Bradleya 22/2004 pages 9 – 14

Resolving the phylogenetic placement of Blossfeldia liliputana (Cactaceae): reticulate evolution, chloroplast inheritance, and graft-chimeras

Root Gorelick School of Life Sciences, Arizona State University, Tempe, AZ 85287-4501, U.S.A. (email: [email protected]).

Summary: The perplexing ancestral phylogenetic und solche Phylogenien werden durch reticulate placement of Blossfeldia liliputana based on Evolution verkompliziert. Das Problem mit der chloroplast DNA can possibly be explained by (1) zweiten Hypothese liegt darin begründet, dass the hybrid origin and uniparental inheritance of die meisten kultivierten Exemplare von chloroplasts in Blossfeldia and (2) the DNA Blosssfeldia gepfropft sind, und dass sich samples originating from the grafting stock niemand sicher ist, ob sich nicht leicht upon which the Blossfeldia was cultivated. The Pfropfchimären bilden. Solange diese beiden problem with the first of these hypotheses is that Hypothesen nicht ausgetestet worden sind, wird nobody knows whether chloroplasts are inherited Blossfeldia am Besten als hochabgeleitetes Glied from one or both parents in Blossfeldia nor how der Unterfamilie betrachtet, wie es this inheritance pattern may have changed in auch von den morphologischen Daten gestützt ancestors of Blossfeldia. Phylogenetic recon- wird. structions of species trees assume that the mode of organelle inheritance is known, and these phy- Introduction logenies are confounded by reticulate evolution. With trepidation, Nyffeler (2002) proposed that The problem with the second hypothesis is that Blossfeldia liliputana Werdermann is a basal most cultivated specimens of Blossfeldia are member of the subfamily Cactoideae Schumann. grafted and nobody is certain whether or not Yet, phylogenies based exclusively on morpho- these readily form graft-chimeras. Until both logical characters place Blossfeldia liliputana as hypotheses are tested, B. liliputana should be a highly derived taxon. Evidence for this place- considered a highly derived member of the sub- ment comes from such notable characters as family Cactoideae, as indicated by morphological having fewer stomata than any other terrestrial data. (other than parasites and aquatics), stomata that are sunken in areolar pits, Zusammenfassung: Die auf Grund von extremely thin epidermis, and lack of a hypo- Chloroplasten-DNA-Untersuchungen rätselhafte dermis (Barthlott & Porembski, 1996). phylogenetische Stellung von Blossfeldia The present consensus position is that liliputana kann möglicherweise (1) durch einen Blossfeldia Werdermann is in the tribe hybridogenen Ursprung von Blossfeldia und uni- Notocacteae Buxbaum, a position largely founded parentale Chloroplastenvererbung erklärt on a superficial resemblance to Parodia werden, oder (2) weil Chloroplasten untersucht Spegazzini and Frailea Britton & Rose, with wurden, die aus der Propfunterlage der kul- somewhat similar seeds and a hairy testa (Hunt tivierten Pflanzen stammen. Das Problem mit et al., 1990). Seeds, however, provide the only der ersten dieser beiden Hypothesen ist die similarity. Blossfeldia lacks several important Tatsache, dass niemand weiss, ob die characters of Notocacteae, such as presence of Chloroplasten bei Blossfeldia nur durch eine oder acicular bristles in the receptacle areoles. durch beide Elternpflanzen vererbt werden, und Blossfeldia are much reduced in their ob dieses Vererbungsmuster bei Vorläufern von structures and seem to have more in common Blossfeldia anders gewesen sein könnte. with tribe Trichocereeae Buxbaum than with Phylogenetische Rekonstruktionen von Notocacteae. Blossfeldia flowers have a well- Artverwandtschafen gehen davon aus, dass der defined cauline zone, with a mass of white kinked Modus der Organellenvererbung bekannt ist, uniseriate trichomes at their base, which is not

Bradleya 22/2004 9 found in tribe Notocacteae, but it is present in http://mobot.mobot.org/W3T/Search/ipcn.html) nal parent, or both. Three prickly pears ( tively small , especially for those taxa with subtribe Rebutiinae of the tribe Trichocereeae, and Ramsey & Schemske (1998) show that ane- basilaris Engelmann & Bigelow, O. engelmannii more than four sets of homologous chromosomes e.g. Rebutia Schumann sensu stricta (Mottram, uploidy is rare or absent in cacti. Salm-Dyck, O. vulgaris Miller) and one hedgehog (Pinkava et al., 1998), with the notable exception pers. comm.). Blossfeldia fruits also share many Intergeneric hybrids do occur in the (Echinocereus engelmannii (Parry ex of Opuntia sensu stricta in which polyploids morphological characters with subfamily Cactoideae, albeit rarely. The tetraploid Engelmann) Rümpler) have strictly maternal appear to be more robust (Pinkava, 2002). The Rebutiinae, such as fleshy scales overtopping the Bergerocactus emoryi (Engelm.) Britton & Rose inheritance. Two epiphytic cacti, both in the evidence is therefore equivocal for whether poly- rim of the umbilicus (Mottram, pers. comm.). hybridizes with both the diploid Myrtillocactus tribe Rhipsalideae A. P. de Candolle – Hatiora ploidy results in larger, smaller, or equal sized Thus, there is morphological evidence for place- cochal (Orcutt) Britton & Rose to form the hexa- gaertneri (Regel) Barthlott and Schlumbergera cacti. ment of Blossfeldia in either of the two closely ploid xMyrtgerocactus lindsayi Moran and the truncata (Haworth) Moran) – have many chloro- Rapid genetic and epigenetic changes follow- related tribes Notocacteae and Trichocereeae, tetraploid Pachycereus pringlei (S. Watson) plasts in their sperm cells and therefore most ing polyploid formation are believed to disrupt neither of which is generally considered ancestral Britton & Rose to form the octaploid likely have chloroplasts inherited from both development (Osborn et al., 2003) and could within the Cactoideae. xPacherocactus orcuttii (K. Brandegee) parents (Corriveau & Coleman, 1988). thereby explain general size reductions. Genetic Why did Nyfeller’s (2002) analysis result in a G.D.Rowley (Moran, 1962a, b; Pinkava et al., Furthermore, these two species do not have changes include gains or losses of DNA frag- glaringly anomalous phylogenetic position for B. 1998). The diploid Hylocereus undatus (Haworth) dimorphic sperm cells, providing further evi- ments (Osborn et al., 2003). The developmental liliputana amongst the Cactoideae? His phy- Britton & Rose has been hybridized with the dence of biparental inheritance (Corriveau, pers. pathways of most organisms, including plants, logeny was based solely on trnK/matK and trnL- tetraploid Selenicereus megalanthus (Schumann comm.). With dimorphism, one sperm cell might are largely regulated by epigenetic signals. trnF sequences of chloroplast DNA. Gorelick ex Vaupel) Moran to form a viable hexaploid (Tel- contain no chloroplasts and always end up fer- Epigenetic signals are often mediated by mole- (2002) proposed that the problem is that B. Zur et al., 2003). I hypothesize that the hexaploid tilising the egg cell, while the sperm cell with cules that are attached to DNA, such as cytosine liliputana is of hybrid origin and that chloroplast B. liliputana was similarly formed via hybridiza- chloroplasts always fertilises the polar nuclei. It methylation, heterochromatin formation, and DNA is only maternally inherited. Here, I review tion of a diploid and tetraploid parent. My sus- is not known whether chloroplast inheritance is histone acetylation (Holliday & Pugh, 1975; and critique that argument, noting that until picion is that both parents were members of the uniparental or biparental in B. liliputana or any Riggs, 1975; Russo et al., 1996). These epigenetic further cytological or molecular work with B. Notocacteae or Trichocereeae, which have several other member of the tribes Notocacteae or signals act by suppressing gene products (see liliputana is done, we will not know whether that tetraploid lineages (Pinkava et al., 1998). Trichocereeae. Gorelick, 2003a, b for review of mechanisms). argument is valid. I also discuss the mechanisms Most flowering plants have chloroplasts that Why might the fluidity of modes of organelle The proportion of the genome modified by epige- and equivocal evidence for Blossfeldia morphol- are solely inherited from the female parent inheritance over evolutionary time pose a netic signals, especially of cytosine methylation ogy arising via allopolyploidy. Additionally, I (Mogensen, 1996). Hence phylogenies based on problem in the phylogenetic reconstruction of the and heterochromatin, increases dramatically fol- propose an alternative hypothesis based on uniparentally inherited chloroplast DNA would Cactaceae? One reason chloroplast DNA is used lowing formation of a new polyploid (Matzke & graft-chimeras for why analysis of chloroplast only reflect half of the ancestry – the female half in phylogenetics of cacti is that organelle DNA Matzke, 1998; Liu et al., 2000), thereby disrupt- DNA may have incorrectly ascribed the phyloge- – of any plant of allopolyploid origin. Gorelick evolves much quicker than nuclear DNA because ing regulatory pathways. Such disruptions could netic placement of B. liliputana: The chloroplast (2002) argued that B. liliputana could then have organelle genomes are haploid. This means that mean that only neotenous (paedomorphic; repro- DNA sampled may have been from the stock a highly derived paternal parent and an ances- a maternally inherited haploid genome will have ducing in a juvenile state) individuals could rather than from the Blossfeldia scion. tral maternal parent and would thereby be clas- one-fourth the effective population size of a survive and successfully reproduce (see Ahokas sified as ancestral if solely using chloroplast biparentally inherited diploid genome (Moore, & Manninen, 2000). Spontaneous changes in Polyploidy and inheritance of chloroplasts DNA data. Furthermore, hybridization (reticu- 1995). The effective population size of the haploid these epigenetic signals can cause reversion in all B. liliputana is hexaploid, which is virtually a late evolution) results in the transfer of chloro- genome doubles when the organelle is or part of a plant from adult to juvenile growth guarantee of allopolyploidy. The only parsimo- plast DNA amongst lineages, which confounds biparentally inherited. The coalescent theory forms, or vice versa (Brink, 1962). In fact, the nious way to produce hexaploids is via hybridis- phylogenetic reconstruction of species trees underlying phylogenetic reconstructions there- additional cytosine methylation and chromatin ation between a diploid and a tetraploid parent. (Moore, 1995). fore crucially relies on knowing whether inheri- formation caused by polyploidization may itself Hexaploidy may have resulted from the hybridi- In most flowering plants, chloroplast DNA is tance is uniparental or biparental (Hudson, be the cause of neoteny (cf. Golubovsky, 2002) or sation of a diploid and a tetraploid parent in exclusively maternally inherited because plastids 1990). Yet, all phylogenetic studies of cacti have other changes in development and its timing which both parents had unreduced gametes (i.e. are never incorporated into their sperm cells. In simply assumed uniparental inheritance. (Dremen, 1936; Ficq & Pavan, 1957; Delevoryas, containing the same amounts of DNA as the virtually all flowering plants, four microspore 1980; Levin, 1983 citing Gottschalk, 1976). parents’ somatic cells). This can arise by the eggs cells are produced by meiosis. Each of these Is Blossfeldia morphology consistent with Neoteny is known to confound phylogenetic and sperm cells not having undergone both haploid microspores then undergoes a mitotic allopolyploidy? inferences in cacti (Barthlott & Hunt, 1993). meiotic divisions (Bretagnolle, 2001) or by them division to produce a generative and a vegetative Polyploidy, especially in recently created poly- Polyploidy may also disrupt the proper for- both undergoing two meiotic reduction divisions cell. During this mitotic division, plastids, includ- ploids, is known to result in novel morphological mation of wood fibres because of the increased followed by a chromosomal doubling (Friedman, ing chloroplasts, are excluded from the genera- and developmental characteristics. volume needed to hold the extra chromosomal 1999). Alternatively, both parents could have tive cell in most flowering plants (Hagemann & Most studies that have examined a broad material within a cell (Stebbins, 1938; Otto & undergone normal meiotic divisions and a Schröder, 1989; Mogensen, 1996). The generative swath of angiosperm taxa show that polyploids Whitton, 2000). Polyploidy causes nuclei and triploid offspring was produced, which later pro- cell later undergoes another mitotic division to typically are larger plants with larger cells and cells to be larger, which is believed to disrupt duced hexaploid descendants via a subsequent form two sperm cells. However, in a minority of organs than their diploid ancestors, albeit with cambial activity (Darlington, 1932). When wood autopolyploidization. Both alternatives should be species, plastids are parcelled a reduced number of parts, e.g. fewer fibres are not produced or are deformed, stems considered reasonable possibilities for the evolu- into the generative and sperms cells (Corriveau parts (Ramsey & Schemske, 2002). This result, will be short. Polyploid-induced disruption of tion of Blossfeldia or any hexaploid flowering & Coleman, 1988), hence are deposited into the however, is probably biased by the corpus of work developmental pathways may also have been the plants (Ramsey & Schemske, 1998). Other than zygote, and therefore paternally inherited plas- on domesticated crops in which artificial selec- cause of stomatal loss in Blossfeldia. allopolyploidy, the only other possible origin for tids can be retained in seedlings (also note that tion has been exclusively for larger individuals, I sincerely believe that allopolyploidy is the hexaploidy in Blossfeldia is aneuploidy it is possible for other mechanisms to selectively especially for larger fruits (Hilu, 1993). By con- cause for the derived morphology of Blossfeldia. followed by spontaneous chromosome doubling eliminate organelles of one parent or the other in trast, polyploid cacti seem to be of roughly the However, because I have not conclusively demon- (autopolyploidy). However, this is highly the embryo, as occurs with human male mito- same size as their putative diploid ancestors, at strated this, I propose another hypothesis – that unlikely. Perusal of the Index to Plant chondria). It appears that only six species of cacti least judging by the list of known polyploid taxa. is due to Roy Mottram (pers. comm.) – that Chromosome Numbers (Goldblatt & Johnson) have been tested for whether chloroplasts are Furthermore, virtually all cactus genera for explains the phylogenetic relationship inferred (updated and published annually online at inherited solely from the maternal parent, pater- which polyploidy has been documented are rela- by Nyfeller (2002).

10 Bradleya 22/2004 Bradleya 22/2004 11 found in tribe Notocacteae, but it is present in http://mobot.mobot.org/W3T/Search/ipcn.html) nal parent, or both. Three prickly pears (Opuntia tively small plants, especially for those taxa with subtribe Rebutiinae of the tribe Trichocereeae, and Ramsey & Schemske (1998) show that ane- basilaris Engelmann & Bigelow, O. engelmannii more than four sets of homologous chromosomes e.g. Rebutia Schumann sensu stricta (Mottram, uploidy is rare or absent in cacti. Salm-Dyck, O. vulgaris Miller) and one hedgehog (Pinkava et al., 1998), with the notable exception pers. comm.). Blossfeldia fruits also share many Intergeneric hybrids do occur in the cactus (Echinocereus engelmannii (Parry ex of Opuntia sensu stricta in which polyploids morphological characters with subfamily Cactoideae, albeit rarely. The tetraploid Engelmann) Rümpler) have strictly maternal appear to be more robust (Pinkava, 2002). The Rebutiinae, such as fleshy scales overtopping the Bergerocactus emoryi (Engelm.) Britton & Rose inheritance. Two epiphytic cacti, both in the evidence is therefore equivocal for whether poly- rim of the umbilicus (Mottram, pers. comm.). hybridizes with both the diploid Myrtillocactus tribe Rhipsalideae A. P. de Candolle – Hatiora ploidy results in larger, smaller, or equal sized Thus, there is morphological evidence for place- cochal (Orcutt) Britton & Rose to form the hexa- gaertneri (Regel) Barthlott and Schlumbergera cacti. ment of Blossfeldia in either of the two closely ploid xMyrtgerocactus lindsayi Moran and the truncata (Haworth) Moran) – have many chloro- Rapid genetic and epigenetic changes follow- related tribes Notocacteae and Trichocereeae, tetraploid Pachycereus pringlei (S. Watson) plasts in their sperm cells and therefore most ing polyploid formation are believed to disrupt neither of which is generally considered ancestral Britton & Rose to form the octaploid likely have chloroplasts inherited from both development (Osborn et al., 2003) and could within the Cactoideae. xPacherocactus orcuttii (K. Brandegee) parents (Corriveau & Coleman, 1988). thereby explain general size reductions. Genetic Why did Nyfeller’s (2002) analysis result in a G.D.Rowley (Moran, 1962a, b; Pinkava et al., Furthermore, these two species do not have changes include gains or losses of DNA frag- glaringly anomalous phylogenetic position for B. 1998). The diploid Hylocereus undatus (Haworth) dimorphic sperm cells, providing further evi- ments (Osborn et al., 2003). The developmental liliputana amongst the Cactoideae? His phy- Britton & Rose has been hybridized with the dence of biparental inheritance (Corriveau, pers. pathways of most organisms, including plants, logeny was based solely on trnK/matK and trnL- tetraploid Selenicereus megalanthus (Schumann comm.). With dimorphism, one sperm cell might are largely regulated by epigenetic signals. trnF sequences of chloroplast DNA. Gorelick ex Vaupel) Moran to form a viable hexaploid (Tel- contain no chloroplasts and always end up fer- Epigenetic signals are often mediated by mole- (2002) proposed that the problem is that B. Zur et al., 2003). I hypothesize that the hexaploid tilising the egg cell, while the sperm cell with cules that are attached to DNA, such as cytosine liliputana is of hybrid origin and that chloroplast B. liliputana was similarly formed via hybridiza- chloroplasts always fertilises the polar nuclei. It methylation, heterochromatin formation, and DNA is only maternally inherited. Here, I review tion of a diploid and tetraploid parent. My sus- is not known whether chloroplast inheritance is histone acetylation (Holliday & Pugh, 1975; and critique that argument, noting that until picion is that both parents were members of the uniparental or biparental in B. liliputana or any Riggs, 1975; Russo et al., 1996). These epigenetic further cytological or molecular work with B. Notocacteae or Trichocereeae, which have several other member of the tribes Notocacteae or signals act by suppressing gene products (see liliputana is done, we will not know whether that tetraploid lineages (Pinkava et al., 1998). Trichocereeae. Gorelick, 2003a, b for review of mechanisms). argument is valid. I also discuss the mechanisms Most flowering plants have chloroplasts that Why might the fluidity of modes of organelle The proportion of the genome modified by epige- and equivocal evidence for Blossfeldia morphol- are solely inherited from the female parent inheritance over evolutionary time pose a netic signals, especially of cytosine methylation ogy arising via allopolyploidy. Additionally, I (Mogensen, 1996). Hence phylogenies based on problem in the phylogenetic reconstruction of the and heterochromatin, increases dramatically fol- propose an alternative hypothesis based on uniparentally inherited chloroplast DNA would Cactaceae? One reason chloroplast DNA is used lowing formation of a new polyploid (Matzke & graft-chimeras for why analysis of chloroplast only reflect half of the ancestry – the female half in phylogenetics of cacti is that organelle DNA Matzke, 1998; Liu et al., 2000), thereby disrupt- DNA may have incorrectly ascribed the phyloge- – of any plant of allopolyploid origin. Gorelick evolves much quicker than nuclear DNA because ing regulatory pathways. Such disruptions could netic placement of B. liliputana: The chloroplast (2002) argued that B. liliputana could then have organelle genomes are haploid. This means that mean that only neotenous (paedomorphic; repro- DNA sampled may have been from the stock a highly derived paternal parent and an ances- a maternally inherited haploid genome will have ducing in a juvenile state) individuals could rather than from the Blossfeldia scion. tral maternal parent and would thereby be clas- one-fourth the effective population size of a survive and successfully reproduce (see Ahokas sified as ancestral if solely using chloroplast biparentally inherited diploid genome (Moore, & Manninen, 2000). Spontaneous changes in Polyploidy and inheritance of chloroplasts DNA data. Furthermore, hybridization (reticu- 1995). The effective population size of the haploid these epigenetic signals can cause reversion in all B. liliputana is hexaploid, which is virtually a late evolution) results in the transfer of chloro- genome doubles when the organelle is or part of a plant from adult to juvenile growth guarantee of allopolyploidy. The only parsimo- plast DNA amongst lineages, which confounds biparentally inherited. The coalescent theory forms, or vice versa (Brink, 1962). In fact, the nious way to produce hexaploids is via hybridis- phylogenetic reconstruction of species trees underlying phylogenetic reconstructions there- additional cytosine methylation and chromatin ation between a diploid and a tetraploid parent. (Moore, 1995). fore crucially relies on knowing whether inheri- formation caused by polyploidization may itself Hexaploidy may have resulted from the hybridi- In most flowering plants, chloroplast DNA is tance is uniparental or biparental (Hudson, be the cause of neoteny (cf. Golubovsky, 2002) or sation of a diploid and a tetraploid parent in exclusively maternally inherited because plastids 1990). Yet, all phylogenetic studies of cacti have other changes in development and its timing which both parents had unreduced gametes (i.e. are never incorporated into their sperm cells. In simply assumed uniparental inheritance. (Dremen, 1936; Ficq & Pavan, 1957; Delevoryas, containing the same amounts of DNA as the virtually all flowering plants, four microspore 1980; Levin, 1983 citing Gottschalk, 1976). parents’ somatic cells). This can arise by the eggs cells are produced by meiosis. Each of these Is Blossfeldia morphology consistent with Neoteny is known to confound phylogenetic and sperm cells not having undergone both haploid microspores then undergoes a mitotic allopolyploidy? inferences in cacti (Barthlott & Hunt, 1993). meiotic divisions (Bretagnolle, 2001) or by them division to produce a generative and a vegetative Polyploidy, especially in recently created poly- Polyploidy may also disrupt the proper for- both undergoing two meiotic reduction divisions cell. During this mitotic division, plastids, includ- ploids, is known to result in novel morphological mation of wood fibres because of the increased followed by a chromosomal doubling (Friedman, ing chloroplasts, are excluded from the genera- and developmental characteristics. volume needed to hold the extra chromosomal 1999). Alternatively, both parents could have tive cell in most flowering plants (Hagemann & Most studies that have examined a broad material within a cell (Stebbins, 1938; Otto & undergone normal meiotic divisions and a Schröder, 1989; Mogensen, 1996). The generative swath of angiosperm taxa show that polyploids Whitton, 2000). Polyploidy causes nuclei and triploid offspring was produced, which later pro- cell later undergoes another mitotic division to typically are larger plants with larger cells and cells to be larger, which is believed to disrupt duced hexaploid descendants via a subsequent form two sperm cells. However, in a minority of organs than their diploid ancestors, albeit with cambial activity (Darlington, 1932). When wood autopolyploidization. Both alternatives should be flowering plant species, plastids are parcelled a reduced number of parts, e.g. fewer flower fibres are not produced or are deformed, stems considered reasonable possibilities for the evolu- into the generative and sperms cells (Corriveau parts (Ramsey & Schemske, 2002). This result, will be short. Polyploid-induced disruption of tion of Blossfeldia or any hexaploid flowering & Coleman, 1988), hence are deposited into the however, is probably biased by the corpus of work developmental pathways may also have been the plants (Ramsey & Schemske, 1998). Other than zygote, and therefore paternally inherited plas- on domesticated crops in which artificial selec- cause of stomatal loss in Blossfeldia. allopolyploidy, the only other possible origin for tids can be retained in seedlings (also note that tion has been exclusively for larger individuals, I sincerely believe that allopolyploidy is the hexaploidy in Blossfeldia is aneuploidy it is possible for other mechanisms to selectively especially for larger fruits (Hilu, 1993). By con- cause for the derived morphology of Blossfeldia. followed by spontaneous chromosome doubling eliminate organelles of one parent or the other in trast, polyploid cacti seem to be of roughly the However, because I have not conclusively demon- (autopolyploidy). However, this is highly the embryo, as occurs with human male mito- same size as their putative diploid ancestors, at strated this, I propose another hypothesis – that unlikely. Perusal of the Index to Plant chondria). It appears that only six species of cacti least judging by the list of known polyploid taxa. is due to Roy Mottram (pers. comm.) – that Chromosome Numbers (Goldblatt & Johnson) have been tested for whether chloroplasts are Furthermore, virtually all cactus genera for explains the phylogenetic relationship inferred (updated and published annually online at inherited solely from the maternal parent, pater- which polyploidy has been documented are rela- by Nyfeller (2002).

10 Bradleya 22/2004 Bradleya 22/2004 11 Graft-chimeras (=Gymnocalycium ‘Hibotan’ + Hylocereus), i.e. Tephrocactus Lemaire sensu lato (including APPLEQUIST, W.L. & WALLACE, R.S. (2001). Most specimens of Blossfeldia that are outside of three plants grafted together with the variegated Maihueniopsis Spegazzini, Puna R. Kiesling, Phylogeny of the portulacaceous cohort based their native habitat are grafted. When any pair Gymnocalycium sandwiched as an ‘inter-stem’ and some species of Austrocylindropuntia on ndhF sequence data. Syst. Bot. 26: 406–419. of plants are grafted, there exists the possibility between the Hylocereus stock and Blossfeldia Backeberg) have from two to fifteen times the AXSMITH, B.J., TAYLOR, E.L. & TAYLOR, T.N. that the scion may be a chimera. A chimera scion. The distinctive colours of the variegated typical diploid complement of chromosomes (1998). The limitations of molecular system- means that a shoot contains a mixture of cells Gymnocalycium tissues would then be obvious in (Pinkava et al., 1998). It will be useful to ascer- atics: a palaeobotanical perspective. Taxon 47: from both stock and scion. These chimeras are the Blossfeldia scion if a periclinal chimera forms. tain the ploidy levels of other such geophytic taxa 105–108. often classified as periclinal, in which one layer such as Cintia knizei Riha and Rimacactus laui BARTHLOTT, W. & HUNT, D.R. (1993). Cactaceae. of cells is from one plant and the remaining Conclusion (Lüthy) Mottram. If high ploidy levels are found Pages 161–196 in K. Kubitzki, Rowher, J.G. & layers from another plant (e.g. Echinopsis ‘Haku- I have presented evidence that B. liliputana is of in most or all diminutive geophytic South Bittrich, V. (eds.) The families and genera of Jo’), or sectorial, in which there is a transverse likely allopolyploid origin. Whether the morpho- American cacti, then this provides corroborative vascular plants. Volume II. Flowering plants, division between the cells of different origin logical characters that set Blossfeldia apart from support for the allopolyploid hypothesis for the dicotyledons: Magnoliid, Hamamelid and (Tilney-Bassett, 1986). Such graft-chimeras other members of the subfamily Cactoideae could derived phylogenetic placement of Blossfeldia. Caryophilid families. Springer-Verlag, Berlin. have been formed between species in many fam- be novel and a result of allopolyploidy is still an The alternative, which I have not yet dis- BARTHLOTT, W. & POREMBSKI, S. (1996). Ecology ilies and can be formed naturally or artificially open question, but I have outlined why this is cussed, is that B. liliputana truly is a basal/ances- and morphology of Blossfeldia liliputana (Tilney-Bassett, 1986). Natural occurrences of plausible. tral member of the subfamily Cactoideae, as indi- (Cactaceae): a poikilohydric and almost asto- chimeras include haustoria of parasitic plants I have presented two hypotheses for why cated by molecular data. Patrick Griffith (2004) mate succulent. Bot. Acta 109: 161–166. infiltrating their host’s cells (Mauseth et al., chloroplast DNA would erroneously implicate proposes that diminutive geophytic South BRETAGNOLLE, F. (2001). Pollen production and 1992), which occurs in some cacti (Mauseth et al., Blossfeldia as an ancestral member of the sub- American cacti, such as Blossfeldia, most closely spontaneous polyploidization in diploid popu- 1984, 1985). Artificial graft-chimeras of cacti family Cactoideae: (1) maternally inherited resemble the ancestors of all cacti. His hypothe- lations of Anthoxanthum alpinum. Biol. J. have been reported between cacti in a few tribes chloroplasts in conjunction with allopolyploidy sis seems plausible in light of recent molecular Linn. Soc. 72: 241–247. of the subfamily Cactoideae (Heath, 1992; and (2) contamination of samples via cells from phylogenies showing that the nearest outgroups BRINK, R.A. (1962). Phase change in higher Rowley, 2003), including the tribe Trichocereeae, a grafting stock. Ironically, these hypotheses are to cacti are diminutive members of the family plants and somatic cell heredity. Quart. Rev. but not Notocacteae. Heath (1992) lists the fol- related. Much information about plastid inheri- Portulacaceae, many of which are geophytes Biol. 37: 1–22. lowing such graft chimeras in cacti: tance has been garnered from chimeric plants in (Hershkovitz & Zimmer, 1997; Applequist & CORRIVEAU, J.L. & COLEMAN, A.W. (1988). Rapid kotschoubeyanus + Trichocereus (Echinopsis) which a layer of tissue has incurred a somatic Wallace, 2001). But, these phylogenies were all screening method to detect potential spachianus = +Ariocereus `Ewaldii’; Echinopsis mutation coding for chlorophyll deficiencies based on the assumptions of maternally inherited biparental inheritance of plastid DNA and eyriesii + Gymnocalycium mihanovichii var. making the entire plant variegated (Kirk & chloroplast genomes and no reticulate evolution. results for over 200 angiosperm species. Amer. friedrichii = +Echinocalycium `Japonicum’; Tilney-Bassett, 1978). Both hypotheses need to be It will require more data and more robust J. Bot. 75: 1443–1458. Gymnocalycium + Hylocereus (no specific epi- tested before definitive inferences regarding the phylogenetic methods before chloroplast DNA DARLINGTON, C.D. (1932). Recent advances in thets listed) = +Hylocalycium. If grafted speci- phylogenetic placement of B. liliputana are data alone will allow us to discern whether B. cytology, 1st edition. Churchill, London. mens of Blossfeldia are used for molecular phy- made. The mode of inheritance of chloroplasts – liliputana is an ancestral or a derived member of DELEVORYAS, T. (1980). Polyploidy in logenetic analysis, care must be taken to insure whether they be maternal, paternal, or the Cactoideae. Furthermore, there needs to be gymnosperms. Pages 215–218 in W. H. Lewis that the genomes being analysed are of biparental – should also be tested using cytolog- more than just chloroplast DNA evidence to (ed.) Polyploidy: biological significance. Blossfeldia and not of the grafting stock. Even if ical (Corriveau & Coleman, 1988) or molecular justify assigning Blossfeldia as an ancestral Plenum Press, New York. clones of Blossfeldia were initially grafted and (Milligan, 1992) techniques. The allopolyploidy member of the Cactoideae, especially in light of DREMEN, H. (1936). Aposporic parthenogenesis in then removed from the stock (typically, and graft-chimera hypotheses are not mutually the preponderance of morphological evidence a triploid apple, Malus hupehensis. J. Arnold. Pereskiopsis), they may still retain vestiges of the exclusive. Both could independently contribute to indicating that it is highly derived (Axsmith et Arboret. 17: 90–105. graft-chimera, as appears to be the case with erroneous phylogenetic placement of B. liliputana al., 1998). Until the above two hypotheses are FICQ, A. & PAVAN, C. (1957). Autoradiography of Echinopsis ‘Haku-Jo’. Three of the four samples as an ancestral taxon. tested, B. liliputana should probably be consid- polytene chromosomes of Rhynchosciara of Blossfeldia that Nyfeller (2002) used for his The graft-chimera hypothesis can be tested by ered a highly derived member of the subfamily angelae at different stages of larval develop- chloroplast DNA analysis were grafted. grafting Blossfeldia onto a variegated plant, Cactoideae. ment. Nature 180: 983–984. If Blossfeldia specimens are periclinal which will (by necessity) itself be a chimera. More FRIEDMAN, W.E. (1999). Expression of the cell chimeras with the stock’s cells on the outer sophisticated genetic techniques (from allozymes Acknowledgments cycle in sperm of Arabidopsis: implications for layers, then these should be readily identifiable. to microarrays) could also be used to detect oth- Thanks to Annette Coleman, Jim Mauseth and understanding patterns of gametogenesis and If the specimen has a hypodermis, a thick cuticle erwise cryptic graft-chimeral tissues. Patrick Griffith for patiently answering my fertilization in plants and other eukaryotes. or layer of epidermal wax, or a relatively large The allopolyploid origin hypothesis needs to queries while preparing this manuscript. Thanks Development 126: 1065–1075. number of stomata (especially if they are not be tested via construction of phylogenies that are to Joe Corriveau for reviewing the manuscript GIBSON, A. C. (1973). Comparative anatomy of sunken in stomatal pits), then this is diagnostic based on both organelle and nuclear DNA, using and for unearthing and sharing his two-decade- secondary xylem in Cactoideae (Cactaceae). of the stock, and not the Blossfeldia (Barthlott & phylogenetic methods that account for reticulate old unpublished data. Special thanks to Roy Biotropica 5: 29–65. Porembski, 1996). evolution and inferred modes of organelle inher- Mottram for his suggestion of graft-chimeras, his GOLDBLATT, P. & JOHNSON, D.E., (EDS.) (1997). Roy Mottram (pers. comm.) has not found any itance. This will be a highly non-trivial task. It willingness to dissect precious B. liliputana spec- Index to plant chromosome numbers. Missouri evidence for periclinal chimeras in grafted spec- will be essential to determine whether chloro- imens and his generous in-depth comments, Botanical Garden, St. Louis. imens of Blossfeldia. All five specimens that he plasts are maternally, paternally, or biparentally regardless of any differences we may have GOLUBOVSKY, M. (2002). Paternal familial examined had a thin epidermis with no visible inherited in Blossfeldia and any ancestral taxa regarding cactus phylogenies. twinning: hypothesis and genetic/medical stomata. This does not rule out the existence of that are believed to have hybridized to form the implications. Twin Res. 5: 75–86. periclinal chimeras, but only eliminates the pos- ancestor of Blossfeldia, using fluorescent staining References GORELICK, R. (2002). DNA sequences and cactus sibility of those in which the stock contributes of chloroplasts and other plastids in pollen grains AHOKAS, H. & MANNINEN, M.L. (2000). classification: a short review. Bradleya 20: cells to the epidermal layer of the scion. One pos- and sperm cells of Blossfeldia specimens that Introgressive hexaploid oats from the Avena 1–4. sible way to detect graft-chimeral tissues in the have never been grafted. abyssinica (female) x A. sativa hybrid: perfor GORELICK, R. (2003a). Evolution of dioecy and sex interior of scions is to graft Blossfeldia on top of High ploidy levels have also been found in mance, grain lipids and proteins. Euphytica chromosomes via methylation driving Muller’s a variegated chimera such as +Hylocalycium other diminutive geophytic South American cacti. 111: 153–160. ratchet. Biol. J. Linn. Soc. 80: 353–368.

12 Bradleya 22/2004 Bradleya 22/2004 13 Graft-chimeras (=Gymnocalycium ‘Hibotan’ + Hylocereus), i.e. Tephrocactus Lemaire sensu lato (including APPLEQUIST, W.L. & WALLACE, R.S. (2001). Most specimens of Blossfeldia that are outside of three plants grafted together with the variegated Maihueniopsis Spegazzini, Puna R. Kiesling, Phylogeny of the portulacaceous cohort based their native habitat are grafted. When any pair Gymnocalycium sandwiched as an ‘inter-stem’ and some species of Austrocylindropuntia on ndhF sequence data. Syst. Bot. 26: 406–419. of plants are grafted, there exists the possibility between the Hylocereus stock and Blossfeldia Backeberg) have from two to fifteen times the AXSMITH, B.J., TAYLOR, E.L. & TAYLOR, T.N. that the scion may be a chimera. A chimera scion. The distinctive colours of the variegated typical diploid complement of chromosomes (1998). The limitations of molecular system- means that a shoot contains a mixture of cells Gymnocalycium tissues would then be obvious in (Pinkava et al., 1998). It will be useful to ascer- atics: a palaeobotanical perspective. Taxon 47: from both stock and scion. These chimeras are the Blossfeldia scion if a periclinal chimera forms. tain the ploidy levels of other such geophytic taxa 105–108. often classified as periclinal, in which one layer such as Cintia knizei Riha and Rimacactus laui BARTHLOTT, W. & HUNT, D.R. (1993). Cactaceae. of cells is from one plant and the remaining Conclusion (Lüthy) Mottram. If high ploidy levels are found Pages 161–196 in K. Kubitzki, Rowher, J.G. & layers from another plant (e.g. Echinopsis ‘Haku- I have presented evidence that B. liliputana is of in most or all diminutive geophytic South Bittrich, V. (eds.) The families and genera of Jo’), or sectorial, in which there is a transverse likely allopolyploid origin. Whether the morpho- American cacti, then this provides corroborative vascular plants. Volume II. Flowering plants, division between the cells of different origin logical characters that set Blossfeldia apart from support for the allopolyploid hypothesis for the dicotyledons: Magnoliid, Hamamelid and (Tilney-Bassett, 1986). Such graft-chimeras other members of the subfamily Cactoideae could derived phylogenetic placement of Blossfeldia. Caryophilid families. Springer-Verlag, Berlin. have been formed between species in many fam- be novel and a result of allopolyploidy is still an The alternative, which I have not yet dis- BARTHLOTT, W. & POREMBSKI, S. (1996). Ecology ilies and can be formed naturally or artificially open question, but I have outlined why this is cussed, is that B. liliputana truly is a basal/ances- and morphology of Blossfeldia liliputana (Tilney-Bassett, 1986). Natural occurrences of plausible. tral member of the subfamily Cactoideae, as indi- (Cactaceae): a poikilohydric and almost asto- chimeras include haustoria of parasitic plants I have presented two hypotheses for why cated by molecular data. Patrick Griffith (2004) mate succulent. Bot. Acta 109: 161–166. infiltrating their host’s cells (Mauseth et al., chloroplast DNA would erroneously implicate proposes that diminutive geophytic South BRETAGNOLLE, F. (2001). Pollen production and 1992), which occurs in some cacti (Mauseth et al., Blossfeldia as an ancestral member of the sub- American cacti, such as Blossfeldia, most closely spontaneous polyploidization in diploid popu- 1984, 1985). Artificial graft-chimeras of cacti family Cactoideae: (1) maternally inherited resemble the ancestors of all cacti. His hypothe- lations of Anthoxanthum alpinum. Biol. J. have been reported between cacti in a few tribes chloroplasts in conjunction with allopolyploidy sis seems plausible in light of recent molecular Linn. Soc. 72: 241–247. of the subfamily Cactoideae (Heath, 1992; and (2) contamination of samples via cells from phylogenies showing that the nearest outgroups BRINK, R.A. (1962). Phase change in higher Rowley, 2003), including the tribe Trichocereeae, a grafting stock. Ironically, these hypotheses are to cacti are diminutive members of the family plants and somatic cell heredity. Quart. Rev. but not Notocacteae. Heath (1992) lists the fol- related. Much information about plastid inheri- Portulacaceae, many of which are geophytes Biol. 37: 1–22. lowing such graft chimeras in cacti: Ariocarpus tance has been garnered from chimeric plants in (Hershkovitz & Zimmer, 1997; Applequist & CORRIVEAU, J.L. & COLEMAN, A.W. (1988). Rapid kotschoubeyanus + Trichocereus (Echinopsis) which a layer of tissue has incurred a somatic Wallace, 2001). But, these phylogenies were all screening method to detect potential spachianus = +Ariocereus `Ewaldii’; Echinopsis mutation coding for chlorophyll deficiencies based on the assumptions of maternally inherited biparental inheritance of plastid DNA and eyriesii + Gymnocalycium mihanovichii var. making the entire plant variegated (Kirk & chloroplast genomes and no reticulate evolution. results for over 200 angiosperm species. Amer. friedrichii = +Echinocalycium `Japonicum’; Tilney-Bassett, 1978). Both hypotheses need to be It will require more data and more robust J. Bot. 75: 1443–1458. Gymnocalycium + Hylocereus (no specific epi- tested before definitive inferences regarding the phylogenetic methods before chloroplast DNA DARLINGTON, C.D. (1932). Recent advances in thets listed) = +Hylocalycium. If grafted speci- phylogenetic placement of B. liliputana are data alone will allow us to discern whether B. cytology, 1st edition. Churchill, London. mens of Blossfeldia are used for molecular phy- made. The mode of inheritance of chloroplasts – liliputana is an ancestral or a derived member of DELEVORYAS, T. (1980). Polyploidy in logenetic analysis, care must be taken to insure whether they be maternal, paternal, or the Cactoideae. Furthermore, there needs to be gymnosperms. Pages 215–218 in W. H. Lewis that the genomes being analysed are of biparental – should also be tested using cytolog- more than just chloroplast DNA evidence to (ed.) Polyploidy: biological significance. Blossfeldia and not of the grafting stock. Even if ical (Corriveau & Coleman, 1988) or molecular justify assigning Blossfeldia as an ancestral Plenum Press, New York. clones of Blossfeldia were initially grafted and (Milligan, 1992) techniques. The allopolyploidy member of the Cactoideae, especially in light of DREMEN, H. (1936). Aposporic parthenogenesis in then removed from the stock (typically, and graft-chimera hypotheses are not mutually the preponderance of morphological evidence a triploid apple, Malus hupehensis. J. Arnold. Pereskiopsis), they may still retain vestiges of the exclusive. Both could independently contribute to indicating that it is highly derived (Axsmith et Arboret. 17: 90–105. graft-chimera, as appears to be the case with erroneous phylogenetic placement of B. liliputana al., 1998). Until the above two hypotheses are FICQ, A. & PAVAN, C. (1957). Autoradiography of Echinopsis ‘Haku-Jo’. Three of the four samples as an ancestral taxon. tested, B. liliputana should probably be consid- polytene chromosomes of Rhynchosciara of Blossfeldia that Nyfeller (2002) used for his The graft-chimera hypothesis can be tested by ered a highly derived member of the subfamily angelae at different stages of larval develop- chloroplast DNA analysis were grafted. grafting Blossfeldia onto a variegated plant, Cactoideae. ment. Nature 180: 983–984. If Blossfeldia specimens are periclinal which will (by necessity) itself be a chimera. More FRIEDMAN, W.E. (1999). Expression of the cell chimeras with the stock’s cells on the outer sophisticated genetic techniques (from allozymes Acknowledgments cycle in sperm of Arabidopsis: implications for layers, then these should be readily identifiable. to microarrays) could also be used to detect oth- Thanks to Annette Coleman, Jim Mauseth and understanding patterns of gametogenesis and If the specimen has a hypodermis, a thick cuticle erwise cryptic graft-chimeral tissues. Patrick Griffith for patiently answering my fertilization in plants and other eukaryotes. or layer of epidermal wax, or a relatively large The allopolyploid origin hypothesis needs to queries while preparing this manuscript. Thanks Development 126: 1065–1075. number of stomata (especially if they are not be tested via construction of phylogenies that are to Joe Corriveau for reviewing the manuscript GIBSON, A. C. (1973). Comparative anatomy of sunken in stomatal pits), then this is diagnostic based on both organelle and nuclear DNA, using and for unearthing and sharing his two-decade- secondary xylem in Cactoideae (Cactaceae). of the stock, and not the Blossfeldia (Barthlott & phylogenetic methods that account for reticulate old unpublished data. Special thanks to Roy Biotropica 5: 29–65. Porembski, 1996). evolution and inferred modes of organelle inher- Mottram for his suggestion of graft-chimeras, his GOLDBLATT, P. & JOHNSON, D.E., (EDS.) (1997). Roy Mottram (pers. comm.) has not found any itance. This will be a highly non-trivial task. It willingness to dissect precious B. liliputana spec- Index to plant chromosome numbers. Missouri evidence for periclinal chimeras in grafted spec- will be essential to determine whether chloro- imens and his generous in-depth comments, Botanical Garden, St. Louis. imens of Blossfeldia. All five specimens that he plasts are maternally, paternally, or biparentally regardless of any differences we may have GOLUBOVSKY, M. (2002). Paternal familial examined had a thin epidermis with no visible inherited in Blossfeldia and any ancestral taxa regarding cactus phylogenies. twinning: hypothesis and genetic/medical stomata. This does not rule out the existence of that are believed to have hybridized to form the implications. Twin Res. 5: 75–86. periclinal chimeras, but only eliminates the pos- ancestor of Blossfeldia, using fluorescent staining References GORELICK, R. (2002). DNA sequences and cactus sibility of those in which the stock contributes of chloroplasts and other plastids in pollen grains AHOKAS, H. & MANNINEN, M.L. (2000). classification: a short review. Bradleya 20: cells to the epidermal layer of the scion. One pos- and sperm cells of Blossfeldia specimens that Introgressive hexaploid oats from the Avena 1–4. sible way to detect graft-chimeral tissues in the have never been grafted. abyssinica (female) x A. sativa hybrid: perfor GORELICK, R. (2003a). Evolution of dioecy and sex interior of scions is to graft Blossfeldia on top of High ploidy levels have also been found in mance, grain lipids and proteins. Euphytica chromosomes via methylation driving Muller’s a variegated chimera such as +Hylocalycium other diminutive geophytic South American cacti. 111: 153–160. ratchet. Biol. J. Linn. Soc. 80: 353–368.

12 Bradleya 22/2004 Bradleya 22/2004 13 GORELICK, R. (2003b). Transposable elements MILLIGAN, B.G. (1992). Is organelle DNA strictly suppress recombination in all meiotic eukary- maternally inherited? Power analysis of a otes, including automictic ancient asexuals: a binomial distribution. Amer. J. Bot. 79: reply to Schön and Martens. J. Nat. Hist. 37: 1325–1328. 903–909. MOGENSEN, H.L. (1996). The hows and whys of GOTTSCHALK, W. (1976). Die bedeutung der cytoplasmic inheritance in seed plants. Amer. polyploidie für die evolution der pflanzen. J. Bot. 83: 383–404. Gustav Fisher Verlag, Stuttgart. MOORE, W.S. (1995). Inferring phylogenies from GRIFFITH, M. P. (2004). What did the first cactus mtDNA variation: mitochondrial-gene trees look like? An attempt to reconcile the mor- versus nuclear-gene trees. Evolution 49: phological and molecular evidence. Taxon (in 718–726. press). MORAN, R. (1962a). Pachycereus orcuttii: a puzzle HAGEMANN, R. & SCHRÖDER, M.B. (1989). The solved. Cact. Succ. J. (US) 34: 88–94. cytological basis of the plastid inheritance in MORAN, R. (1962b). The unique Cereus. Cact. angiosperms. Protoplasma 152: 57–64. Succ. J. (US) 34: 184–188. HEATH, P.V. (1992). The nothogenera of the NYFFELER, R. (2002). Phylogenetic relationships Cactaceae. Calyx 1: 93–140. in the cactus family (Cactaceae) based on evi- HERSHKOVITZ, M.A. & ZIMMER, E.A. (1997). On dence from trnK/matK and trnL-trnF the evolutionary origins of the cacti. Taxon 46: sequences. Amer. J. Bot. 89: 312–326. 217–232. OSBORN, T.C., PIRES, J.C., BIRCHLER, J.A., AUGER, HILU, K.W. (1993). Polyploidy and the evolution D.L., CHEN, Z.J., LEE, H.S., COMAI, L., of domesticated plants. Amer. J. Bot. 80: MADLUNG, A., DOERGE, R.W., COLOT, V. & 1494–1499. MARTIENSSEN, R.A. (2003). Understanding HOLLIDAY, R. & PUGH, J.E. (1975). DNA mechanisms of novel gene expression in poly- modification mechanisms and gene activity ploids. Trends Genet. 19: 141–147. during development. Science 187: 226–232. OTTO, S.P. & WHITTON, J. (2000). Polyploid HUDSON, R.R. (1990). Gene geneologies and the incidence and evolution. Ann. Rev. Genet. 34: coalescent process. Oxford Surveys Evol. Biol. 401–437. 7: 1–44. PINKAVA, D.J. (2002). On the evolution of HUNT, D.R., TAYLOR N.P. & IOS WORKING PARTY. continental North American (1990). The genera of Cactaceae: progress (Cactaceae). Succ. Pl. Res. 6: 59–98. towards consensus. Bradleya 8: 85–107. PINKAVA, D.J., REBMAN, J.R. & BAKER, M.A. KIRK, J. T. O. & TILNEY-BASSETT, R.A.E. (1978). (1998). Chromosome numbers in some cacti of The plastids: their chemistry, structure and western North America, VII. Haseltonia 6: inheritance, 2nd edition. Elsevier/North 32–41. Holland, Amsterdam. RAMSEY, J. & SCHEMSKE, D. W. (1998). Pathways, LEVIN, D.A. (1983). Polyploidy and novelty in mechanisms, and rates of polyploid formation flowering plants. Amer. Nat. 122: 1–25. in flowering plants. Ann. Rev. Ecol. Syst. 29: LIU, B., HU, B., DONG, Y.Z., LIU, Z.L. & HE, M.Y. 467–501. (2000). Speciation-induced heritable cytosine RAMSEY, J. & SCHEMSKE, D.W. (2002). methylation changes in polyploid wheat. Prog. Neopolyploidy in flowering plants. Ann. Rev. Nat. Sci. 10: 601–606. Ecol. Syst. 33: 589–639. MATZKE, M.A. & MATZKE, A.J.M. (1998). RIGGS, A.D. (1975). X inactivation, Polyploidy and transposons. Trends Ecol. differentiation, and DNA methylation. Evol. 13: 241. Cytogenet. Cell Genet. 14: 9–11. MAUSETH, J.D., HSIAO, S.C. & MONTENEGRO, G. ROWLEY, G.D. (2003). Cristates, variegates and (1992). Vegetative body of the parasitic the asymmetric universe. Brit. Cact. Succ. J. angiosperm Ombrophytum subterraneum 21: 41–44. (Balanophoraceae). Bull. Torrey Bot. Club RUSSO, V. E. A., MARTIENSSEN, R. & RIGGS, A.D. 119: 407–417. (1996). Epigenetic mechanisms of gene regu- MAUSETH, J D., MONTENEGRO, G. & WALCKOWIAK, lation. Cold Spring Harbor Laboratory Press, A.M. (1984). Studies of the holoparasite Cold Spring Harbor. Tristerix aphyllus (Loranthaceae) infecting STEBBINS, G.L. (1938). Cytological characteristics Trichocereus chilensis (Cactaceae). Can. J. associated with the different growth habits in Bot.-Rev. Can. Bot. 62: 847–857. the dicotyledons. Amer. J. Bot. 25: 189–198. MAUSETH, J.D., MONTENEGRO, G. & WALCKOWIAK, TEL-ZUR, N., ABBO, S., BAR-ZVI, D. & MIZRAHI, Y. A.M. (1985). Host infection and flower forma- (2003). Chromosome doubling in vine cacti tion by the parasite Tristerix aphyllus hybrids. J. Hered. 94: 329–333. (Loranthaceae). Can. J. Bot.-Rev. Can. Bot. TILNEY-BASSETT, R.A.E. (1986). Plant chimeras. 63: 567–581. Edward Arnold, London.

14 Bradleya 22/2004