Sabtu, 08 Maret 2008

Mitochondrial DNA sequences reveal the photosynthetic relatives of Rafflesia, the world's largest flower

Proc Natl Acad Sci U S A. 2004 January 20; 101(3): 787–792.
Published online 2004 January 8. doi: 10.1073/pnas.0305562101.
PMCID: PMC321759
From the Cover
Mitochondrial DNA sequences reveal the photosynthetic relatives of Rafflesia, the world's largest flower
Todd J. Barkman,* Seok-Hong Lim,* Kamarudin Mat Salleh, and Jamili Nais§
*Department of Biological Sciences, Western Michigan University, Kalamazoo, MI 49008; School of Environmental and Natural Resources Science, Universiti Kebangsaan Malaysia, 43600 Bangi, Selangor, Malaysia; and §Sabah Parks, 88806 Kota Kinabalu, Sabah, Malaysia
To whom correspondence should be addressed. E-mail:
Edited by Jeffrey D. Palmer, Indiana University, Bloomington, IN, and approved November 7, 2003
Received September 1, 2003.
All parasites are thought to have evolved from free-living ancestors. However, the ancestral conditions facilitating the shift to parasitism are unclear, particularly in plants because the phylogenetic position of many parasites is unknown. This is especially true for Rafflesia, an endophytic holoparasite that produces the largest flowers in the world and has defied confident phylogenetic placement since its discovery >180 years ago. Here we present results of a phylogenetic analysis of 95 species of seed plants designed to infer the position of Rafflesia in an evolutionary context using the mitochondrial gene matR (1,806 aligned base pairs). Overall, the estimated phylogenetic tree is highly congruent with independent analyses and provides a strongly supported placement of Rafflesia with the order Malpighiales, which includes poinsettias, violets, and passionflowers. Furthermore, the phylogenetic placement of Mitrastema, another enigmatic, holoparasitic angiosperm with the order Ericales (which includes blueberries and persimmons), was obtained with these data. Although traditionally classified together, Rafflesia and Mitrastema are only distantly related, implying that their endoparasitic habits result from convergent evolution. Our results indicate that the previous significant difficulties associated with phylogenetic placement of holoparasitic plants may be overcome by using mitochondrial DNA so that a broader understanding of the origins and evolution of parasitism may emerge.


Using the recent ordinal classification of angiosperms (11) as a guide, we explicitly sampled at least one family from 43 of 45 monophyletic orders to produce a broad mtDNA phylogenetic framework. Sampling included 92 species from 80 angiosperm families. Of primary interest for this study was the inclusion of Rafflesia keithii and its close parasitic relative Rhizanthes zippelii. Both of these species traditionally have been placed in Rafflesiaceae sensu stricto. Mitrastema yamamotoi (Mitrastemonaceae), another extreme holoparasite that has long been placed in Rafflesiaceae sensu lato, also was sampled to determine its phylogenetic affinities. Like Rafflesia, this parasite grows as an endophyte within its host (various members of the oak family, Fagaceae), emerging only during flower production (3). Because the morphology of Mitrastema is so divergent from that of photosynthetic plants, its affinities have remained obscure, with no relatives ever proposed aside from other endophytic parasites like Rafflesia (27). Three gymnosperms, Pinus, Ginkgo, and Zamia, were included as outgroups to root phylogenetic estimates. Table 1, which is published as supporting information on the PNAS web site, lists all of the species included in this study, the GenBank accession numbers for all of the sequences analyzed, and the voucher numbers for the newly generated sequences. Molecular methods, including DNA extraction, and DNA sequencing were performed as previously described (19). Details of the PCR methodology, including the primer sequences used, are available as Supporting Text, which is published as supporting information on the PNAS web site. In total, 55 matR mtDNA sequences were generated for this study.

clustalx (28) was used to produce a preliminary alignment of the matR sequences that was followed by minor manual adjustments. The aligned data matrix is available from Tree-BASE ( (S981–M1631). Regions of uncertain alignment were excluded before analysis; however, their inclusion did not alter the fundamental conclusions of this study. After exclusion, there were 1,499 included characters and 636 parsimony-informative characters. Unweighted parsimony analyses were conducted by using paup* v. 4.0b10 (29) with 100 random addition sequences, and tree bisection–reconnection swapping. Bootstrap support (BS) (30) and jackknife support (JS) (with 33% character deletion) values were obtained from 2,000 replicates by using “fast” stepwise addition. Although the fast stepwise addition analyses are expected to provide estimates of support that are less than those obtained when comprehensive branch-swapping analyses are performed (31, 32), such analyses were not computationally feasible with this data set. modeltest v. 3.06 (33) was used to determine the best-fit model of nucleotide substitution for the data set, and this best-fit model [K81 (34), assuming unequal nucleotide frequencies and a Γ parameter allowing for rate heterogeneity] then was implemented during Bayesian analyses performed by using mrbayes v. 3.0b4 (35). Four chains were run simultaneously for one million generations, and these were sampled every 100 generations. The first 10,000 generations were discarded as the “burn-in” period, and posterior probabilities (PP) for individual clades then were obtained from the remaining samples. Neighbor-joining (NJ) analyses were performed by assuming the optimal model of nucleotide substitution chosen with modeltest, and BS was obtained from 2,000 replicates by using paup* v. 4.0b10.


Parsimony analyses of matR from 95 species of seed plants resulted in 61,911 equal-length trees of 2,464 steps [consistency index = 0.5657]. The strict consensus tree (Fig. 1) represents an mtDNA-only estimate of broad, angiospermwide phylogeny, and it is largely congruent with estimates based on independent plastid and nuclear DNA sequences (911, 36). Levels of parsimony BS (P-BS) and Bayesian PP are shown on all nodes receiving estimates higher than 50 or 0.5, respectively. The complete majority-rule consensus tree obtained from the Bayesian analysis is available as Fig. 3, which is published as supporting information on the PNAS web site. Parsimony JS (P-JS) values and NJ-BS values could not be shown in Fig. 1 but are presented below. The complete set of JS values and the NJ tree with BS values are available in Figs. 4 and 5, respectively, which are published as supporting information on the PNAS web site. In general, the P-BS for all nodes obtained with fast stepwise addition was lower than the JS, NJ-BS, or PP and may be conservative. The P-BS values and PP shown for each node may therefore be interpreted as the lower and upper bounds of node reliability, respectively (37). The strict consensus tree suggests that the members of the ANITA (Amborella, Nymphaeales, and Austrobaileyales) clade (19, 22, 23) are the basal-most angiosperms, followed by an unresolved set of relationships among magnoliid orders, including all monocots. Expected major eudicot relationships include monophyletic rosids and asterids. All recently recognized orders are monophyletic when represented by more than one family, and, in most cases, BS and PP are high at the ordinal level (Fig. 1).

Fig. 1.Fig. 1.
Strict consensus tree from an unweighted parsimony analysis of matR mtDNA sequences (number of trees = 61,911, tree length = 2,464, consistency index = 0.5657). BS from the parsimony analysis is listed before the slash, and PP from the Bayesian analysis (more ...)

Within the context of this global angiosperm mtDNA phylogeny, the placement of Rafflesia as sister to the Malpighiales is supported (P-BS = 93, P-JS = 97, NJ-BS = 98, PP = 1.0). Parsimony analyses of amino acid sequences also suggested the same relationship (P-BS = 77, P-JS = 89) (Fig. 6, which is published as supporting information on the PNAS web site). The nested position of Rafflesia within the monophyletic Oxalidales + Malpighiales (P-BS < js =" 56," bs =" 57," pp =" 1.0)" href="" class="cite-reflink">10, 11). Not supported by these data were traditional placements of Rafflesia + Rhizanthes with either Piperales (38) or Santalales (27).

Placement of the enigmatic Southeast Asian holoparasitic plant Mitrastema (Mitrastemonaceae) with the Ericales (Fig. 1) was obtained by using the matR mtDNA phylogenetic framework produced here (P-BS = 70, P-JS = 82, NJ-BS = 91, PP = 1.0). This phylogenetic placement of Mitrastema with Ericales was surprising but is supported by parsimony analyses of amino acid data as well (P-BS = 76, P-JS = 91) (Fig. 6). Whereas Mitrastema has long been classified with Rafflesia + Rhizanthes, or close to them, these taxa are only distantly related.


Holoparasite Phylogenetic Placement. The positions of the holoparasites, Rafflesia + Rhizanthes and Mitrastema, inferred with matR sequence data, are robust for several reasons. First, our results are not method-dependent, because the same strongly supported placements were obtained with parsimony, NJ, and Bayesian analysis. Second, long-branch attraction (39) does not seem to have misled these analyses, because the inclusion of random or misaligned sequences (40) did not influence the placement of Rafflesia + Rhizanthes or Mitrastema (Figs. 7 and 8, which are published as supporting information on the PNAS web site). Furthermore, it is clear that these lineages have not been artifactually placed with any of the other longest branches in this data set (Fig. 2A). Third, RNA editing and the potential for processed paralogy in plant mt genomes can confound phylogenetic estimates (41); however, it is unlikely that the placement of these holoparasites has been affected, because the removal of known matR RNA edit sites from the data set (4244) did not change the level of support for the inferred relationships (Fig. 9, which is published as supporting information on the PNAS web site). Finally, because we have sampled the host plants of the parasites (or their close relatives), contamination cannot be invoked as an explanation for our results.

Fig. 2.Fig. 2.
(A) A randomly chosen phylogram of one of the equal-length trees from the matR parsimony analysis in Fig. 1. Branch lengths are shown only above lineages that had lengths of >50 steps. (B) Phylogram showing estimated 18S branch lengths on a topology (more ...)

Horizontal gene transfer (45, 46) of matR between Rafflesia and/or Rhizanthes and a member of the Malpighiales, or between Mitrastema and a member of the Ericales, is also an unlikely explanation for these results for several reasons. First, although host-to-parasite transfer of macromolecules is possible (47), Rafflesia and Rhizanthes are only parasitic on species of Tetrastigma (Vitaceae) to which they are clearly unrelated (Fig. 1). Likewise, Mitrastema only parasitizes members of the Fagales. Second, horizontal transfer seems to be rare in flowering plants overall (45) and would have to be postulated to have affected only Rafflesia, Rhizanthes, and Mitrastema, because all other nodes with high BS or JS in this analysis are comparably placed by independent estimates of angiosperm phylogeny (11). Finally, phylogenetic analysis of 128 coxI mtDNA sequences suggests comparable phylogenetic placements of Rafflesia with Malpighiales and Mitrastema with Ericales, although with lower support (C. W. dePamphilis, personal communication). Although confident placement of these parasites has been achieved with these data, more studies, including more taxa and characters, are needed to refine the positions of Rafflesia + Rhizanthes and Mitrastema to determine whether they are nested within, or sister to, the Malpighiales and Ericales, respectively. In either case, expanded circumscription of these orders is needed to include these holoparasites.

Rafflesia and Its Photosynthetic Relatives. The relationship between Rafflesia + Rhizanthes and Malpighiales estimated with matR mtDNA has not been suggested in any angiosperm classification or phylogenetic analysis. Not since 1822, when Robert Brown noted similarities in floral structure between Rafflesia and the Passifloraceae (12), has a potential relationship to the members of the currently circumscribed Malpighiales been suggested. Within the order, Rafflesia is morphologically most similar to Passifloraceae, with which it shares a hypanthium (perigone tube of Rafflesia), an androgynophore (central column of Rafflesia), and an annular corona (diaphragm of Rafflesia), a suite of characteristics that are otherwise somewhat rare in angiosperms (27). Some of these morphological features are found in other members of Malpighiales; therefore, the potential homology of these features may be ascertained only when the placement of Rafflesia, relative to a broader sampling of the order, is determined. The phylogenetic placement suggested by our data implies that a massive increase in flower size has occurred since the origin of Rafflesia, because the largest Malpighialian flowers are only 10 cm in diameter. Future comparative studies of MADS box and other floral developmental genes from Rafflesia and Malpighiales may reveal the genetic basis for this evolutionary floral size increase.

The placement of Rafflesia with Malpighiales also makes particularly tantalizing the 100-year-old observation that an individual of the tropical vine, Passiflora caerulea, was parasitic on another angiosperm (48). Although this may have been an aberrant observation of Passiflora, it is interesting to note that because Rafflesia exists as a vine-like endophyte within its host, the association of the viney growth form with parasitism is found in at least three separate orders: Laurales (49), Solanales (50), and Malpighiales. Although there may be different ancestral conditions giving rise to parasitism, the close physical association between vines and the plants that support them suggests that this interaction may foster the transition to a parasitic lifestyle.

The Asterid Holoparasite Mitrastema. The general placement of Mitrastema within the asterids has not been suggested previously, but its gamopetalous corolla supports this position. The specific placement of Mitrastema with the Ericales is further supported by a suite of morphological characteristics, including opposite and decussate leaves, parietal placentation, and circumscissile fruit dehiscence, that are also found in various members of the order but which collectively do not link it with any family in particular. The traditionally suggested close relationship between Rafflesia and Mitrastema is not supported by our data and probably reflects a taxonomic emphasis on their similar endoparasitic lifestyle. In fact, the floral morphology of Rafflesia is very different from Mitrastema, which has small (≈2.54 cm in diameter), white, bisexual flowers, a dehiscent staminal tube, and a superior ovary (Fig. 1). Because Mitrastema is no more divergent in floral morphology from Rafflesia than are any of the other traditional Rafflesiaceae genera, Cytinus, Bdallophyton, Apodanthes, and Pilostyles, it is possible that they represent independent parasitic lineages as well and should be targeted for future studies using mtDNA sequences.

Holoparasite mtDNA Evolution. The use of mt sequences to study parasitic plant phylogeny seems promising because the holoparasitic angiosperm family Hydnoraceae was recently placed by using a combination of mtDNA with plastid and nuclear sequences (51). One reason for the utility of mtDNA for studying the phylogeny of holoparasites is that the evolutionary dynamics governing these sequences seem to be similar in parasitic plants and their free-living relatives. Fig. 2 A shows reconstructed branch lengths from one of the equal-length trees obtained in the parsimony analysis of matR. It is clear that branch lengths are heterogeneous across angiosperms in general and that numerous photosynthetic plants are as divergent as Rafflesia or nearly so, yet all are confidently placed [except Ceratophyllum, which has defied confident placement to date (11)]. Furthermore, a likelihood ratio test (52) was performed to compare the level of selective constraint (estimated by the nonsynonymous/synonymous rate ratio, dN/dS) on matR in the holoparasitic lineages (Rafflesia, Rhizanthes, and Mitrastema) and photosynthetic lineages. Although the estimated dN/dS was higher for the holoparasitic lineages (1.0646) than for the photosynthetic lineages (0.7010), this difference was not statistically significant (P > 0.05). Thus, although some holoparasitic plants show elevated substitution rates and significant decreases in selective constraint for chloroplast genes such as rbcL and rps2 compared with their photosynthetic relatives (5, 6), matR has likely been maintained for efficient mt functioning even in parasitic plants.

Problems with Holoparasite 18S Nuclear DNA. In addition to mtDNA, nuclear sequence data could provide an important independent genomic estimate of holoparasitic relationships as well. However, the 18S nuclear ribosomal DNA and other loci involved in protein translation have been problematic in studying other extreme parasites like microsporidia (2), and these genes are unlikely to provide useful data for Rafflesia or most other plant holoparasites. To illustrate this point for Rafflesia (Mitrastema has not yet been sequenced), we have obtained 70 nuclear 18S DNA sequences from GenBank for taxa that are also represented in our matR data set. Table 2, which is published as supporting information on the PNAS web site, lists all species names and GenBank accession numbers. Fig. 2B shows branch lengths for the 18S sequence data on a tree that was constrained to represent the currently accepted estimate of angiosperm relationships (10, 11). Assuming a placement of Rafflesia as sister to the Malpighiales, it is clear that this holoparasite has a highly divergent 18S sequence that is 4 times longer (197 steps) than the next longest sequence in the data set (the branch separating angiosperms and gymnosperms; 49 steps), and it is 6–8 times longer than most other branches. As expected with long-branch attraction, Rafflesia is placed with the other longest branch in the data set in an unconstrained parsimony analysis (Fig. 2C). This result corroborates an earlier study of 18S that found Rafflesia artifactually placed as sister to angiosperms and indicates that this gene is of limited utility for studying this divergent holoparasite (18). Even phylogenetic analyses of the slowly evolving 16S plastid ribosomal DNA clearly show extreme divergences in holoparasites (4), indicating that this gene will also be of limited utility in studying parasitic plants like Rafflesia. Classes of nuclear gene sequences other than those involved in translation, such as loci involved in primary metabolism, may instead provide better candidates for studying holoparasite relationships (2). Future studies of nuclear genes from Rafflesia and other extreme holoparasitic plants are needed because no other loci besides the 18S ribosomal DNA have been sequenced.

Implications. The results of this study have clear implications for studies of angiosperm phylogeny and parasite evolution. First, we show that mtDNA sequences provide a third independent genomic estimate of angiosperm relationships useful for resolving a variety of phylogenetic questions among both anciently and more recently evolving lineages. Second, placement of several other parasitic plants [composing nearly half of the taxa that remain of uncertain positions at the ordinal level (11)] may now be possible using the approach adopted here. Third, the eventual phylogenetic placement of all parasitic plants within flowering plant phylogeny will enable detailed comparative studies aimed at inferring ancestral states that may have facilitated the repeated evolution of parasitism in angiosperms.

Supplementary Material

Supporting Information
Click here to view.

This paper was submitted directly (Track II) to the PNAS office.
Abbreviations: mt, mitochondrial; BS, bootstrap support; JS, jackknife support; PP, posterior probabilities; NJ, neighbor joining; P-BS, parsimony BS; P-JS, parsimony JS.
Data deposition: The sequences reported in this article have been deposited in the GenBank database (accession nos. AY453070–AY453124).

Combes, C. (2001) Parasitism: The Ecology and Evolution of Intimate Interactions (Univ. of Chicago Press, Chicago).
Keeling, P. J., Lucker, M. A. & Palmer, J. D. (2000) Mol. Biol. Evol. 17:, 23–31. [PubMed].
Kuijt, J. (1969) The Biology of Parasitic Flowering Plants (Univ. of California Press, Berkeley).
Nickrent, D. L., Duff, R. J., Colwell, A. E., Wolfe, A. D., Young, N. D., Steiner, K. E. & dePamphilis, C. W. (1998) in Molecular Systematics of Plants II: DNA Sequencing, eds. Soltis, D. E., Soltis, P. S. & Doyle, J. J. (Kluwer Academic, Boston), pp. 211–241.
dePamphilis, C. W., Young, N. D. & Wolfe, A. D. (1997) Proc. Natl. Acad. Sci. USA 94:, 7367–7372. [PubMed].
Leebens-Mack, J. & dePamphilis, C. W. (2002) Mol. Biol. Evol. 19:, 1292–1302. [PubMed].
Nickrent, D. L., Ouyang, Y., Duff, R. J. & dePamphilis, C. W. (1997) Plant Mol. Biol. 34:, 717–729. [PubMed].
dePamphilis, C. W. & Palmer, J. D. (1990) Nature 348:, 337–339. [PubMed].
Soltis, P. S., Soltis, D. E. & Chase, M. W. (1999) Nature 402:, 402–404. [PubMed].
Soltis, D. E., Soltis, P. S., Chase, M. W., Mort, M. E., Albach, D. C., Zanis, M., Savolainen, V., Hahn, W. H., Hoot, S. B., Fay, M. F., et al. (2000) Bot. J. Linn. Soc. 133:, 381–461.
The Angiosperm Phylogeny Group (2003) Bot. J. Linn. Soc. 141:, 399–436.
Brown, R. (1822) Trans. Linn. Soc. London 13:, 201–234.
Nais, J. (2001) Rafflesia of the World (Natural History Publications, Kota Kinabalu, Malaysia).
Beaman, R. S., Decker, P. J. & Beaman, J. H. (1988) Amer. J. Bot. 75:, 1148–1162.
Takhtajan, A. L. (1997) Diversity and Classification of Flowering Plants (Columbia Univ. Press, New York).
Heywood, V. H., Moore, D. M., Richardson, I. B. K. & Stern, W. T. (1993) Flowering Plants of the World (Oxford Univ. Press, New York).
Nickrent, D. L. & Starr, E. M. (1994) J. Mol. Evol. 39:, 62–70. [PubMed].
Lipscomb, D. L., Farris, J. S., Källersjö, M. & Tehler, A. (1998) Cladistics 14:, 303–338.
Barkman, T. J., Chenery, G., McNeal, J. R., Lyons-Weiler, J., Ellisens, W. J., Moore, G., Wolfe, A. D. & dePamphilis, C. W. (2000) Proc. Natl. Acad. Sci. USA 97:, 13166–13171. [PubMed].
Bowe, L. M., Coat, G. & dePamphilis, C. W. (2000) Proc. Natl. Acad. Sci. USA 97:, 4092–4097. [PubMed].
Chaw, S.-M., Parkinson, C. L., Cheng, Y., Vincent, T. M. & Palmer, J. D. (2000) Proc. Natl. Acad. Sci. USA 97:, 4086–4091. [PubMed].
Parkinson, C. L., Adams, K. L. & Palmer, J. D. (1999) Curr. Biol. 9:, 1485–1488. [PubMed].
Qiu, Y.-L., Lee, J., Bernasconi-Quadroni, F., Soltis, D. E., Soltis, P. S., Zanis, M., Zimmer, E. A., Chen, Z., Savolainen, V. & Chase, M. W. (1999) Nature 402:, 404–407. [PubMed].
Nickrent, D. L., Parkinson, C. L., Palmer, J. D. & Duff, R. J. (2000) Mol. Biol. Evol. 17:, 1885–1895. [PubMed].
Wolfe, K. H., Li, W.-H. & Sharp, P. M. (1987) Proc. Natl. Acad. Sci. USA 84:, 9054–9058. [PubMed].
Duff, R. J. & Nickrent, D. L. (1997) J. Mol. Evol. 45:, 631–639. [PubMed].
Cronquist, A. (1981) An Integrated System of Classification of Flowering Plants (Columbia Univ. Press, New York).
Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmougin, F. & Higgins, D. G. (1997) Nucleic Acids Res. 24:, 4876–4882.
Swofford, D. L. (2002) PAUP*: Phylogenetic Analysis Using Parsimony (* and Other Methods) (Sinauer, Sunderland, Massachusetts), Version 4.0b10.
Felsenstein, J. (1985) Evolution 39:, 783–791.
Mort, M. E., Soltis, P. S., Soltis, D. E. & Mabry, M. L. (2000) Syst. Biol. 49:, 160–171. [PubMed].
DeBry, R. W. & Olmstead, R. G. (2000) Syst. Biol. 49:, 171–179. [PubMed].
Posada, D. & Crandall, K. A. (1998) Bioinformatics 14:, 817–818. [PubMed].
Kimura, M. (1981) Proc. Natl. Acad. Sci. USA 78:, 454–458. [PubMed].
Huelsenbeck, J. P. & Ronquist, F. (2001) Bioinformatics 17:, 754–755. [PubMed].
Savolainen, V., Chase, M. W., Hoot, S. B., Morton, C. M., Soltis, D. E., Bayer, C., Fay, M. F., deBruijn, A. Y., Sullivan, S. & Qiu, Y.-L. (2000) Syst. Biol. 49:, 306–362. [PubMed].
Douady, C. J., Delsuc, F., Boucher, Y., Doolittle, W. F. & Douzery, E. J. P. (2003) Mol. Biol. Evol. 20:, 248–254. [PubMed].
Bouman, F. & Meijer, W. (1994) Plant Syst. Evol. 193:, 187–212.
Felsenstein, J. (1978) Syst. Zool. 27:, 401–410.
Qiu, Y.-L., Lee, J., Whitlock, B. A., Bernasconi-Quadroni, F. & Dombrovska, O. (2001) Mol. Biol. Evol. 18:, 1745–1753. [PubMed].
Bowe, L. M. & dePamphilis, C. W. (1996) Mol. Biol. Evol. 13:, 1159–1166. [PubMed].
Thomson, M. C., Macfarlane, J. L., Beagley, C. T. & Wolstensholme, D. R. (1994) Nucleic Acids Res. 22:, 5745–5752. [PubMed].
Begu, D., Mercado, A., Farre, J.-C., Moenne, A., Holuigue, L., Araya, A. & Jordana, X. (1998) Curr. Genet. 33:, 420–428. [PubMed].
Giege, P. & Brennicke, A. (1999) Proc. Natl. Acad. Sci. USA 96:, 15324–15329. [PubMed].
Bergthorsson, U., Adams, K. L., Thomason, B. & Palmer, J. D. (2003) Nature 424:, 197–201. [PubMed].
Won, H. & Renner, S. S. (2003) Proc. Natl. Acad. Sci. USA 100:, 10824–10829. [PubMed].
Haupt, S., Oparka, K. J., Sauer, N. & Neumann, S. (2001) J. Exp. Bot. 52:, 173–177. [PubMed].
Pée-Laby, M. E. (1904) Rev. Gen. Bot. 16:, 453–457.
Rohwer, J. G. (2000) Syst. Bot. 25:, 60–71.
Stefanovic, S., Krueger, L. & Olmstead, R. G. (2002) Amer. J. Bot. 89:, 1510–1522.
Nickrent, D. L., Blarer, A., Qiu, Y.-L., Soltis, D. E., Soltis, P. S. & Zanis, M. (2002) Amer. J. Bot. 89:, 1809–1817.
Yang, Z. (1998) Mol. Biol. Evol. 15:, 568–573. [PubMed].

Tidak ada komentar:


Template Design By: