmstead and Reeves (1995), but here also includes representatives of other families. Taxa followed by an asterisk (*) are ''transitional'' genera classified alternatively in Scrophulariaceae or Orobanchaceae. Abbreviations for families in Scrophulariales in addition to those in Fig. 8.2: ACA = Acanthaceae, GLB = Globulariaceae, LNT= Lentibulariaceae, PED = Pedaliaceae, PTG = Plantaginaceae. 222 MOLECULAR SYSTEMATICS OF PLANTS that many holoparasitic Scrophulariaceae (e.g., were determined by Colwell (1994). That study Aleetra, Harveya, Lathraea, Orobanehe, and examined nine parasitic genera; subsequent se• Striga) retain an open reading frame for the gene quencing has increased to 18 the number of 18S (Fig. 8.3). The independence of structural muta• rDNA sequences of Scrophulariaceae s. 1. now tions among and/or within genera for the pseu• available for analysis (Nickrent and Duff, 1996). dogene sequences (see Structural Analysis of Using Glycine, Lyeopersieon, and Ipomoea as RUBIS CO) suggests that disruption of the gene outgroups, a clade is retrieved (Fig. 8.4) contain• occurred independently following the loss of pho• ing the nonparasitic Linaria and Chionophila; this tosynthesis in Bosehniakia, Conopholis, Hy• clade is sister to the remaining Scrophulariaceae obanehe, Epifagus, and Orobanehe (dePamphilis s.1. These results based on 18S rDNA sequences and Palmer, 1990; Wolfe et al., 1992c; Colwell, indicate, as did rps2 and rbeL, that parasitism 1994; Wolfe and dePamphilis, 1997). Combined, arose just once in this group. Although sampling these data suggest that a rbeL ORF may have been was limited compared to analyses of the chloro• retained in these holoparasitic lineages after the plast genes, a number of relationships are con• loss of photosynthesis and that perhaps the gene cordant, such as clades containing: (1) Pedieu• was functional at some point following the adap• laris, Orthoearpus, and Castilleja; (2) Harveya tation to heterotrophy. and Hyobanehe; and (3) Epifagus and Conopho• lis. Unlike the results from the chloroplast genes, Phylogeny of 8crophulariaceae s. I. 18S rDNA sequence analysis resulted in a strong Using Nuclear 188 rONA association (91 % bootstrap support) between the "transitional" genus Lathraea and an Orobanehe A general review of the organization of nuclear clade that is composed of only North American 18S rDNA and its use in phylogenetic studies of taxa. The rDNA tree also indicates (as does rps2) plants can be found in Chapter 1. Phylogenetic that this clade is not closely related to the North relationships among hemi- and holoparasitic American genera Epifagus, Conopholis, and Scrophulariaceae using l8S rDNA sequences Bosehniakia, making Orobanchaceae (in the tra-

,...--- Pedicularls lanceolata 18S rONA Pedicularls racemosa Orthocarpus erlanthus ...... -.-i----- Orthocarpus lutea Castilleja mlnlata Parasitism ...------Lathraea clandestina 99 Orobanche ludovlciana Orobanche multiflora ..__ 99__ .... C = Orobanche fasciculata-CO '" Orobanche fBSciculata-CA L--..J~_-o[======HyobancheHarveya speciosa atropurpurea Eplfagus virglnlana .---1 96 Conopholls alplns Conopholis americana ..--- Boschnlakla rassica L~---- Linaria vulgaris ...... _-- Chlonophila Jamesii L_~8~2_-{===:-;;:;:;: Lycopersicon esculentum (SOL) Ipomoea hederacea (CON) ...... ------Glycine max (FAB) 5 steps Figure 8.4. Parsimony analysis of nuclear 18S rONA sequences from Scrophulariaceae s. I. Strict consensus phylogram of four trees at 246 steps (CI without autapomorphies = 0.553; RI = 0.643. Bootstrap values (from 100 replications) shown above branches; nodes without values were supported in less than 50% of the replications. The arrow indicates a parasite clade including genera traditionally assigned to Orobanchaceae. Nonphotosynthetic holoparasites are indicated by thick lines. Linaria and Chionophila are nonparasitic Scrophulariaceae. Lycopersicon (Solanaceae), Ipomoea (Convolvulaceae) and Glycine (Fabaceae) were included as outgroups. PARASITIC PLANTS 223 ditional sense) paraphyletic. Although nuclear ships such as the derivation of Balanophoraceae 18S rDNA sequences usually provide insuffi• from Viscaceae. As with others before and after cient data at the rank of family and below, in• him, Fagerlind's system failed to deal with con• creased rates of nucleotide substitution, espe• vergent features in unrelated groups. cially in orobanchaceous and transitional genera, At least three systems for interfamilial rela• have allowed greater resolution of some relation• tionships in Santalales have been proposed that ships. From the same taxa sampled for rps2 and differ mainly in their circumscription of Ere• rbeL, 18S rDNA sequences should be obtained, molepidaceae and Santalaceae. The first places thereby allowing direct comparisons of resulting Santalaceae as sister to Viscaceae and Eremolep• trees and analyses of combined data sets. idaceae as sister to Loranthaceae (Kuijt, 1968, 1969). The second also places Santalaceae as sis• ter to Viscaceae, but does not consider Ere• MOLECULAR PHYLOGENETIC molepidaceae as distinct from the former family STUDIES OF SANTALALES (Wiens and Barlow, 1971). The third hypothesis AND NONASTERID HOLOPARASITES (Bhandari and Vohra, 1983) allies Loranthaceae Systematic and Phylogenetic Problems with Viscaceae (the latter including Eremolepi• daceae). All three of these systems include Opil• Traditional classifications place Santalales in iaceae as a part of Olacaceae; the latter family Rosidae, often near Celastrales (Cronquist, being the least derived in the order because it 1981; Takhtajan, 1987). In the system of Cron• contains both parasitic and nonparasitic mem• quist (1981), Santalales comprise ten families bers. These differing hypotheses have provided (including Balanophoraceae) and this order is the impetus for molecular phylogenetic investi• allied with Rafflesiales (Rafflesiaceae and Hyd• gations (Nickrent and Franchina, 1990). noraceae). The phylogenetic system proposed by Takhtajan (1987) is similar, except for the Phylogeny of Santalales Using Nuclear placement of the nonasterid holoparasites (Fig. 18S rONA and rbcL 8.1). Takhtajan (1987) places Rafflesiaceae, Hydnoraceae, and Balanophoraceae (minus At present, 62 nuclear 18S rDNA (Nickrent and Cynomoriaceae) in subclass Magnoliidae, not Duff, 1996) and 37 rbeL sequences exist for rep• Rosidae. For Santalales, concepts of interfamil• resentatives of six families of Santalales (Fig. ial relationships date to the nineteenth century. A 8.1). For Viscaceae and Misodendraceae, all noncladistic phylogeny of the order was pro• genera have been sampled. Phylogenetic analy• posed by Fagerlind (1948); however, this system ses for the same suite of 37 taxa using 18S was overly influenced by the presence of Allium rDNA and rbeL sequence data, separately and in type embryo sacs and reductions seen in the gy• combination, allow the utilities of these mole• noecium, thereby resulting in unlikely relation- cules to be compared directly (Table 8.2). The

Table 8.2. Comparison of the results of phylogenetic analyses of Santalales using 18S rDNA, rbeL, and 18S rDNAlrbeL combined. a Tree Characteristic 18SrDNA rbeL 18S + rbeL Length of shortest tree (steps) .1047 1134 2212 Number of shortest trees 15 8 11 Number of characters 1741 1410 3151 Number of variable characters 433 468 901 Number of potentially informative characters 226 275 501 Consistency IndexlC.I. minus uninformative sites 0.510.39 0.54/0.44 0.5110.40 Homoplasy IndexIH.1. minus uninformative sites 0.5/0.61 0.46/0.56 0.4810.60 Retention Index 0.57 0.61 0.58 Rescaled Consistency Index 0.29 0.33 0.30

"Sequences for the same suite of 38 taxa were used in a heuristic search using PAUP. 224 MOLECULAR SYSTEMATICS OF PLANTS number of variable characters and the number of family. The combined analysis indicates that phylogenetically informative characters for the Opiliaceae form a well-supported (100% boot• two molecules are similar, as are the consistency strap) clade that is sister to the remaining mem• indices. The analysis of the 37-taxon 18S rDNA bers of the order. In contrast to traditional classi• data set yields trees generally concordant with fications, sequence data indicate that Opiliaceae those obtained following analysis of the 62- are distinct from and evolutionarily more de• taxon matrix. Trees summarizing family-level rived than Olacaceae. Using rbeL data alone, relationships derived from the 18S rDNA and Opiliaceae are intercalated between a group of rbeL analyses are generally quite similar and New World and Old World Santalaceae (the lat• differ mainly in the placement of Opiliaceae ter including Eremolepidaceae). (Fig. 8.5). Given the overall concordance be• Analyses of 18S rDNA and rbeL sequences tween the topologies, the data sets were com• separately and in combination show that Santa• bined and analyzed together. The topology of laceae are not monophyletic but a grade that cul• the combined-analysis phylogram is congruent minates in Viscaceae (Fig. 8.5). Two groups of with that obtained from the separate 18S rDNA Santalaceae are resolved representing mainly data set, including the position of Opiliaceae; New World genera (Aeanthosyris, Buckley, hence it will be used in the following discussion Jodina, and Pyrularia) and Old World genera of relationships. (Exoearpos, Osyris, and Santalum) plus Ere• In agreement with traditional phylogenetic molepidaceae (Antidaphne, Eubrachion, and systems, Olacaceae represent the least derived Lepidoceras). These latter three genera are not family in the order. Despite limited generic sam• monophyletic using either molecular data set, pling, neither 18S nor rbeL phylograms indicate thus supporting the concept of Wiens and Barlow that the family is monophyletic (Fig. 8.5). In the (1971) who used karyological and morphologi• combined 18SIrbeL analysis, Sehoepfia is sister cal evidence to suggest that Eremolepidaceae are to Misodendrum, and this clade is sister to Lo• aerial parasites of Santalaceae. Viscaceae are the ranthaceae. Anatomical evidence supporting the most derived family of the order, and bootstrap distinctiveness of Sehoepfia from other Olaca• support for the clade is high using either 18S or ceae include its possession of aliform-confluent rbeL sequences. Molecular analyses indicate that parenchyma (Sleumer, 1984) and tracheidlves• Viscaceae is distinct from Loranthaceae, as was sel features (Reed, 1955). The latter study also previously shown by data derived from biogeo• noted similarities in the pollen of Sehoepfia with graphy, floral morphology, and cytology (Barlow, the more derived family Santalaceae. Taken as a 1964; Barlow and Wiens, 1971; Wiens and Bar• whole, it is worth considering possible phyloge• low, 1971; Barlow, 1983). Strongly supported in• netic relationships between Sehoepfia and gen• tergeneric relationships are obtained only for era of Loranthaceae that are root parasites. Bio• PhoradendronIDendrophthora and Korthalsellal geographic information and the long branch Ginalloa. Despite complete generic sampling in connecting Sehoepfia with Misodendrum sug• the family, 18S and rbeL sequence data, and ac• gest the latter genus may represent a relictual celerated substitution rates, relationships among taxon that evolved on the Gondwanan landmass all genera are not fully resolved, suggesting that along with the aerial parasites of Loranthaceae. the evolution of the viscaceous genera may rep• The three genera of Loranthaceae used for the resent a "hard polytomy" (Maddison, 1989), that combined analysis form a clade (100% boot• is, a true rapid radiation. strap) that is sister to the MisodendrumlSehoepf• The foregoing discussion demonstrates that ia clade. Analysis of a larger 18S data set (23 of analyses of a nuclear and plastid gene are con• the 74 genera, data not shown) resolves two cordant in resolving relationships within Santa• clades, one composed of New World and the lales. In addition, hypotheses proposed by the other of Old World mistletoes (Nickrent and three traditional classification systems can now Duff, 1996). A more rapidly evolving gene is re• be evaluated in light of new evidence. For ex• quired to provide sufficient phylogenetic signal ample, it is apparent that aerial parasitism has to address relationships among all genera of this evolved in Santalales in at least four separate PARAsmc PLANTS 225

___ Phoradendron serotinum 18S rONA + rbeL 100 ~~~_~ Phoradendron callfomicum Dendropthora clavata Korthalsel/a latlss/ma Korthaisella l/ndsayVcomplanata Glnalloa arnottlana ..___ -t=NotOthixos leiophyllus Notothixos subaureus ___ Vlscum album ..iiiI_ Vlscum articulatum 56 .._____ .. __ Arceuthobium vertlcil/if/orum ,.:!5:i!,5.. __ Eubrachlon amblguum * Lep/doceras chilense * 80 r:=:::I- Ant/daphne vlscoldes * --,.__ Osyris lanceolata Santalum album .....____ Exocarpos bidwllliVaphylla Jod/na rhombifolia Acanthosyris falcatalasipapote _-- Pyrularla pubera 52 _---- Buckleya dlstichophyl/a ---- Thesium impeditum ....._____ Comandra umbel/ata CansJera leptostachya Opilia amentacea Champereia manillana Agonandra macrocarpa ..__ Moquln/ella rubra TupeJa antarctica Loranthaceae Gaiadendron punctatum ..._____ Misodendrum brachystachyum Misodendraceae '-____ Schoepfia schreberi 1_--- Strombosla phlllpp/nensis He/steria conc/nna 1--_ Minquartia guianensis Olacaceae 1-___ Ximen/a americana ....______Olax aphylla ....______Alnus glutinosa outgroup 20 steps

18S rONA rbeL

Viscaeeae .----- Viseaeeae Santalaeeae Santalaceae + 'Eremolepidaceae' + 'Eremolepldac:eae' Opillaceae Oplliaeeae '---- Santalaeeae Loranthaeeae Loranthaeeae Misodendraeeae Misodendraceae Schoepfia 1.-____ Olacaeeae Schoepfia

I.-_____ outgroup ""'------Olaeaeeae '------outgroup

Figure 8.5. Strict consensus phylogram of 11 trees of length 2,212 derived from a parsimony analysis of nuclear 18S rONA and rbcL sequences combined from 37 taxa of Santalales and Alnus (outgroup). For taxa with two specific epithets, 18S rONA was sequenced from the first species and rbcL from the second. The phylograms enclosed as insets indicate the topologies of trees derived from analyses of the 18S rONA and rbcL sequences separately. Numbers above the nodes indi• cate bootstrap values (from 100 replications). Nodes without values were supported in less than 50% of the replications. See Table 8.2 for additional details pertaining to this analysis. The genera indicated by an asterisk have traditionally been classified in Eremolepidaceae but are here considered part of Santalaceae s. 1. The multiple origins of aerial parasitism are indicated by taxa with thick lines. 226 MOLECULAR SYSTEMATICS OF PLANTS lineages (Misodendraceae, Loranthaceae, "Ere• with difficulty owing to the extreme reduction molepidaceae," and Viscaceae; thick lines on and/or modification of morphological structures Fig. 8.5). Aside from such phylogenetic implica• that have accompanied the evolution of these lin• tions, what molecular evolutionary insights eages. Although sequences of rbcL have proved might be gained by comparing Santalales to extremely valuable in elucidating relationships other parasitic plants? In contrast to Scrophulari• throughout the angiosperms (see Chapter 17), this aceae, the holoparasitic trophic mode has appar• gene does not amplify for the nonasterid holopar• ently never evolved among the diverse lineages asites using standard PCR procedures (D. Nick• of root and stem parasitic Santalales. Although rent, unpubl.). Because nuclear 18S rONA se• Olacaceae have been proposed as the evolution• quences have proved useful for addressing ary origination point for Balanophoraceae (Cron• phylogenetic relationships in other parasitic quist, 1981), such an association is not supported plants, data from this molecular marker were ob• by nuclear 18S rONA sequence analyses (see tained for representatives of all nonasterid next section). If such a relationship is real, then holoparasites (Fig. 8.1). Relative rate tests no intermediate (less derived) taxa exist that showed that 18S sequences in these holoparasites might provide a link between the two families. It were evolving, on average, 3.5 times faster than is noteworthy that, like Balanophoraceae, base nonparasitic and most hemiparasitic plants (Nick• substitution rates in 18S rONA are increased rent and Starr, 1994), thus complicating their use among Viscaceae in general and Arceuthobium in phylogenetic analyses. When these divergent in particular. For the latter genus, base substitu• sequences were included in analyses with non• tion rates are also increased for rbcL Signifi• parasites, the resulting topologies frequently cantly, only about 30% of the carbon needed for showed aberrant relationships attributed to "long• growth is fixed in shoot tissue of Arceuthobium branch attractions" (Felsenstein, 1978). Similar (Hull and Leonard, 1964). Because these para• results were seen using parsimony, neighbor-join• sites are obtaining a large portion of their photo• ing, and maximum likelihood methods. synthate from the host, can we expect the even• Long-branch artifacts are sometimes the result tual evolution of holoparasitism? It appears that of incomplete sampling where intermediate lin• such a trophic mode is not compatible with the eages that could serve to "break up" a long aerially parasitic habit. Seedlings of Arceutho• branch have been omitted (Chase, Soltis, Olm• bium (Tocher et al., 1984) and those of other stead et al., 1993). Recently, increased taxon mistletoes are actively photosynthetic and must density for nuclear 18S rONA has been achieved exist autotrophically prior to attachment to the such that, at present, more than 400 complete nu• host branch. Such is not the case for root clear 18S sequences exist for angiosperms holoparasites that germinate in response to (Nickrent and Soltis, 1995; Soltis, Soltis, Nick• chemical stimulants (carried in a moist rhizo• rent et al., 1997). A 223-taxon data set (that in• sphere) and attach to closely proximal host roots cluded a greater sampling of monosulcate taxa (Press and Graves, 1995). Mistletoe seeds do not than previous studies) allowed a reexamination undergo dormancy, are not adapted to long-term of the placement of the nonasterid holoparasites. storage, and germinate in the absence of chemi• Sequences from the following holoparasites cal cues; therefore, attachment to the host must were added to the alignment: Balanophorales be rapid. These features impose critical evolu• (Balanophora jungosa, Corynaea crassa, Helo• tionary constraints upon mistletoes that do not sis cayennensis,. Ombrophytum subterraneum, exist for root holoparasites. Rhopalocnemis phalloides, and Scybalium jaimaicense), Cynomoriales (Cynomorium coc• Phylogenetic Studies of the Nonasterid cineum), Cytinales (Cytninus ruber), Raffiesiales Holoparasites Using Nuclear 18S rONA (Rafflesia keithii and Rhizanthes zippelii), and Sequences Hydnorales (Hydnora africana and Prosopanche americana). Given the size of this data set, only As discussed previously, traditional means of heuristic search strategies could be used and classifying the nonasterid holoparasites have met branch swapping did not go to completion. PARASITIC PLANTS 227

The tree resulting from this analysis (not MOLECULAR EVOLUTIONARY STUDIES shown) retained the major topological features OF PARASITIC PLANTS found in the more extensive analysis reported in Soltis, Soltis, Nickrent et al. (1997). None of the Molecular evolutionary processes occurring in nonasterid holoparasites were associated with parasitic plants represent some of the most ex• Santalales. As seen in previous analyses, Raffle• treme and dynamic known among all angio• siales and Balanophorales exhibited long• sperms. As evidenced by advances made from branch attractions, for example Rafflesia and studying Epifagus, Conopholis, and Cuscuta, Rhizanthes were intercalated in the middle of these plants can serve as important genetic mod• Balanophorales-an unlikely relationship. This els for facilitating the characterization of genome composite clade was placed with taxa allied evolution. It is therefore worthwhile to compare with Saxifragales, an equally unlikely relation• and contrast genetic processes found in evolu• ship. This analysis showed that Cytinus and Raf• tionarily distinct groups in an attempt to uncover flesia were only distantly related, thus support• common themes, mechanisms, and evolutionary ing the segregation of Cytinaceae as a distinct trends. As shown below for RUBISCO, fine• family of Rosidae s. 1. (Takhtajan et aI., 1985; scale examination of the types of mutations that Takhtajan, 1987). The disposition of the remain• have occurred at the molecular level help illumi• ing families of Rafflesiales has been hampered nate the progression of change among diverse by the lack of nuclear 18S rDNA sequences members of Scrophulariaceae. from Apodanthaceae, Mitrastemonaceae, and Bdallophyton. 18S rDNA sequences of repre• Structural Analysis of RUBISCO Large sentatives of Balanophorales show that Cyno• Subunit Protein in Scrophulariaceae moriaceae differ markedly (more than 100 steps) from other members of the order, thus The structure and function of the RUBISCO supporting the segregation of this family as pro• large subunit has been explored in great depth posed by Tahktajan (1987) and Thorne (1992). (Roy and Nierzwicki-Bauer, 1991; van der Vies In large- and small-scale analyses, Hydno• et al., 1992; Schreuder et al., 1993; Adam, 1995; raceae are consistently placed as sister to the Kellogg and Juliano, 1997 and references "paleoherb I" clade (sensu Donoghue and therein). A comparative analysis of the amino Doyle, 1989), that is, Aristolochiaceae, Chloran• acid substitutions of 499 flowering plants (Kel• thaceae, Lactoridaceae, Piperaceae, all members logg and Juliano, 1997) reveals a conservative of Magnoliidae. In contrast to Balanophorales mode of evolution of the protein in that there is and Rafflesiales, 18S rDNA rate increases for a high degree of toggling among biochemically Hydnoraceae are less than half that seen in other similar amino acids along the length of the holoparasites such as Rafflesia and Bal• polypeptide. The important structural motifs are anophora; hence, long-branch attraction ap• particularly conserved in most flowering plants. pears to be less problematic. Furthermore, when To examine the effect of relaxed functional con• taxa such as Rafflesia and Balanophora are in• straints on the evolution of the RUBISCO large cluded in the same analysis as Hydnoraceae, the subunit in parasitic Scrophulariaceae, several association of the latter family with the paleo• approaches were used, including: (1) an analysis herbs is retained. The relationship between Hyd• of the synonymous and nonsynonymous substi• noraceae and Aristolochiaceae is concordant tution rates (Table 8.3); (2) mapping amino acid with the concepts proposed by Solms-Laubach substitutions onto the strict consensus tree using (1894), Harms (1935), and more recently MacClade 3.0 (Maddison and Maddison, 1992); Cocucci (1983). The results with Hydnoraceae (3) a PAM250 analysis (Dayhoff et al., 1978); demonstrate the utility of molecular phyloge• and (4) assessment of substitutions in important netic methods in providing independent evi• structural motifs (Table 8.4); Wolfe and dePam• dence useful in resolving long-standing prob• philis, 1997). lems in the higher-level classification of these Average synonymous (K) and nonsynony• unusual holoparasites. mous (K) substitution rates and K/Kn for rbcL 228 MOLECULAR SYSTEMATICS OF PLANTS

Table 8.3. Average substitution rates of rbeL sequences for parasitic Scrophulariaceae and average PAM250 scores for RUBISCO large subunit polypeptide. Ka Ka Parasitic Plant Category No. Species S N K/KN PAM25()b Nonparasitic 16 0.2022 0.0281 7.19 1113 Photosynthetic hemiparasitic 7 0.2227 0.0223 9.98 1124 Nonphotosyntheticc hemiparasitic 11 0.2463 0.0270 9.12 1112 HoloparasiticJ 6 0.2482 0.0450 5.52 902

-Synonymous (Ks) and nonsynonymous (KN) substitution rates based on Jukes-Cantor corrections detennined using the computer program MEGA (Kumar et aI., 1993). bNieotiana tabaeum as a reference protein. e Hemiparasitic lineages with holoparasitic members and holoparasitic rbeL ORFs (see Fig. 8.3). d rbeL pseudogene sequences only.

from different categories of parasitic Scrophu• The rbcL nucleotide sequences were trans• lariaceae are listed in Table 8.3. As heterotrophy lated to inferred protein sequences. Deletions increases in Scrophulariaceae, Ks also increases. were maintained in the pseudogene sequences, In contrast, Kn does not appear to be affected but insertions were removed to retain the accu• greatly until the selective constraint is removed racy of the amino acid alignment. The step and pseudogenes are formed. There is a bias to• changes along the length of the polypeptide ward nonsynonymous substitutions in parasitic were plotted against the strict consensus tree plants with rbcL open reading frames, as evi• based on rbcL sequences using MacClade. denced by the higher K/Kn ratios compared to The primary finding of this analysis is the ap• nonparasitic Scrophulariaceae. However, the av• parent loss of conserved sequences along the erage K/Kn for rbcL pseudogene sequences is RUBISCO large subunit polypeptide with in• much lower, suggesting a relatively higher rate creasing heterotrophy. For nonparasitic Schroph• of nonsynonymous substitutions after selection ulariaceae, there are long sequences in which no is no longer a factor. amino acid replacements are observed. This

Table 8.4. PAM250 analysis of important rbeL structural motifs for representative Scrophulariaceae.·

Structural Motifs

No. of Between Dirner- Active PAM250 Taxon Events Ol ~ Intradimer Subunits dirner Site Other Difference Hoioparasites Orobanche ramosa 20 6 3 2 2 0 9 103 Boschniakia hookeri 13 4 2 0 3 0 1 4 64 Boschniakia strobilacea 10 2 2 0 2 0 1 3 57 Hyobanche sanguinea 7 3 1 0 3 41 Hyobanche atropurpurea 6 4 0 0 25 Other Holoparasites Lathraea clandestina 6 1 0 0 1 0 4 32 Orobanche corymbosa 2 0 1 0 0 0 0 1 5 Orobanche fasciculata 0 0 0 0 0 0 2 Herniparasites Alectra sessilijlora 0 0 0 0 0 0 1 6 Striga hermonthica 0 1 0 0 0 0 0 8 Nonparasites Alonsoa unilabiata 0 1 0 0 0 0 0 3 Chelone obliqua 0 0 0 0 0 0 5

-See text for further explanation and interpretation. PARASITIC PLANTS 229 same pattern was found by Kellogg and Juliano acid substitution in the other taxa. The total (1997) for 499 rbcL sequences from the Chase, PAM250 scores for the entire polypeptide and Soltis, Olmstead et al. (1993) study. Kellogg and for each critical structural motif were compared Juliano (1997) noted that important structural between the reference protein and each taxon. domains were conserved across a wide sampling Although the amount of amino acid replacement of flowering plants and that there were addi• increases with increasing heterotrophy, the over• tional conserved sequences of no known func• all effect of these replacements on the protein tion. Hemi- and holoparasitic plants of Scrophu• structure as inferred from the PAM250 analysis lariaceae have fewer conserved regions across is minimal. The only PAM250 scores signifi• the length of the RUBISCO large subunit cantly different from the reference protein are polypeptide. However, it is important to note pseudogene sequences. The translated protein that many biochemically similar amino acid re• sequences from the holoparasite rbcL ORFs are placements (e.g., replacing leucine with nearly identical to those of photosynthetic plants isoleucine) do not affect the structure and/or in terms of their inferred structure. Hence, de• function of the protein and are, therefore, neutral spite increased synonymous and nonsynony• replacements. Plotting the amino acid replace• mous substitutions in nonphotosynthetic hemi• ments against the strict consensus tree gives a parasites, the functionality of RUBISCO does first indication of molecular evolution of the not appear to be negatively impacted, which im• protein, but does not elucidate the effect of mol• plies that RUBISCO is functional in parasitic ecular evolution on the structure and/or function plants even after the loss of photosynthesis (see of the protein. next section). To assess the effect of amino acid replace• ments on the structure and function of the Plastome Reduction RUBISCO large subunit polypeptide, Wolfe and in Holoparasitic Plants dePamphilis (1995, 1997) conducted a PAM250 analysis (Dayhoff et al., 1978) on the translated For those holoparasites whose plastomes have protein sequences for selected taxa from all cat• been more fully characterized at the sequence egories of parasitic plants in Schrophulariaceae. level, the loss of photosynthesis is accompanied Dayhoff et al. (1978), using empirical data for by the loss or modification of photosynthetic many proteins (hemoglobins, fibrinopeptides, genes, a process that reduces the overall genome cytochrome c), constructed a table (the PAM250 size. Empirical evidence for genome reduction matrix) of amino acid substitution probabilities; can be seen in Scrophulariaceae s. 1., Cuscuta, in other words, a table that shows the probabili• and the nonasterid holoparasites. The genes re• ties of the substitution of one amino acid for an• sponsible for plastid gene expression (rONA, ri• other. Amino acids that have similar biochemi• bosomal proteins) are most often retained, as ev• cal properties are more likely to be substituted idenced by data on more extreme holoparasites with one another than with biochemically dis• such as Cytinus. In Epifagus, plastid-encoded similar amino acids, and this is reflected in the tRNA genes appear more labile as evidenced by values given in the PAM250 matrix. These sub• their loss and apparent replacement by nuclear• stitution probabilities are measured in terms of encoded tRNAs (Wolfe et aI., 1992b). Can the PAM (percent accepted mutations) with one unit reductional trend be taken to the extreme where representing one amino acid substitution per 100 the plastid and/or its plastome is lost entirely? residues. Based on ultrastructural studies·of holoparasite Using the reference protein from the out• plastids, Dodge and Lawes (1974) concluded group, Nicotiana tabacum, PAM250 scores that plastids are retained as storage organs or for were calculated over the entire length of the other functions. Because the plastome of Epi• polypeptide and from the critical structural mo• fagus contains only four genes that are not tifs (Tables 8.3, 8.4) by assigning a PAM250 involved in gene expression (dePamphilis value for each amino acid of Nicotiana and and Palmer, 1990; Wolfe et al., 1992c), anyone comparing the PAM250 score for each amino may serve some indispensable function in the 230 MOLECULAR SYSTEMATICS OF PLANTS nonphotosynthetic plastid, thus explaining the et al., 1986, 1991; Machado and Zetsche, 1990; retention of the entire plastome. Haberhausen and Zetsche, 1992), the rbeL locus Studies of the rbeL locus in holoparasitic is minimally expressed in terms of gene tran• Scrophulariaceae (Fig. 8.6) are especially infor• scription and/or protein activity. No rbeL ex• mative in that they show a range in the degree of pression was detected in several species of structural modifications experienced by differ• Orobanehe (Thalouarn et aI., 1994); however, ent lineages, thereby implying that such changes such studies have not been conducted on species occur independently and result in different de• with intact rbeL ORFs (e.g., O. eorymbosa or grees of pseudogene formation. It is of interest O. Jascieulata). An examination of the 5' and 3' that the distribution of rbeL pseudogenes on the untranslated regions (UTRs) of the rbeL locus in phylogeny of Scrophulariaceae (Fig. 8.3) is dif• Orobanehe (Wolfe and dePamphilis, 1997) re• ferent among Old and New World species of vealed that the promoter and/or ribosome bind• Orobanehe (the Old World species Orobanehe ing sites are maintained in O. eorymbosa and O. eernua and O. ramosa have pseudo genes, Jascieulata, but the 5' UTRs of the pseudogene whereas the New World O. eorymbosa and sequences have structural mutations or nu• O. Jascieulata do not). It is therefore important cleotide substitutions. The 3' UTRs of all to determine not just the presence of rbeL genes species of Orobanehe examined to date (Wolfe but also whether they are expressed in holopara• and dePamphilis, 1997) have intact stem-loop sites. structures but have major deletions upstream of It is clear from recent studies of parasitic the palindromic sequence. Phylogenetic analy• Scrophulariaceae and Cuscutaceae (Machado ses, gene expression studies, and sequence and Zetsche, 1990; Haberhausen et aI., 1992; analyses of rbeL all suggest that the retention of Delavault et aI., 1995; Wolfe and dePamphilis, rbeL ORFs in O. eorymbosa, O. Jascieulata, and 1997) that rbeL ORFs are maintained in many in other lineages of holoparasitic Scrophulari• independent lineages of parasitic plants. In two aceae can be explained by the recency of their holoparasites, Lathraea (Bricaud et aI., 1986; transition to holoparasitism. Thalouarn et aI., 1989; Thalouarn et aI., 1991; Wolfe and dePamphilis (1997) presented Delavault et aI., 1995, 1996) and Cuseuta (Press three hypotheses to explain the retention of rbeL

Nicotiana _ • • _____• .. .. . tabacum __ •• Hyobanche .. .. . atropurpurea _. ----- __ •• Hyobanche _. _____ 1 ... 1 __ • sangulnea * •• Boschniakia _ hookerl A I I. Boschniakla _____K!pili_ AA A T strobl/acea A A I. Orobanche ______K_ M X A .. _._- • cemua A A A Orobanche ramosa • A K A M~ EX - A I I I I I I I I I I I I I I I I 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 Length (kb) Figure 8.6. Structural maps of rbeL pseudogenes from orobanchaceous holoparasites compared with Nieotiana tabaeum. Solid black lines indicate sequence not affected by structural changes in most other taxa. Blank regions demarcate insertions from other taxa. Open triangles represent multinucleotide insertions, " = single nucleotide insertions. Gray lines are multinucleotide deletions, * = single nucleotide deletion. Note that the rbeL locus of Hyobanehe atropurpurea lacks structural mutations, but is a pseudogene, presumably because of a premature stop codon. Conopholis and Epifagus rbeL loci are not shown (missing in Conopholis, unalignable fragment in Epifagus). PARASITIC PLANTS 231 in holoparasitic plants of Scrophulariaceae and stop codon is 4/63. Under these assumptions, Cuscutaceae: (1) the plant still requires the RU• and if all mutations accumulating in the rbcL se• BISCO protein for low levels of autotrophic car• quences are independent, then, if the sequences bon fixation; (2) the oxygenase activity may are pseudogenes, the probability of retaining an function in glycolate metabolism; and (3) its ORF without a stop codon is: P = (59/63)n, maintenance or loss is a result of stochastic where n is the number of nucleotide substitution events. For the latter, the presence or absence of differences between them. Using this probabil• the gene in closely related members of a lineage ity model, Wolfe and dePamphilis (1997) calcu• may simply reflect the fact that insufficient time lated a probability of 2.79%; therefore it is has passed for the accumulation of deleterious unlikely that chance alone explains the main• mutations. Whereas the carboxylase activity of tenance of the rbcL ORF in O. corymbosa and RUBISCO fixes carbon dioxide into reduced O. Jasciculata. carbon (which is subsequently metabolized by As suggested by the differential occurrence the plant), the oxygenase activity is a competi• of pseudogenes in different species of Oro• tive reaction involved in glycine and serine banche, the short time scale for evolutionary di• biosynthesis (glyoxylate metabolism). Holopar• versification may explain differences in the asitic plants receive reduced carbon from their plastome structure of Epifagus and Conopholis, host, eliminating the need for RUBISCO activ• shown to be sister taxa by nuclear 18S rONA, ity (Press et ai., 1991); however, some species plastid 16S rONA, and rps2 data. The 16S123S maintain very low levels of photosynthetic pig• rONA spacer of Epifagus contains two nearly ment production and photosynthesis, or perform full-length tRNA pseudogenes whereas these photosynthesis only during a discrete phase of genes are essentially absent in the spacer of the life cycle such as seedling development or Conopholis, thus suggesting a recent loss. host attachment. This may explain the retention Moreover, both 18S rONA and rps2 data indi• of a minimally expressed rbcL in the stem para• cate that Boschniakia is sister to the Epifa• site, Cuscuta reflexa (Haberhausen et al., 1992). gus/Conopholis clade, hence the ancestor of the It is unclear how important the glycolate path• latter two taxa was also a holoparasite (whose way is for amino acid biosynthesis in photosyn• plastome must have contained inverted re• thetic or parasitic plants. Press et al. (1986) re• peats). Therefore, the loss of one copy of the in• ported a decrease in photorespiration with verted repeat in Conopholis took place after its increasing heterotrophy. Although unlikely, it divergence from Epifagus, likely less than 5 is possible that the oxygenase activity of myr ago (Muller, 1981). The case of the malar• RUBISCO is maintained in parasitic plants until ial Plasmodium (an apicomplexan) contrasts the adaptation to heterotrophy is complete, and with such rapid plastome reorganizations. Api• acquisition of host amino acids reduces the ne• complexans are thought to have evolved from cessity for photorespiration. Glycolate metabo• photosynthetic chromophytes, hence the malar• lism has also been suggested as a possible cause ial plastome has been retained for possibly 800 for the retention of rbcL in the plastome of the myr or more (Escalante and Ayala, 1995). As nonphotosynthetic euglenoid Astasia longa with holoparasitic plants, the plastome of Plas• which has experienced deletions of nearly all modium has a compact organization and retains other photosynthetic genes (Gockel et ai., 1994). a specific suite of genes involved in gene ex• To determine whether stochastic factors were pression and inverted repeats. These two exam• the sole reason for maintenance of rbcL ORFs in ples demonstrate the complexity of selectional O. corymbosa and O. Jasciculata, Wolfe and de• factors involved in either maintaining or delet• Pamphilis (1997) developed a probability model ing major structural features of the plastome. Al• using the following assumptions: (1) O. corym• though the selective environments of malarial bosa and O. Jasciculata had a nonphotosynthetic parasites and plants are clearly quite different, ancestor, and all rbcL sequence divergence oc• both have followed parallel pathways in shap• curred without photosynthetic constraint; (2) the ing the overall plastome structure, and both sug• probability of random mutations generating a gest that this genome may contribute essential 232 MOLECULAR SYSTEMATICS OF PLANTS metabolic products required by the parasite hemiparasites show no indication of acceleration. (Wilson et aI., 1996). These data suggest that substitution rate accelera• tion of nuclear rDNA occurs only in lineages RATE ACCELERATIONS where photosynthesis is absent or diminished, thereby requiring an advanced state of nutritional One of the fundamental questions in evolution• dependence upon the host. It is not presently un• ary biology centers around the concept of the derstood why these changes in life history are molecular clock, specifically the application of a correlated with changes at nuclear rDNA loci, al• strict molecular clock that infers equal rates of though relaxation of selectional constraints on nucleotide substitution for organisms with equal rRNA structure and function, small effective pop• generation times (the generation-time effect). As ulation size, and molecular drive have been pro• has been shown for rbcL in monocots (Bousquet posed (Nickrent and Starr, 1994). et al., 1992; Gaut et aI., 1992), rps2 in parasitic Scrophulariaceae (dePamphilis et al., 1997), and Plastid Genes: rps2 and rbcL 18S rDNA in Santalales (Nickrent and Starr, 1994), a strict molecular clock cannot be univer• Phylogenetic comparisons of Epifagus to Nico• sally applied, at least for these loci and taxa. As tiana and more distant outgroups revealed longer discussed below in Rate Variation among branches leading to Epifagus for virtually all of Genomes and Lineages, rate accelerations occur the retained protein-coding genes and plastid-en• in genes from all three subcellular genomes in coded rRNA genes (Morden et aI., 1991; Wolfe parasitic plants, hence these organisms provide et al., 1992b, 1992c). Increased numbers of sub• ideal models for studying molecular evolution• stitutions were observed at both synonymous and ary questions such as the effect of rate hetero• nonsynonymous sites between the parasitic and geneity on phylogeny reconstruction. nonparasitic lineages. These studies provided the first indication that the plastome of Epifagus Nuclear 18S rRNA Genes may be subject to increased rates of molecular evolution, even at functionally constrained The first documentation of rate acceleration in genes. As shown on the rps2 phylogeny (dePam• nuclear I8S rDNA in plants was the study by philis et aI., 1997), the branch lengths for both Nickrent and Starr (1994) wherein relative rates Epifagus and Conopholis are considerably tests were conducted using one hemiparasite longer than those of other Scrophulariaceae, and (Arceuthobium) and four holoparasites (Bal• formal relative rates tests show that both have in• anophora, Prosopanche, Rajjlesia, and Rhizan• creased evolutionary rates relative to nonpara• thes). These tests showed that 18S rDNA se• sitic and most hemiparasitic lineages (dePam• quences in these holoparasites were evolving, on philis et aI., 1997). Also evolving at an average, at least three times faster than nonpara• accelerated rate are the ptDNAs of Orobanche sitic and most hemiparasitic plants. Rate hetero• ramosa (non synonymous only), the Buchnera• geneity of this magnitude has not been detected Striga-Cycnium clade (nonsynonymous sites among over 200 angiosperm sequences that have only), and Euphrasia (synonymous sites only). been examined (D. Nickrent, unpubI.); however, These complex patterns of rate acceleration indi• statistically significant rate increases have been cate not only that rate asymmetries in this group measured in some Scrophulariaceae, Cuscuta, are not uniquely tied to the loss of photosynthe• Pholisma (Lennoaceae), and mycoheterotrophic sis, but that they may have more than one distinct angiosperms (A. Colwell and D. Nickrent, un• underlying cause, resulting in rate accelerations pubI.). These four nonasterid holoparasites and at synonymous and non synonymous sites. the dwarf mistletoe represent four distinct orders, For rbcL, synonymous and nonsynonymous thus accelerated substitution rates have occurred substitution rates in nonparasitic and hemipara• independently in an ancestor of each of these lin• sitic (photosynthetic) Scrophulariaceae are very eages. In Santalales, accelerated rates are not uni• similar to those of all other flowering plants ex• versally seen, as many other nonparasites and amined to date (Wolfe and dePamphilis, 1995, PARASITIC PLANTS 233

1997). This similarity is probably the result of angiosperms given their overall sequence con• the strong functional constraints imposed on the servation (see Chapter 1). The 16S rDNA se• protein to maintain important structural motifs quences from five nonasterid holoparasites were such as the alpha-beta barrels forming the active included in a phylogenetic analysis with other site, the active site itself, intradimer and dimer• land plants, algae, and cyanobacteria to show dimer interactions, and intersubunit interactions their affinity to angiosperms and graphically (Kellogg and Juliano, 1997). Parasitic plants that demonstrate rate increases. The topology of the have lost photosynthesis are no longer under se• strict consensus phylogram (Fig. 8.7) retains the lective constraints to maintain the structure and major features obtained by others using different function of proteins involved in photosynthetic methods of phylogenetic reconstruction such as reactions. This has become evident in examina• distance-based methods and neighbor joining tion of holoparasitic Scrophulariaceae. For ex• (summarized by Nelissen et al., 1995). Despite ample, the entire rbcL locus is missing in their extremely long branches, the holoparasite Conopholis (Colwell, 1994), and only a remnant 16S rDNA sequences are most closely related to of the gene remains in Epifagus (dePamphilis other angiosperm plastid 16S rDNAs. This posi• and Palmer, 1990; Wolfe et al., 1992c). tion is significant because previous analyses us• ing nuclear 18S rDNA sequences of extreme Plastid Genes: 168 rONA holoparasites resulted in the migration of these long-branch taxa to the base of the tree (Nick• Plastid-encoded 16S rDNA sequences have sel• rent and Starr, 1994). Similar perturbations of dom been used in phylogenetic studies of the position of divergent taxa on mitochondrial

r---r--"~ Mitrastema yamamotoi 165 rONA ~ Hydnora africanB ...... ::._- Bdallophyton amerlcanum 30 steps .....::...- Cytinus ruber ....._- Cynomorlum coccineum Oryzs sativa ZNmays Pisum sstlvum Vlclataba Glycine max Daucus carota Nlcotlana tabacum fiuscuta reflexa 1 Eplfagus vlrglnlana Conopholls amerlcs=n:..:;a=--__a=ng.::;.:ioaperm..;.:;.;;=8 ___ --I Pinus thunbergll gymnosperm Psl/otum nudum Botrychium bltematum ] pteridophytes Isoetes me/anopoda Lycopodium dlgltatum lycophyta Sphagnum palustre Marchsntia polymorpha ] bryophytes origin Coleochsetae orbicularis ] chlorophyta8 plastids L..-____ Chlamydomonas reinhardii Pyrenomonas salina 135 Astasia longa ] cryptophyte ,----1 Euglena gracilis euglenophyte8 ~ '---- Pylaiella linoralis phaeophyta L..-__ Cyanidium caldaricum rhodophyte Cyanophora paradoxa glaucophyte L..-__ Anscystis nidulans '---- Anabaena sp. ] cyanobacteria Figure 8.7. Strict consensus phylogram of four minimum-length trees of 2,576 steps derived from a parsimony analysis of 30 plastid-encoded l6S rONA sequences and two cyanobacterial outgroup taxa (Anacystis and Anabaena). Numbers above the branches of the nonphotosynthetic taxa indicate number of substitutions (steps) and are shown in bold. The nonasterid holoparasite clade (arrow) is present within the angiosperms; however, intergeneric relationships are influenced by long• branch attraction artifacts. 234 MOLECULAR SYSTEMATICS OF PLANTS

16S rRNA trees has been observed (Olsen, on helices 6, 23-1, 29, and 48. Further molecular 1987). For this character-based analysis (parsi• and biochemical work is needed to determine mony) of the holoparasites, the types of whether these genes are expressed and whether changes, not simply the number of changes, ap• the rRNA is functional. parently determined the placement of the clade. To quantify rate heterogeneity among various On the other hand, interrelationships among the 16S rONA sequences, formal relative rates tests nonasterid holoparasites are likely artifactual were conducted. The parametric test described because topological positions simply reflect an by Wu and Li (1985) as implemented for nu• increasing number of substitutions. clear 18S rONA sequences in holoparasites by Most angiosperm 16S rONA sequences differ Nickrent and Starr (1994) was used. In the ab• by 2-3% when compared with tobacco, whereas sence of information on actual divergence times, the nonasterid holoparasites show an increas• this method employs a variance estimation to ingly greater number of mutations: Cynomorium determine whether substitution rates (K) be• (7.3%), Cytinus (8.0%), Bdallophyton (12.7%), tween two lineages differ. Because the time of Mitrastema (14.9%), Hydnora (19.4%), Pi• divergence of the holoparasites relative to lostyles (30.4%), and Corynaea (35.9%) (Nick• monocots and dicots is uncertain, Marchantia rent et aI., 1997a). The high sequence variation was chosen as the reference. Use of the phyloge• suggests that these 16S rONA sequences may netically closer Pinus as the reference did not possibly be pseudogenes. As with Cytinus, how• change the results. The magnitude of rate in• ever, these rRNAs can be folded into secondary crease among the nonasterid holoparasites structures and most retain the typical comple• matches the overall number of substitutional dif• ment of 50 helices (Nickrent et al., 1997a). The ferences described above with Cynomorium at single exception is the 16S rRNA of Pilostyles, one extreme and Hydnora at the other (Fig. 8.8). likely the most unusual16S rRNA structure yet The highly divergent sequences of Pilostyles documented; it not only lacks four helices (9, and Corynaea would show K values even greater 10, 11, and 37), but also contains large insertions than Hydnora and are therefore not shown. The

250 Test Nicotiafla Marchafl'ia 1 2 3

200 ., ...0 K3 .-.... 150 ~'U N ~ I 100 :;! ...... ~ 50

0

II) ~ til ~ ~ .! e t: III e .5 ~ ~ ~ .5 ~ u 01 0 u ,5 e 0 Q, ..c: '§ ~ .l2 ~ ~ g. ~ ..c: III -6 <:J (,) e ~ g. e ~ t: 0 :::: ~ 8 t: ~- ~ ~en

Figure 8.8. Histogram showing results of relative rate tests using plastid 16S rDNA sequences. The II land plants indicated on the abscissa are the test organisms (taxon 1), Nicotiana was taxon 2, and Marchantia was the reference or outgroup (taxon 3). The three-taxon tree uses K" the number of nucleotide substitutions per site for taxon 1, K2 for taxon 2, and K3 for taxon 3. The difference in nucleotide substitutions per site (K' 3 - K23 ) is multiplied by 1,000 for graphical purposes. Standard error values are included above the bar for each taxon. The fastest rates are observed in the five nonasterid holoparasites. PARASITIC PLANTS 235

16S rONA sequences of Epifagus and Conopho• and Nickrent, submitted) that include four lis exhibited slightly accelerated rates relative to holoparasites ·(Cytinus, Hydnora, Corenaea, most angiosperms (in agreement with Wolfe et Epifagus), and Nicotiana. These sequences were al., 1992a), but not of the magnitude seen in the aligned with six published sequences and com• nonasterid holoparasites. pared with Glycine (Table 8.5). As with the 18S and 16S rONA data, Cytinus exhibited the Mitochondrial rONA smallest degree of sequence divergence (2.3%) followed by Comaea (2.8%) and Hydnora Increased nucleotide substitution rates among (3.1 %). The moderate divergence of the Epifa• both nuclear 18S rONA and plastid-encoded gus sequence (1.5%) is comparable to diver• 16S rONA of holoparasites raises the question gence levels of its plastid 16S rONA. "what is happening in the mitochondrial Formal relative rates tests using mitochondrial genome?" The plant mitochondrial genome has 19S rONA (similar to those performed on the been largely ignored as a source of data for phy• 18S and 16S rONA sequences) showed that only logenetic and population genetic studies of the nonasterid holoparasites had significantly in• plants (see Chapter 1), partly because the rate of creased substitution rates (Duff and Nickrent, synonymous nucleotide substitution in this 1997). Borderline rate increase was seen for genome is only one-third to one-quarter that of Epifagus. In contrast, rates of sequence evolution chloroplast DNA, despite the high frequency of for a second mitochondrial gene, cox], are not el• genetic recombination (Palmer, 1990, 1992). evated for Epifagus or for Cytinus relative to Small subunit sequences of mitochondrial nonparasitic relatives (C. dePamphilis et al., un• rONA (here designated 19S rONA) exhibit es• publ.). Unlike the 18S and 16S rONA sequences pecially low levels of divergence relative to of Cytinus, Hydnora, and Corynaea, which ex• their chloroplast and nuclear counterparts in an• hibit base composition biases, the mitochondrial giosperms. The six published photosynthetic an• sequences of these three genera show no such giosperm sequences (Table 8.5) exhibit less than bias. In addition to sequence divergence, these 1.3% sequence divergence. This value was de• sequences are characterized by a higher fre• termined from the "core" of the molecule, that quency of indels than sequences of nonparasitic is, excluding two variable (and unalignable) re• plants. Models constructed from these sequences gions (helices 6 and 43). Six new mitochondrial indicate that the majority of mutations do not dis• 19S rONA sequences have been generated (Duff rupt the higher-order structure, thus providing

Table 8.5. Angiosperm mitochondria119S rDNA sequences compared with Glycine.

GenBank Indelsb Helix 6 Helix 43 Species Number Substitutions· (bp) (bp) (bp) Secale cereale ZI4049 18 0 151 366 Triticum aestivum ZI4078 18 0 151 344 Zeamays X60794 20 1(14) 144 350 Oenothera berteriana X61277 10 3(3) 106 338 Glycine max M16859 102 427 Lepidoceras chilense U82641 9 1(1) 103 364 Lupinus luteus Z1l512 11 1(1) 103 461 Nicotiana tabacum U82638 15 0 102 335 Cytinus ruber: U82639 33 4(10) 99 330 Epifagus virginianaC U82642 22 2(2) 87 not sequenced Hydnora africana' U82637 41 3(4) 89 449 Corynaea crassa' U82636 40 3(39) 75 271

'Substitutions from comparisons excluding helix 6 and 43. 'Frequency and size of indels were determined by sequence alignment aided by secondary structure analysis based on a model for Zea (Gutell. 1993). cHoloparasitic taxa. 236 MOLECULAR SYSTEMATICS OF PLANTS strong indirect evidence that these genes are nuclear-encoded Adh and plastid-encoded rbcL functional. (Gaut et aI., 1996). Three basic models of The presence of RNA editing in plant mito• rate variation can be defined: (1) organismal, chondria has recently received increased atten• (2) genome-specific, and (3) gene-specific. The tion (Arts and Benne, 1996; Bowe and dePam• first model requires that DNA replication rate for philis, 1996; Sper-Whitis et al., 1996). Although each subcellular genome be correlated; hence if RNA editing is recognized as a significant one locus is accelerated, all loci will be similarly process in many mitochondrial genes, the affected. In the absence of selection, compar• holoparasite 19S rDNA sequences reported here isons of molecular phylogenies using (say) two are unlikely to be significantly affected as deter• different organellar loci would show each taxon mined by reverse-transcriptase peR experi• to have the same relative branch lengths on the ments (R. Duff and D. Nickrent, unpubl.). De• respective trees. In the genome-specific model, spite the numerous reports of editing among there is no assumption that a correlation exists protein-coding genes in plant mitochondria, no between DNA replication and/or substitution cases of editing of mitochondrial 195 rRNA are rates between organellar genomes. Such a model known (Schuster et al., 1991). Neither the pat• could result from differences in DNA replication tern of substitutions nor the presence of indels and/or repair mechanisms in the distinct organel• follows that seen in reported cases of RNA edit• lar genomes. Here, comparisons of molecular ing. For these reasons, we do not feel editing can phylogenies derived from the different organel• be used to explain the observed pattern of sub• lar loci would not be expected to show the same stitutions in mitochondriall9S rDNA. distribution of rate variation among taxa. Finally, the gene-specific model implies that each subcel• RATE VARIATION AMONG GENOMES lular genome, and possibly individual genes AND LINEAGES within each genome, are constrained by their own selectional environment. This model would The patterns of nucleotide substitution across be manifest as molecular phylogenies with pat• lineages and organellar gene loci provide a terns of rate variation that are not correlated be• framework for addressing the underlying causes tween gene loci. of rate variation. Relatively few studies exist in Sequences of parasitic plant genes derived which rate comparisons are made between dif• from all three subcellular genomes can be used ferent organellar genes across lineages. An ex• to address the above models and to examine the ception is recent work showing that synony• association between rate changes and loss of mous (but not nonsynonymous) substitution photosynthesis. As shown in Table 8.6, the rates are slower for palms than grasses for both patterns of rate variation are complex; some ex-

Table 8.6. Rate comparisons for parasitic plants across genomes.

Plastid Nuclear Mitochondrial Taxon I8S rDNA I6SrDNA rbcL rps2 I9SrDNA

Nonasterid holoparasites ++a ++ absent ? ++ Santalales (minus Viscaceae) ? Arceuthobium ++ + + ? ? Epifagus +/- + I\J + +/- Conopholis +/- + absent + ? Hyobanche + I\J ? Cuscuta +b +d ?

"Molecular evolutionary rates: + + = very fast, + = fast, +/- = borderline fast, - = not accelerated in relative rate tests, tjI = pseudogene. 'From Cuscula gronovii (A. Colwell and D. Nickrent, unpubl.). 'From Cuscuta reflexa (Haberhausen and Zetsche, 1994). dGene from Cuscuta reflexa present but with low transcription levels (Haberhausen et aI., 1992). PARASITIC PLANTS 237 amples appear to support one model, whereas tosynthesis show increased rates, as evidenced others do not. Consider first the nonasterid by rps2 in Alectra orobanchoides or Hy• holoparasites. Formal relative rate tests, per• obanche. Further support for the latter concept formed using sequences from nuclear, plastid, is seen following relative rate tests of nuclear and mitochondrial small subunit rDNA, re• rDNA in ericaceous mycotrophs (e.g., Mono• vealed that all three genomes are accelerated rel• tropa) or nonphotosynthetic orchids (e.g., ative to autotrophic plants. These results support Coraliorhiza) where no rate increases were ob• the organismal model of rate variation. served (A. Colwell and D. Nickrent, unpubl.). In other instances, the available data do not Results from rps2 analyses of Scrophulariaceae support the organismal model. Considering a indicate that rate increases can occur in hemi• particular gene locus, different substitution rates parasites prior to the loss of photosynthesis (de• can be observed among different parasites of the Pamphilis et aI., 1997), thus providing evidence same lineage. For 18S rDNA and rbcL, rates are supporting the recency and rapidity of this elevated for Viscaceae, but not for other families trophic change. of Santalales (Nickrent and Soltis, 1995; Nick• The relationship between rate accelerations in rent and Duff, 1996), demonstrating that rate ac• plastomes and the nuclear genome is not clear at celeration is not a general evolutionary feature present. One might predict that changes in the of the order but has occurred only following the plastome occur first and that the comparatively divergence of Viscaceae (see Fig. 8.5). Simi• more conservative nuclear ribosomal loci fol• larly, rate acceleration for rps2 can be seen in low. That the two events are linked is evidenced Epifagus, Conopholis, and Orobanche, but not by the paucity of cases, at least in parasitic in other holoparasitic Scrophulariaceae such as plants, where plastome genes are clearly accel• Hyobanche (dePamphilis et aI., 1997). Rate ele• erated but nuclear rDNA genes are not. Such vations may occur differentially among genes correlations suggest a greater degree of signal within a single genome. For example, Arceutho• transduction from the plastid to the nucleus than bium shows elevated rates for rbcL, but not for has previously been suspected (Susek and l6S rDNA, which indicates that elevated substi• Chory, 1992). tution rates can exist in genes of fully functional chloroplasts, but that conservative loci may lag CONCLUSIONS behind more variable ones. In Scrophulariaceae, the pattern of rate increase (or pseudogene for• Parasitic plants represent some of the most un• mation) for rbcL does not parallel that of l6S usual organisms on earth, and each independent rDNA, again, with the latter showing more con• lineage represents a natural genetic experiment servative rates of change. Finally, differences in whose organelles and genes have evolved under substitution rates may be observed between dif• relaxed functional constraints. Although Cus• ferent subcellular genomes within the same cuta is amenable to culture and direct experi• species. For Arceuthobium, Hyobanche, and mental manipulation, very few holoparasitic Cuscuta, rates for nuclear but not plastid small parasitic plants have been established as model subunit rDNA are elevated. These data do not systems for biochemical and genetic research. tend to support either the organismal or the Certainly, attempts to culture holoparasites genome-specific model of rate variation, thus would be hindered by lack of knowledge of leaving only the gene-specific model as an ex• their life histories. To date, studies of Santa• planation. lales, nona sterid holoparasites, Schrophulari• The relationship between the loss of photo• aceae s. 1., and Cuscuta have yielded a wealth of synthesis and rate accelerations in nuclear and new insights into molecular phylogenetic and plastid genes can be interpreted using the data molecular evolutionary processes. Despite this presented in Table 8.6. As shown by Arceutho• progress, a number of questions about the evolu• bium, not all plants with rate increases (for the tionary process still remain, which will hope• nuclear or plastid genomes) are nonphotosyn• fully stimulate continued study of these fascinat• thetic. Conversely, not all plants that lose pho- ing plants. 238 MOLECULAR SYSTEMATICS OF PLANTS

LITERATURE CITED Dayanandan, and V. A. Albert. 1993. Phylogenetics of seed plants: an analysis of nucleotide sequences from Adam, Z. 1995. A mutation in the small subunit of ribulose- the plastid gene rbcL. Annals of the Missouri Botani• 1,5-biphosphate carboxylase/oxygenase that reduces cal Garden 80:528-580. the rate of its incorporation into holoenzyme. Photo• Cocucci, A. E. 1983. New evidence from embryology in an• synthesis Research 43:143-147. giosperm classification. Nordic Journal of Botany An-Ming, L. 1990. A preliminary cladistic study of the fam• 3:67-73. ilies of the superorder Lamiiflorae. Botanical Journal Colwell, A. E. 1994. Genome evolution in a non-photosyn• of the Linnean Society 103:39-57. thetic plant, Conopholis americana. Ph.D. dissertation, Arts, G. 1., and R. Benne. 1996. Mechanism and evolution of Washington University, St. Louis, Missouri. RNA editing in kinetoplastida. Biochimica Biophysica Cronquist, A. 1981. An Integrated System of Classification Acta 1307:39-54. of Flowering Plants. Columbia University Press, New Barlow, B. A. 1964. Classification of the Loranthaceae and York. Viscaceae. Proceedings of the Linnaean Society of Cronquist, A. 1988. The Evolution and Classification of New South Wales 89:268-272. Flowering Plants. New York Botanical Garden, New Barlow, B. A. 1983. Biogeography of Loranthaceae and Vis• York. caceae. In The Biology of Mistletoes, eds. M. Calder Dayhoff, M. 0., R. M. Schwartz, and B. C. Orcutt. 1978. A and P. Bernhardt, pp. 19-45. Academic Press, New model of evolutionary change in proteins. In Atlas of York. Protein Sequence and Structure, ed. M. O. Dayhoff, Barlow, B. A., and D. Wiens. 1971. The cytogeography of pp. 345-352. National Biomedical Research Founda• the 10ranthaceous mistletoes. Taxon 20:291-312. tion, Silver Springs, Maryland. Bhandari, N. N., and S. C. A. Vohra. 1983. Embryology and Delavault, P., V. Sakanyan, and P. Thalouarn. 1995. Diver• affinities of Viscaceae. In The Biology of Mistletoes, gent evolution of two plastid genes, rbcL and atpB, in eds., M. Calder and P. Bernhardt, pp. 69-86. Academic a non-photosynthetic parasitic plant. Plant Molecular Press, New York. Biology 29:1071-1079. Boeshore, I. 1920. The morphological continuity of Scro• Delavault, P. M., N. M. Russo, N. A. Lusson, and P. A. phulariaceae and Orobanchaceae. Contributions of the Thalouarn. 1996. Organization of the reduced plastid Botanical Laboratory of the Morris Arboretum genome of Lathraea clandestina, an achlorophyllous 5:139-177. parasitic plant. Physiologia Plantarum 96: 674-682. Bommer, D., G. Haberhausen, and K. Zetsche. 1993. A large dePamphilis, C. W. 1995. Genes and genomes. In Parasitic deletion in the plastid DNA of the holoparasitic flow• Plants, eds. M. C. Press and 1. D. Graves, pp. 176-205. ering plant Cuscuta rejlexa concerning two ribosomal Chapman & Hall, London. proteins (rpI2, rpI23), one transfer RNA (tm!) and an dePamphilis, C. w., and 1. D. Palmer. 1989. Evolution and ORF 2280 homologue. Current Genetics 24: 171-176. function of plastid DNA: a review with special refer• Bousquet, J., S. H. Strauss, A. H. Doerksen, and R. A. Price. ence to nonphotosynthetic plants. In Physiology, Bio• 1992. Extensive variation in evolutionary rate of rbcL chemistry, and Genetics of Nongreen Plastids, eds. C. gene sequences among seed plants. Proceedings of the D. Boyer, J. C. Shannon, and R. C. Hardison, pp. National Academy of Sciences U.S.A. 89:7844-7848. 182-202. American Society of Plant Physiologists, Bowe, L. M., and C. W. dePamphilis. 1996. Effects of RNA Rockville, Maryland. editing and gene processing on phylogenetic recon• dePamphiJis, C. w., and 1. D. Palmer. 1990. Loss of photo• struction. Molecular Biology and Evolution synthetic and chlororespiratory genes from the plastid 13:1159-1166. genome of a parasitic flowering plant. Nature (Lon• Bricaud, C. H., P. Thalouarn, and S. Renaudin. 1986. Ribu• don) 348:337-339. lose 1,5-bisphosphate carboxylase activity in the dePamphilis, C. w., N. D. Young, and A. D. Wolfe. 1997. holoparasite Lathraea clandestina L. Journal of Plant Evolution of plastid gene rps2 in a lineage ofhemipar• Physiology 125:367-370. asitic and holoparasitic plants: many losses of photo• Burtt, B. L. 1965. The transfer of Cyrtandromoea from Ges• synthesis and complex patterns of rate variation. Pro• neriaceae to Scrophulariaceae, with notes on the clas• ceedings of the National Academy of Sciences U.S.A. sification of that family. Bulletin of the Botanical Sur• 94:7367-7372. vey 7:73-88. Dodge, 1. D. and G. B. Lawes. 1974. Plastid ultrastructure in Chase, M. w., D. E. Soltis, R. G. Olmstead, D. Morgan, D. some parasitic and semi-parasitic plants. Cytobiology H. Les, B. D. Mishler, M. R. Duvall, R. A. Price, H. G. 9:1-9. Hills, Y.-L. Qiu, K. A. Kron, 1. H. Rettig, E. Conti, J. Donoghue, M. J., and J. A. Doyle, 1989. Phylogenetic analy• D. Palmer, J. R. Manhart, K. J. Sytsma, H. J. Michaels, sis of angiosperms and the relationships of Hamameli• W. J. Kress, K. G. Karol, W. D. Clark, M. Hedren, B. dae. In Evolution, Systematics, and Fossil History of S. Gaut, R. K. Jansen, K.-J. Kim, C. F. Wimpee, 1. F. the HamameJidae, eds. P. R. Crane and S. Blackmore, Smith, G. R. Fumier, S. H. Strauss, Q.-Y. Xiang, G. M. pp. 17-45. Clarendon Press, Oxford. Plunkett, P. S. Soltis, S. M. Swensen, S. E. Williams, P. Downie, S. R., and J. D. Palmer. 1992. Use of chloroplast A. Gadek, C. J. Quinn, L. E. Eguiarte, E. Golenberg, DNA rearrangements in reconstructing plant phy• G. H. Learn, Jr., S. W. Graham, S. C. H. Barrett, S. logeny. In Molecular Plant Systematics, eds. P. S. PARASITIC PLANTS 239

Soltis, D. E. Soltis, and J. J. Doyle, pp. 14-35. Chap• Harms, H. 1935. Hydnoraceae. In Die Natiirlichen Pflanzen• man & Hall, New York. familien, eds. A. Engler and H. Harms, pp. 282-295. Duff, R. J., and D. L. Nickrent. 1997. Characterization of W. Engelmann, Leipzig. mitochondrial small-subunit ribosomal RNAs from Hong, D.-Y. 1984. Taxonomy and evolution of the Veron• holoparasitic plants. Journal of Molecular Evolution iceae (Scrophulariaceae) with special reference to pa• 45:631-639. lynology. Opera Botanica 75: 1-60. Ems, S. C., C. W. Morden, C. K. Dixon, K. H. Wolfe, C. W. Hull, R. J., and O. A. Leonard. 1964. Physiological aspects dePamphilis, and J. D. Palmer. 1995. Transcription, of parasitism in mistletoes (Arceuthobium and splicing and editing of plastid RNAs in the nonphoto• Phoradendron). 2. The photosynthetic capacity of synthetic plant Epifagus virginiana. Plant Molecular mistletoes. Plant Physiology 39: 1008-1017. Biology 29:721-733. Kellogg, E. A., and N. D. Juliano. 1997. The structure and Eplee, R. E. 1~81. Striga's status as a plant parasite in the function of RuBisCo and their implications for system• United States. Plant Disease 56:951-954. atic studies. American Journal of Botany 84:413-428. Escalante, A. A., and F. J. Ayala. 1995. Evolutionary origin Kuijt, J. 1968. Mutual affinities of Santalalean families. of Plasmodium and other Apicomplexa based on Brittonia 20:136-147. rRNA genes. Proceedings of the National Academy of Kuijt, J. 1969. The Biology of Parasitic Flowering Plants. Sciences U.S.A. 92:5793-5797. University of California Press, Berkeley. Fagerlind, F. 1948. Beitrage ziir Kenntnis der Gynaceum• Kumar, S., K. Tamura, and M. Nei. 1993. MEGA: Molecu• morphologie und Phylogenie der Santalales-Familien. lar Evolutionary Genetics Analysis, version 1.01. The Svensk Botanisk Tidskrift 42: 195-229. Pennsylvania State University, University Park. Felsenstein, J. 1978. Cases in which parsimony or compati• Li, W.-H. 1983. Evolution of duplicate genes and pseudo• bility will be positively misleading. Systematic Zool• genes. In Evolution of Genes and Proteins, eds. M. Nei ogy 27:401-410. and R. K. Koehn, pp. 14-37. Sinauer Associates, Sun• Freyer, R., K. Neckermann, R. M. Maier, and H. Kossel. derland, Massachusetts. 1995. Structural and functional analysis of plastid Machado, M. A., and K. Zetsche. 1990. A structural, func• genomes from parasitic plants: loss of an intron within tional and molecular analysis of plastids of the the genus Cuscuta. Current Genetics 27:580-586. holoparasites Cuscuta reflexa and Cuscuta europaea. Gaut, B., S. Muse, W. Clark, and M. Clegg. 1992. Relative Planta 181:91-96. rates of nucleotide substitution at the rbcL locus of Maddison, W. P. 1989. Reconstructing character evolution monocotyledonous plants. Journal of Molecular Evo• on polytomous cladograms. Cladistics 5:365-377. lution 35:292-303. Maddison, W. P., and D. R. Maddison. 1992. MacClade, Gaut, B. S., B. R. Morton, B. C. McCaig, and M. T. Clegg. Analysis of Phylogeny and Character Evolution, ver• 1996. Substitution rate comparisons between grasses sion 3.01. Sinauer Associates, Sunderland, Massachu• and palms: synonymous rate differences at the nuclear setts. gene Adh parallel rate differences at the plastid gene Mayfield, S. P., C. B. Yohn, A. Cohen, and A. Danon. 1995. rbcL. Proceedings of the National Academy of Sci• Regulation of chloroplast gene expression. Annual Re• ences U.S.A. 93:10274-10279. view of Plant Physiology and Plant Molecular Biology Gockel, G., W. Hachtel, S. Baier, C. Fliss, and M. Henke. 46:147-166. 1994. Genes for components of the chloroplast Minkin, J. P., and W. H. Eshbaugh. 1989. Pollen morphology translational apparatus are conserved in the reduced of the Orobanchaceae and rhinanthoid Scrophulari• 73-kb plastid DNA of the nonphotosynthetic eugle• aceae. Grana 28:1-18. noid flagellate Astasia longa. Current Genetics 26: Morden, C. w., K. H. Wolfe, C. W. dePamphilis, and J. D. 256-262. Palmer. 1991. Plastid translation and transcription Gutell, R. R. 1993. Collection of small subunit (l6S and genes in a nonphotosynthetic plant: intact, missing and 16S-like) ribosomal RNA structures. Nucleic Acids pseudo genes. European Molecular Biology Organiza• Research 21:3051-3054. tion Journal 10:3281-3288. Haberhausen, G., and K. Zetsche. 1992. Nucleotide se• Muller, J. 1981. Fossil pollen records of extant angiosperms. quences of the rbcL gene and the intergenic promoter Botanical Review 47:1-142. region between the divergently transcribed rbcL and Musselman, L. J. 1980. The biology of Striga, Orobanche, atpB genes in Ipomoea purpurea (L.). Plant Molecular and other root-parasitic weeds. Annual Review ofPhy• Biology 18:823-825. topathology 18:463-489. Haberhausen, G., and K. Zetsche. 1994. Functional loss of Nelissen, B., Y. Van de Peer, A. Wilmotte, and R. all ndh genes in an otherwise relatively unaltered plas• De Wachter. 1995. An early origin of plastids within the tid genome of the holoparasitic flowering plant Cus• cyanobacterial divergence is suggested by evolution• cuta reflexa. Plant Molecular Biology 24:217-222. ary trees based on complete l6S rRNA sequences. Haberhausen, G. K., Valentin, and K. Zetsche. 1992. Organ• Molecular Biology and Evolution 12: 1166-1173. ization and sequence of photosynthetic genes from the Nickrent, D. L., and R. J. Duff. 1996. Molecular studies of plastid genome of the holoparasitic flowering plant parasitic plants using ribosomal RNA. In Advances in Cuscuta reflexa. Molecular and General Genetics Parasitic Plant Research, eds. M. T. Moreno, J. I. 232:154-161. Cubero, D. Berner, D. Joel, L. J. Musselman, and 240 MOLECULAR SYSTEMATICS OF PLANTS

C. Parker, pp. 28-52. Junta de Andalucia, Direcci6n Reed, C. F. 1955. The comparative morphology of the Ola• General de Investigaci6n Agraria, Cordoba, Spain. caceae, Opiliaceae, and Octoknemaceae. Memorias da Nickrent, D. L., and C. R. Franchina. 1990. Phylogenetic re• Sociedade Broteriana 10:29-79. lationships of the Santalales and relatives. Journal of Roy, H., and S. A. Nierzwicki-Bauer. 1991. RuBisCO: Molecular Evolution 31:294-301. genes, structure, assembly, and evolution. In The Pho• Nickrent, D. L., and D. E. Soltis. 1995. A comparison of an• tosynthetic Apparatus: Molecular Biology and Opera• giosperm phylogenies based upon complete 18S rDNA tion, eds. L. Bogorad and I. K. Vasil, pp. 347-364. and rbcL sequences. Annals of the Missouri Botanical Academic Press, San Diego. Garden 82:208-234. Schreuder, H. A., S. Knight, P. M. G. Curmi, I. Andersson, Nickrent, D. L., and E. M. Starr. 1994. High rates of nu• D. Cascio, C.-I. Branden, and D. Eisenberg. 1993. For• cleotide substitution in nuclear small-subunit (18S) mation of the active site of ribulose-l,5-bisphosphate rDNA from holoparasitic flowering plants. Journal of carboxylase/oxygenase by a disorder-order transition Molecular Evolution 39:62-70. from the unactivated to the activated form. Proceed• Nickrent, D. L., Y. Ouyang, and C. W dePamphilis. 1995. ings of the National Academy of Sciences U.S.A. Presence of plastid genes in representatives of the 90:9968-9972. holoparasitic families Balanophoraceae, Hydnoraceae, Schuster, W, R. Ternes, V. Knoop, R. Hiesel, B. Wissinger, and Rafflesiaceae. American Journal of Botany and A. Brennicke. 1991. Distribution of RNA editing (suppl.) 82:75. sites in Oenothera mitochondrial mRNAs and rRNAs. Nickrent, D. L., R. J. Duff, and D. A. M. Konings. 1997a. Current Genetics 20:397-404. Structural analyses of plastid-encoded 16S rRNAs in Shimada, H., and M. Sugiura. 1991. Fine structural features holoparasitic angiosperms. Plant Molecular Biology of the chloroplast genome: comparison of the se• 74:731-743. quenced chloroplast genomes. Nucleic Acids Research Nickrent, D. L., Y. Ouyang, R. J. Duff, and C. W dePam• 19:983-995. philis. 1997b. Do nonasterid holoparasitic flowering Sleumer, H. O. 1984. Flora Neotropica. Olacaceae Mono• plants have plastid genomes? Plant Molecular Biology graph No 38. New York Botanical Garden, New York. 34:717-729. Solms-Laubach, H. 1894. Hydnoraceae. In Die Natiirlichen Olmstead, R. G., and P. A. Reeves. 1995. Evidence for the Planzenfamilien, Part Ill, eds. A. Engler and K. Pranti, polyphyly of the Scrophulariaceae based on chloro• pp. 282-285. W Engelmann, Leipzig. plast rbcL and ndhF sequences. Annals of the Missouri Soltis, D. E., P. S. Soltis, D. L. Nickrent, L. A. Johnson, W Botanical Garden 82:176-193. J. Hahn, S. B. Hoot, J. A. Sweere, R. K. Kuzoff, K. A. Olsen, G. J. 1987. Earliest phylogenetic branchings: com• Kron, M. W. Chase, S. M. Swensen, E. A. Zimmer, S.• paring rRNA-based evolutionary trees inferred with M. Chaw, L. J. Gillespie, W J. Kress, and K. J. Sytsma. various techniques. Cold Spring Harbor Symposium 1997. Angiosperm phylogeny inferred from 18S ribo• on Quantitative Biology 52:825-837. somal DNA sequences. Annals of the Missouri Botan• Palmer, J. D. 1990. Contrasting modes and tempos of ical Garden 84: 1-49. genome evolution in land plant organelles. Trends in Sper-Whitis, G. L., J. L. Moody, and J. C. Vaughn. 1996. Genetics 6:115-120. Universality of mitochondrial RNA editing in cy• Palmer, J. D. 1992. Mitochondrial DNA in plant systematics: tochrome-c oxidase subunit I (coxf) among the land applications and limitations. In Molecular Plant Sys• plants. Biochimica Biophysica Acta 1307:301-308. tematics, eds. P. S. Soltis, D. E. Soltis, and J. J. Doyle, Stewart, G. R., and M. C. Press. 1990. The physiology and pp. 36-49. Chapman & Hall, New York. biochemistry of parasitic angiosperms. Annual Review Pazy, B., U. Plitmann, and O. Cohen. 1996. Bimodal kary• of Plant Physiology and Plant Molecular Biology otype in Cynomorium coccineum L. and its systematic 41:127-151. implications. Journal of the Linnaean Society, London Susek, R. E., and J. Chory. 1992. A tale of two genomes: role 120:279-281. of a chloroplast signal in coordinating nuclear and Pennell, F. W 1935. The Scrophulariaceae of Eastern Tem• plastid genome expression. Australian Journal of Plant perate North America. Academy of Natural Sciences, Physiology 19:387-399. Philadelphia. Takhtajan, A. L. 1987. Sistema magnoliofitov [in Russian]. Press, M. C., and J. D. Graves. 1995. Parasitic Plants. Chap• Nauka, Leningrad. man & Hall, London. Takhtajan, A. L., N. R. Meyer, and V. N. Kosenko. 1985. Press, M. C., N. Shah, and G. R. Stewart. 1986. The parasitic Morfologiya pyl'tsy i klassifikatsiya semeystva Raf• habit: trends in metabolic reductionism. In Biology flesiaceae s. 1. Botanicheskii Zhurnal 70: 153-162. and Control of Orobanche, ed. S. J. ter Borg, pp. Taylor, G. W, K. H. Wolfe, C. W Morden, C. W dePam• 96-106. LHNPO, Wageningen, Netherlands. philis, and J. D. Palmer. 1991. Lack of a functional Press, M. C., S. Smith, and G. R. Stewart. 1991. Carbon ac• plastid tRNA'Y' gene is associated with loss of photo• quisition and assimilation in parasitic plants. Func• synthesis in a lineage of parasitic plants. Current Ge• tional Ecology 5:278-283. netics 20:515-518. Price, P. 1980. Evolutionary Biology of Parasites. Mono• Thalouarn, P., M.-C. Arnaud, and S. Renaudin. 1989. Evi• graphs in Population Biology No. 15. Princeton Uni• dence of ribulose-bisphosphate carboxylase in the versity Press, Princeton, New Jersey. Scrophulariaceae holoparasite Lathraea clandestina L. PARASITIC PLANTS 241

Comptes Rendu Academy of Science Paris 309: gene map of the plastid-like DNA of the malaria para• 275-280. site Plasmodium Jalciparum. Journal of Molecular Bi• Thalouarn, P., C. Theodet, and S. Renaudin. 1991. Evidence ology 261:155-172. of plastid and nuclear genes for the large and small Wimpee, C. E, R. L. Wrobel, and D. K. Garvin. 1991. A di• subunits of Rubisco in the Scrophulariaceae holopara• vergent plastid genome in Conopholis americana, an site Lathraea clandestina L. Comparison with the au• achlorophyllous parasitic plant. Plant Molecular Biol• totroph Digitalis purpurea L. and hemiparasite ogy 17:161-166. Melampyrum pratense L. Comptes Rendu Academy of Wimpee, C. E, R. Morgan, and R. L. Wrobel. 1992a. An Science Paris 312: 1-6. aberrant plastid ribosomal RNA gene cluster in the Thalouarn, P., C. Theodet, N. Russo, and P. Delavault. 1994. root parasite Conopholis americana. Plant Molecular The reduced plastid genome of a nonphotosynthetic Biology 18:275-285. angiosperm Orobanche hederae has retained the rbcL Wimpee, C. E, R. Morgan, and R. L. Wronel. 1992b. Loss of gene. Plant Physiology and Biochemistry 32:233-242. transfer RNA genes from the plastid 16S-23S riboso• Thieret, 1. W. 1967. Supraspecific classification in the Scro• mal RNA gene spacer in a parasitic plant. Current Ge• phulariceae: a review. Sida 387-106. netics 21:417-422. Thorne, R. E 1992. An updated phylogenetic classification Wolfe, A. D., and C. W. dePamphilis. 1995. Systematic im• of the flowering plants. Aliso 13:365-389. plications of relaxed functional constraints on the RU• Tocher, R. D., S. W. Gustafson, and D. M. Knutson. 1984. BISCO large subunit in parasitic plants of the Scro• Water metabolism and seedling photosynthesis in phulariaceae and Orobanchaceae. American Journal of dwarf mistletoes. In Biology of Dwarf Mistletoes: Botany (suppl.) 82:6. Proceedings of the Symposium, eds. E G. Wolfe, A. D., and C. W. dePamphilis. 1997. Alternate paths Hawksworth and R. E Scharpf, pp. 62-69. USDA, of evolution for the photosynthetic gene rbcL in four Rocky Mt. Forest & Range Experimental Station, Ft. nonphotosynthetic species of Orobanche. Plant Mole• Collins, Colorado. cular Biology 33:965-977. van der Vies, S. M., P. V. Viitanen, A. A. Gatenby, G. H. Wolfe, K. H., D. S. Katz-Downie, C. W. Morden, and J. D. Lorimer, and R. Jaenicke. 1992. Conformational states Palmer. 1992a. Evolution of the plastid ribosomal RNA of ribulosebisphosphate carboxylase and their interac• operon in a nongreen parasitic plant: accelerated se• tion with chaperonin 60. Biochemistry 31:3635-3644. quence evolution, altered promoter structure, and tRNA Walsh, M. A., E. A. Rechel and T. M. Popovich. 1980. Ob• pseudogenes. Plant Molecular Biology 18: 1037-1048. servations of plastid fine structure in the holoparasitic Wolfe, K. H., C. W. Morden, S. C. Ems, and 1. D. Palmer. angiosperm Epifagus virginiana. American Journal of 1992b. Rapid evolution of the plastid translational ap• Botany 67:833-837. paratus in a nonphotosynthetic plant: loss or acceler• Weber, H. C. 1980. Evolution of parasitism in Scrophu• ated sequence evolution of tRNA and ribosomal pro• lariceae and Orobanchaceae. Plant Systematics and tein genes. Journal of Molecular Evolution 35: Evolution 136:217-232. 304-317. Wettstein, R. 1897. Scrophulariceae. In Die Natiirlichen Wolfe, K. H., C. W. Morden, and J. D. Palmer. 1992c. Pflanzenfamilien Nachtrage I-IV, eds. A. Engler and Function and evolution of a minimal plastid genome K. Prantl, pp. 293-299. W Engelmann, Leipzig. from a nonphotosynthetic parasitic plant. Proceedings Wiens, D., and B. A. Barlow. 1971. The cytogeography and of the National Academy of Sciences U.S.A. 89: relationships of the viscaceous and eremolepidaceous 10648-10652. mistletoes. Taxon 20:313-332. Wu, c.-I., and W.-H. Li. 1985. Evidence for higher rates of Wilson, R. J. M., P. W. Denny, P. R. Preiser, K. Rangachari, nucleotide substitution in rodents than in man. Pro• K. Roberts, A. Roy, A. Whyte, M. Strath, D. J. Moore, ceedings of the National Academy of Sciences U.S.A. P. W. Moore, and D. H. Williamson. 1996. Complete 82:1741-1745. 574 INDEX

Xanthophytes, 383 Zebrafish, 469 Xenoiogy, 106, 277 Zingiberales, 499 Zoospores, 518-519, 520, 522 Y chromosome, 258 Zooxanthellae, 390 Yeast, 283, 460 Zygnemataceae, 531 Zygnematales, 525, 526, 527, 529-531, 533 lea, 7,27,44,51,276,283 Zygophyllaceae, 499 Zeaxanthin, 391