Analysis of the DNA-Based Reclassification of the Pleurothallidinae
  

 

By Lou Jost and Lorena Endara  

    Note added Jan 1, 2006: In the last two months many new results have been announced by Alec Pridgeon and colleagues, by Dr. Luer, by Hagen Stenzel, among others. These results help clear up some of the problem areas we have identified in this article. Hopefully we will soon find the time to incorporate these new results into the present article.

 

Like everyone else in the Pleurothallid world, we are trying to figure out what to do with Pridgeon, Solano, and Chase’s proposed reclassification of our favorite subtribe based on DNA analysis. We are trying to come to grips with this by looking closely at the internal logic of the proposal. We think that the discussion so far has oversimplified the issue. Many of their results mirror Dr Luer’s morphology-based taxonomy, so we cannot simply say that the whole DNA analysis is garbage.  On the other hand, some of their results deviate so greatly from the morphological evidence that they invite skepticism.  We must look carefully at the details of each of the changes proposed by Pridgeon and his collaborators, accepting those that are well-supported and holding off on the poorly supported ones. We feel strongly that the correct approach is not simply to accept or reject either this reclassification or the traditional one. We need all the clues we can get to help us with the taxonomy of this difficult group, and it would be a shame to just throw  away either Pridgeon’s work or Dr Luer’s lifetime of insights, as some have done.

            In spite of Pridgeon et al’s rhetoric against a morphology-based taxonomy, we are impressed with the overall convergence between Dr Luer’s morphology-based taxonomy and the DNA-based taxonomy of Pridgeon and company. Dr. Luer’s splitting of Dracula, Trisetella, and Dryadella from Masdevallia, his splitting of Trichosalpinx, Zootrophion, Dresslerella, Frondaria, and Ophidion from Pleurothallis, and his insight that Salpistele consists of two unrelated groups (first mentioned in Luer 1991), are all strongly supported by the work of Pridgeon et al.  In general, Dr Luer’s major subgeneric classifications are also congruent with  the DNA-based groupings, so much so that Pridgeon et al rely heavily on Dr Luer’s groupings to make nomenclatural transfers of  species they have not tested. Dr Luer criticizes this, but we see it rather as a tacit homage to the accuracy of Dr Luer’s work. The major difference between the two classifications is the treatment of the traditional genus Pleurothallis, but this genus had always been a dumping-ground for species with no distinctive features that could lift them into other genera, and it should come as no surprise that it can be broken up with the help of new genetic clues.

 

Dracula and Trichosalpinx, two genera first recognized by Dr. Luer on morphological grounds, and confirmed by the DNA analysis of Pridgeon et al.

 

      So let’s try to look at the details of the proposed new classification without prejudice. Our concept of “genus” plays a critical role in evaluating this reclassification. The genus concept has both a subjective, human component and an objective, biological component. Genera organize the otherwise-unmanageable diversity of life into bite-sized pieces. The size of each bite is, to some extent, up to us---this is the human, subjective dimension of the genus concept. We decide how much morphological diversity we should allow in each genus; if we split genera too finely, or lump too many species in each genus, the concept loses its power as a tool for organizing human knowledge.

            There are, however, some objective biological constraints on the genus concept. In modern taxonomy, the genus concept is wedded to the theory of evolution---our genera must reflect evolutionary descent. If we imagine a “family tree” that shows how species evolved and branched off from each other over the millenia, we are free to decide how far down the branches we should go to define our genera, but we are not free to form a genus by including a few twigs from one branch, a few from another branch, and a few more from another branch. Our genera should be “monophyletic”; once we decide how far up the tree our genus goes, we should include everything that descends from that branch, and nothing else. This way our genera will reflect evolutionary history. This is one of the principle goals of modern taxonomy. When genera reflect evolutionary history, we can use them to learn about how evolution happens. Genera that reflect evolutionary history are far more useful than artificial genera. Artificial genera may make species identification easier, but they are useless for making theoretical inferences about how plants evolve.

 
Suppose we can construct a correct "family tree" for a group of orchids, using DNA analysis or other techniques. If we find that the correct family tree is like the one on the left, with members of different genera mixed with each other on the same branches, then our genera do not reflect their evolutionary history. Such artificial genera are called "polyphyletic" (" many branches"). On the other hand, if the correct family tree is like the one on the right, our genera do reflect evolutionary history. These genera earn the term "natural" or "monophyletic" ("one branch"). The goal of modern taxonomy is to try to figure out evolutionary relationships, and if we have succeeded, our genera will be arranged like those on the right when a family tree is finally known for the group.

 

              DNA analysis looks at small pieces of each organism’s genetic code and, with the help of various mathematical tools, attempts to reconstruct the family tree of the organisms. The results can depend on the amount of the genetic code that gets sampled and the kinds of mathematical tools used. Pridgeon and his collaborators  have done not one but several different studies: a large study involving 185 species but looking at only one segment of their genetic code, a second large study looking at that same segment but adding two additional species, and a small study looking at only 54 pleurothallid species but  analyzing three different, independent segments of their genetic code. The latter analysis produced three independent trees, one for each of the three analyzed segments of the genetic code; Pridgeon et al. also present a tree that combines these results. All these trees differ from each other in some details. We think the discrepancies  indicate that more sampling needs to be done to produce a definitive reclassification. Nevertheless, some broad patterns  appear in all the trees. These patterns seem to us to be real, and should be accepted.(Click here to see the broad patterns that are supported by all the studies.)

 

The discordant elements within the genus Pleurothallis

Below is the basic pleurothallid family tree (see Broad Patterns ). I have shown in red the branches that include members of Pleurothallis (in its traditional sense). The positions of these are agreed upon in all three independent DNA analyses. Apparently Pleurothallis in its traditional sense is polyphyletic. (Note that the name Echinella in the figure has been changed to"Echinosepala" in Pridgeon and Chase, 2002, because Echinella had been used for an algae.)

 

      The most important result of Pridgeon et al. indicates that the genus Pleurothallis is made up of some (but not all) twigs of one branch, and some (but not all) twigs from other branches. All their analyses agree on this point. If we want our genera to reflect evolutionary history, we must either cut these branches lower down, creating a mega-genus that includes all the twigs, or cut the branches higher up, splitting Pleurothallis into several new genera. Because Pleurothallis is already a huge genus, it makes more sense to split it up rather than broaden it still further. People have been trying to split it up for over a century, but there were too many intermediate forms, and not enough clear morphological clues (Luer 1986c). The new DNA analysis gives us a way to look below the surface and unearth new clues, and with these we might at last be able to split the genus into units that reflect evolutionary relationships.

 All the DNA analyses convincingly show that the branch which contains Dr Luer’s Pleurothallis subgenus Acianthera (and a few small segregate subgenera) is widely separated from the rest of Pleurothallis. Furthermore, all these species  cluster together,  isolated from branches containing other genera, in all the different analyses performed by Pridgeon et al. This group of species is also morphologically reasonable (Luer 2002). We are convinced by this evidence, and accept Acianthera as a good genus....it coincides with morphological criteria for the most part, and helps clean up one of the largest and most difficult genera in the plant world.

 

Acianthera: These are the "core species" of the proposed genus Acianthera. By "core species" I mean the species that all analyses agree on; three different segments of DNA all agree that these species belong together. (Drawings of P. sicariae and P. ochreata were not available to me; drawings of the very similar species P. alpina and P. pubescens are substituted.) Click here to see not just the core species but ALL tested species that seem to belong to Acianthera.

All drawings on this site are used with the kind permission of Dr. C. A. Luer  and the Missouri Botanical Garden Press, and are taken from his Icones Pleurothallidinarum monographs published by the Missouri Botanical Garden Press. See References for full citations. The use of Dr. Luer's drawings does not imply that he agrees with me or with Pridgeon et al.

 

             Members of Pleurothallis subgenus Pleurothallis and subgenus Specklinia sit close to each other but on separate branches in all the family trees determined by the DNA analysis. The proposal to raise these subgenera to genus level seems well supported, though the exact limits of these new genera seem uncertain; different  analyses produce different branching patterns. This is one place where we might want to withhold judgement on the precise details pending further sampling, but the separation of Acianthera,  strict Pleurothallis, and Specklinia from the traditional genus Pleurothallis seems undeniable.

 

Pleurothallis (as proposed by Pridgeon et al.): These are the "core species" of the proposed genus Pleurothallis. (By "core species" I mean the species that all analyses agree on; three different segments of DNA all agree that these species belong together.) A drawing of P. cardiantha was not available to me; a drawing of the very similar species P. fornix is substituted.) Click here to see not just the core species but ALL tested species that seem to belong to Pleurothallis (s. s.).

 

Specklinia: These are some of the members of the proposed genus Specklinia. Here I include only those species for which three different segments of DNA were tested. Nevertheless those analyses are not unanimous in their placement of these species.  A drawing of P. costaricensis was not available to me; P. grobyi from the same section is substituted. Click here to see ALL tested species that seem to belong to Specklinia.

 

        

A very surprising proposed genus is Anathallis, also split off from the traditional genus Pleurothallis. In this case, the DNA evidence seems convincing that at least some of the species segregated into this new genus (consisting of Pleurothallis subgenus Acuminatae) are not closely related to Acianthera, Pleurothallis, or Specklinia, in spite of morphological similarities. All the different analyses of Pridgeon et al. agree that these species are related instead to the lepanthiform-sheathed genera Trichosalpinx, Lepanthes, Lepanthopsis, Zootrophion, and Frondaria. Indeed, subgenus Acuminatae do look like big Trichosalpinx species with reduced lepanthiform sheaths, and we have both mistaken them for Trichosalpinx at times. We accept this proposal of Pridgeon et al.

       In addition, analysis of one DNA segment suggests that subgenus Specklinia sect. Muscosae also belongs here. This is the nightmare of taxonomists---species whose affinities can only be determined by laboratory tests, not by external characters.  Nevertheless, in a complex group with much convergent evolution, this situation cannot be ruled out. And a perceptive eye may yet find characters that place them in their proper relationships. For example, Dr. Luer anticipated this result when he observed that many of the lips of Trichosalpinx subgenus Trichosalpinx   “are similar to those of some species in Pleurothallis subgenus Specklinia section Muscosae [one of the constituents of the proposed Anathallis] (Luer 1986a). Dr Luer goes on to say “Except for lacking ciliated sheaths, some Brazilian species [constituents of the proposed Anathallis] are practically inseparable [from Trichosalpinx subg. Trichosalpinx]”(Luer 1997). Anathallis (encompassing at least subgenus Acuminatia) should be accepted in spite of its surprising location in the family tree.

Perhaps a separate new genus should be established for the proposed members of Anathallis coming from Specklinia section Muscosae, since they are morphologically so different from the species coming from Acuminatae.

 

Anathallis: These are two of the species of this proposed genus. Analyses of three different segments of DNA confirm the placement of P. angustilabia and its relatives (subgenus Acuminatae) here. The placement of section Muscosae here (represented in the figure by P. pachyphyta) is based on just one DNA segment, but several species of this section were tested and found to group together here.

 

       It is interesting that Pridgeon et al do not propose splitting Trichosalpinx. Most of their analyses put the two subgenera of Trichosalpinx into different branchs. In fact, in some of their trees T. berlineri is in a different branch than T. blaisdellii. These two species are so similar that it would take a very brave geneticist indeed to split them into different genera. Their appearance in different branches suggests that the methods of Pridgeon et al are not completely reliable for resolving fine details. Nevertheless, we must recognize that this is exceptional in Pridgeon et al’s data set; in most cases, close relatives (as determined by morphology) cluster together. This is why we approach their findings with caution but not with disdain.        

 

        

        Their DNA analyses indicate that three oddball species (P. tripteranthera, P. mentosa, and P. mirabilis) that had been traditionally placed in subgenus Specklinia are rather isolated from that group. Pridgeon et al place these into a new genus, Anthereon, but they admit that no morphological features unite them. Furthermore, the different analyses presented by Pridgeon et al. put these species in different places in the family tree. We prefer to suspend judgement on Anthereon because of this, though the evidence is clear that P. tripterantha and P. mentosa are very closely related in spite of their morphological differences.

       Incidentally, an older name for this proposed genus is Pabstiella, which therefore takes precedence over Anthereon (see Barros 2002). We thank Mario Blanco for pointing out this reference. 

 

Anthereon: Analyses of all three DNA segments place P. tripterantha and P. mentosa (drawing not available; P. determannii is substituted) together, near the species of Pleurothallis s.s. The placement of P. mirabilis is less clear.

 

       Some of Pridgeon et al’s other proposed genera have rough edges. Their proposed genus Phloeophila, containing Luerella pelecaniceps, Pleurothallis peperomioides, and Ophidion pleurothallopsis, combines very different things; but their own data shows that there is a large genetic distance between them, and the placement of Ophidion is inconsistent in the different analyses. It is premature to decide whether to unite them or keep them separate. We withhold judgement on Phloeophila, and suspect that these three elements each deserve separate genera.

 

Phloeophila: Three rather different species make up this proposed group. The DNA evidence for the placement of Ophidion is not very clear.

 

 

            Perhaps the most unpalatable proposal of Pridgeon et al is the broadening of Stelis to include many species that are traditionally put into Pleurothallis. Stelis in its traditional sense is a huge yet easily recognizable genus. Making it even bigger and destroying its unity  should only be done at gunpoint. Fortunately, the DNA analysis does not really force us to accept this proposal of Pridgeon et al. As they themselves admit, this is a matter of deciding how far up the branches we should go to define our genera, and as we mentioned at the beginning of this article, such decisions are partly subjective (although there are recent efforts to make it less so). We can choose to maintain Stelis in its traditional sense without contradicting any of the findings in Pridgeon et al.

            However, doing so forces us to deal with the leftover branches that lie between our strict Stelis branch and the branches of  strict Pleurothallis. Maybe as many as six segregate genera would have to be established if we want to maintain traditional Stelis. Unfortunately the branching patterns in this part of the tree are not very clear, and differ depending on the analysis. We take this as a sign that decisions about segregate genera would be premature. This lack of clarity in the branching patterns is one reason Pridgeon et al proposed to unite the whole mess into a broadened Stelis. Regardless of how we choose to deal with this, all the independent studies agree that there is a distinct group of species that is more closely related to Stelis than to traditional Pleurothallis, and these must be accomodated either in an expanded Stelis or in new genera.

 

Stelis and relatives: These are the core species included in the proposed broadened Stelis. Analyses of all three DNA segments agree that these species are closely related to each other. Click here to see additional tested species that seem to fall into this group, based on analysis of a single DNA segment.

 

       Pridgeon et al’s  proposal to separate subgenus Silenia and Satyria (which share some very unusual morphological characteristics) from Myoxanthus had already been made by Dr Luer (1995). DNA analysis shows that at least subgenus Silenia is indeed separate. The DNA evidence places it close to Myoxanthus and Dresslerella, and morphology doesn’t contradict this. Pridgeon and Chase (2002) propose the genus Echinosepala for these species , and there seems little reason to resist this change.

            Likewise the proposed incorporation of Restrepiopsis into Pleurothallopsis seems reasonable; all the DNA studies agree on this point. Dr. Luer wrote that the flowers of Restrepiopsis are “inseperable from Pleurothallopsis except for the number of pollinia”(Luer 1986a). The only problem with this is the tremendous geographical distance ---2000 km (Luer 2002)---between the lone species of Pleurothallopsis and the species of Restrepiopsis. Still, the DNA evidence here is pretty clear.

            The proposal of the new genus Diodonopsis to hold Masdevallia erinacea and its relatives seems well supported by the DNA data, but we are bothered by the morphological  similarity of these plants to the bulk of Masdevallia. They are so similar to other Masdevallia that Dr. Luer did not even separate them into a subgenus of Masdevallia (Luer 1986b). We hold out on this one pending a larger sample of genes (only one DNA segment of Masdevallia erinacea was tested).

            The finding that  Pleurothallis subgenus Aenigma is closely related to Andinia is somewhat unexpected. The morphological evidence is ambiguous but does not rule this out. We have reservations about Pridgeon et al’s proposal to broaden Andinia to include Aenigma, but we have no reason to deny their conclusion that the two groups are closely related.

 
             The isolated position of Pleurothallis erinacea (subgenus Kraenzlinella) in the tree is bizarre. Though it has two pollinia, it falls into the group of genera with four to eight pollinia,  very far from any other member of traditional Pleurothallis. The genetic distance between it and its nearest neighbor in the tree, Brachionidium, is much larger than any other distance between two neighbors in the tree. Only one segment of DNA was analyzed for this species; perhaps this is a warning against undue reliance on a single gene, or perhaps the DNA analysis is telling us that subgenus Kraenzlinella is very special indeed. It is curious that Pridgeon et al do not even mention this anomaly, and in fact they suggest that Kraenzlinella might belong in their newly redefined genus Pleurothallis (Pridgeon and Chase 2001, p. 268). They fail to mention that the only tested member of Kraenzlinella appears in a completely different section of the tree than their Pleurothallis species. This is disturbingly careless.

 

       In addition to proposing new genera, Pridgeon et al. also propose to eliminate some existing genera, such as Salpistele, Jostia, and Acostaea. Most of these are eliminated to maintain the monophyly of a larger genus which embeds them. If the DNA analysis is well done, we are obligated either to eliminate such a genus or split up the larger genus that it is embedded in. However, the exact placement of these genera in the tree is not certain; Acostaea for example is placed in different branches depending on the analysis, and in one of the trees it is just barely embedded in Specklinia. We think it is premature to eliminate these small, morphologically distinctive genera until we know exactly where they belong in the tree. They may or may not be embedded in larger genera.

            We conclude that the work of Pridgeon et al. should neither be accepted as the last word, nor dismissed out of hand. We feel that the “consensus genera”, those that are unanimously supported in all of Pridgeon et al’s analyses and that seem reasonable on morphological grounds, should be accepted. These include Pleurothallis (in a new, strict sense) and Specklinia, Acianthera, Anathallis, a broadened Pleurothallopsis, and Echinosepala. On the other hand, it would be premature to accept genera whose placement is ambiguous and which are suspicious on morphological grounds, such as Phloeophila. The proposal of Pridgeon et al. to greatly broaden Stelis is perhaps excessive, but if we reject that proposal we must establish one or more new genera to hold the leftovers, which cannot be placed in Pleurothallis (all the independent DNA studies agree on this point). We withhold judgement (but remain open-minded) about the elimination of morphologically well-defined genera like Acostaea, pending sampling of more genes.

 

      We hope that the controversy surrounding the proposals of Pridgeon et al. will lead to more discussion and especially to more detailed DNA analyses that could resolve some of the ambiguities. We hope to see sampling of many more species (Pridgeon et al sampled only 185 out of 4000 species) and of more genes.  We should also look for new morphological characters that correlate with the groups proposed by Pridgeon et al. We are excited by the prospect of new molecular clues to the evolutionary history of the Pleurothallidinae, but we hope that future studies involve more collaboration with traditional taxonomists. We need a synthesis of all available insights, not a gridlock between two opposing camps. 

 

Note: For a different view of these results, see Dr. Luer's recent article in Selbyana 23(1): 57-110.  

 

References

Barros, F. 2002. Notas nomenclaturais em Pleurothallidinae (Orchidaceae), principalmente brasilieras. Bradea 8: 293-297.

Luer, C. A. 1986a. Icones Pleurothallidinarum I: Systematics of the Pleurothallidinae. St. Louis: Missouri Botanical Garden.

 Luer, C. A. 1986b. Icones Pleurothallidinarum II: Systematics of Masdevallia. St. Louis: Missouri BotanicalGarden.

Luer, C. A. 1986c. Icones Pleurothallidinarum III: Systematics of Pleurothallis. St. Louis: Missouri Botanical Garden.

Luer, C. A. 1995. Icones Pleurothallidinarum XII: Systematics of Brachionidium. St. Louis: Missouri Botanical Garden.

Luer, C. A. 1997. Icones Pleurothallidinarum XV: Systematics of Trichosalpinx. St. Louis: Missouri Botanical Garden.

Luer, C. A. 2002. A systematic method of classification of the Pleurothallidinae versus a strictly phylogenetic method. Selbyana 23(1): 57-110

Pridgeon, A., R. Solano, and M. Chase. 2001. Phylogenetic relationships in Pleurothallidinae (Orchidaceae): combined evidence from nuclear and Plastid DNA sequences. American Journal of Botany 88(12): 2286-2308.

Pridgeon, A., and M. Chase. 2001. A phylogenetic reclassification of Pleurothallidinae(Orchidaceae). Lindleyana 16: 235-271.

Pridgeon, A., and M. Chase. 2002. Nomenclatural notes on Pleurothallidinae(Orchidaceae). Lindleyana 17: 98-101.

 

 
Analysis of the DNA-Based Reclassification of the Pleurothallidinae