If It Ain’t Broke, Don’t Fix It: Why Splitting Anolis Is a Bad Idea

Nicholson et al. (2014) provide two reasons that Anolis should be divided into eight genera. The first is simply that this is what modern systematists do, breaking up big groups into little groups. But this is not really a very compelling argument in its own right. As your mother used to tell you, “just because everyone else is jumping into a lake doesn’t mean you should, too.”

Their second reason is more specific. They argue that more finely divided groups are essential for understanding patterns of evolution and diversity:

“When new monophyletic structure is revealed in groups for which such structure was previously unrecognizable, taxonomy should change to incorporate that new information. This process does reveal constructs inherent to the natural world and, therefore, forces us to change the way we train future generations of biologists, design future comparative analyses, and interpret new data.”

But let’s take apart this argument. First, have Nicholson et al. revealed “new monophyletic structure” that was “previously unrecognizable”? The authors themselves point out that these groups have been found in all recent systematic studies and, indeed, the community has been well aware of them. One only needs to go back to Jackman et al. (1999) to see recognition of 17 clades, and discussion of these clades has continued ever since. Nicholson et al. would have readers believe that we thought that Anolis was one big, unstructured group of 400 species, but that is a very incorrect portrayal of the widespread understanding of anole phylogeny.

More importantly, second, has our understanding of anole evolution and diversity suffered because we have considered anoles as one genus instead of eight? This is a pretty hard argument to make. For a quarter of a century, work on anoles has been an exemplar of how to incorporate phylogenetic information into studies of evolutionary diversity. For example, anoles are now known as a model case of convergent evolution—it’s not like we’ve failed to recognize that convergence just because they’re all called “Anolis.” It’s striking in this regard that the sole paper cited by Nicholson et al. (2014) to support their contention that a single genus is problematic is nearly 30 years old! Thirty years of copious anole research shows that recognition of a single, large genus has not hindered anole research in the slightest.

Indeed, it is thought-provoking to compare the contributions made by research on anoles in the Caribbean and on the mainland. Workers on Caribbean anoles almost without exception adhere to the single genus framework, and work on Caribbean anoles has been extensive and is now well-known by the community at large. In contrast, mainland anole researchers are more divided, some favoring a single genus, others favoring multiple genera. It would be hard to compare the quantity and impact of work on Caribbean and mainland anoles and argue that recognizing multiple genera has accelerated research.

Finally, we might take a step back and ask: Is it still the case that big monophyletic groups must be broken up into little monophyletic groups to foster evolutionary research? This viewpoint may have been correct in the 1980’s when the cladistics revolution occurred, but is it really true, in this day and age, that researchers look to taxonomy to derive their evolutionary thinking? I would suggest that this is no longer true: with the huge explosion of phylogenetic thinking, modern researchers no longer look at taxonomic classifications to identify evolutionary groups; rather, they go straight to the phylogenies themselves. To see that this is true, pick up any journal and look at the evolutionary analyses. No one is basing their research on taxonomies; they are basing them on the phylogenies themselves, regardless of the binomial names appended to the terminal taxa. The argument that splitting up clades ever more finely to enhance research is old-fashioned, a hold over from the days when phylogenetic thinking was uncommon and many recognized groups were not monophyletic.

Still, Nicholson et al. are correct: it is the trend to ever more finely split up clades into smaller genera. Why should anoles be different? The answer is that there is an enormous, half-century of literature on these lizards that extends far beyond the fields of herpetology and systematics. Researchers in areas as disparate as physiology, cell biology, development and functional morphology, as well as evolution, ecology, and behavior, have conducted important work on anoles. And this work has been published using the name “Anolis.”

Nicholson et al. (2014) don’t even address the point raised by Poe and many others, that such name changes are more than disruptive, but truly damaging to scholarly research. In the herpetology course I teach, students are given a number of assignments that require them to go into the literature. And they have great trouble tracking down information on species whose names have changed. That’s not even accurate–usually they are simply unaware that there is a literature on a species under its former name. It would be nice to think that they would consult online resources that provide lists of the different names a species has had, but the students usually aren’t savvy enough (even though they are instructed to do so). In this regard, molecular biologists, physiologists and ethologists are no better than undergrads. If they read a paper on Anolis cristatellus from 1983, they will not know to connect that paper to a species now known as Ctenonotus cristatellus. Proponents of name changes tend to brush this under the rug, saying that people are adaptable and will cope, but this view is unrealistic.

Some day, all scholarly resources will be digital and species names will be automatically hyperlinked to their previous identities, but we are a long way from that point. And until that day, there are real costs to changing names, particularly for well-known groups with a long history of research. Given that research on Anolis is vibrant and phylogenetically informed, there is nothing to be gained and a lot to be lost by division into multiple genera.

About Jonathan Losos

Professor and Curator of Herpetology at the Museum of Comparative Zoology at Harvard University. I've spent my entire professional career studying anoles and have discovered that the more I learn about anoles, the more I realize I don't know.

30 thoughts on “If It Ain’t Broke, Don’t Fix It: Why Splitting Anolis Is a Bad Idea

  1. Revision shock. Developmental Biologists went through all this when developmental gene families were discovered in a piecemeal fashion in different species back in the late 80s. Who would now argue in favor of a non-integrated Hox gene nomenclature? But you wouldn’t believe how upset many people were at the time, particularly the gene discoverers.

  2. Very well-reasoned and sensible post, Jonathan. I think we are beginning to see more sense in the field of taxonomy. We can easily have our cake and eat it, too: generic names can remain attached to well-known monophyletic groups, and new subgeneric names can be added as we discover or decide to recognize groups within large genera. Names should not change their meaning unless we discover that they do not actually apply to a monophyletic group. Now THAT’S a principle of modern systematics that everyone should rally around.

  3. Maybe if people would agree to adopting the principles underlying the Phylocode, much of the cognitive dissonance could be dissipated?

  4. I agree with Jonathan and David. It seems to me that splitting is financially driven: in the pet trade “Ctenonotus” sell much better than “Anolis.” And publishing peer-reviewed papers gets you academic advancement and more pay: nothing is easier to do than split!
    Skip

  5. I completely agree with Jonathan’s post and want to add a few points. Two things are being completely forgotten/lost by proponents of splitting known monophyletic groups: the first, as noted, is stability of the nomenclature. This is one of the most important principles of taxonomy and is one of the problems that the phylocode is intended to address. The second point is that splitters have completely failed to about the history/philosophy of the generic ‘rank’. While it can (and has) been argued that ‘species’ are real entities, higher-level taxonomic categories are abstractions. This has always been understood. I wish I had some citations for this, but on short notice, I don’t. But it as always been my understanding that generic classifications were based solely on morphology, i.e., it doesn’t matter how many species (or subclades) there are—if they share a similar form they are the same genus. New genera are used to identify morphological gaps within the clade. As David emphasizes, we’re assuming that all named groups are monophyletic. Granted that ‘morphologically distinct’ is a subjective notion, but what isn’t? A taxonomy that uses the ‘morphological gap’ distinction of generic names, then, actually provides information about phenotype in the taxonomy, which strikes me as much more useful than additional information about phylogenetic substructure. One would have to be more-or-less an idiot in this day and age to assume that a speciose genus like Anolis has no phylogenetic structure that one must be aware of. A final point I’ll note is that as a non-systematist, it is EXTREMELY annoying to that every molecule jock who spits out yet another molecular phylogeny thinks that that is their right to revise an entire taxonomy! This drives me nuts! Molecular phylogenies change with the wind—taxonomies should not! Taxonomies should only change after years of consistency and consensus. The complete re-writing of all of squamate historical biology based on the newest trees of Wiens and his colleagues is a good example of this arrogance. There are already indications that their phylogeny won’t stand up to closer scrutiny. And certainly the morphological phylogeny of Gauthier et al. should give everyone pause. There is NO consensus here. Disagreement is fine. Working toward consensus is good. Creating new taxonomies and reinterpreting biological evolution in light of limited-evidence, non-consensus phylogenies is arrogant, disruptive and damaging to the field. Until someone demonstrates that Anolis is polyphyletic, or argues persuasively that a there are phenotypic gaps worth noting in the taxonomy, THERE IS NO CREDIBLE OR USEFUL REASON TO BREAK THE GENUS UP!

    1. Kurt, you seem to be stuck on the precedence of morphology, but as you know, genes often show us fragmentation and speciation events that were previously unrecognized. In addition, anoles are the model example of convergence, so unless you are a proponent of non-phylogenetic classifications, using morphology alone for anole classifications won’t work.

      By the way, we never indicated anoles lacked phylogenetic structure, by definition, we are seeking to specifically recognize it. That was Jonathan’s interpretation (that “we would have [readers] believe there is no structure”)…The question was one of how MUCH of it to recognize, which we point out will require much more data.

      As for your “molecule jock” reference, I used to laugh at a sign in a lab I once worked in that said “You find ‘em, we grind ‘em”…but you seem to think that folks using genes don’t know morphology. That’s decidedly not the case, and its interesting that people point fingers like that, but you never hear it the other way around (i.e., that stupid morphologist guy doesn’t know ANYTHING about genetics – what an idiot!), when probably the “molecular jock” could make that argument…

      1. Kirsten, my comment was in response to the claim made in both of your papers that the failure to break Anolis into multiple genera has hindered research. The implication of that is that people are unaware of the phylogenetic structure within anoles, which is what I meant when I wrote that your papers “would have [readers] believe there is no structure.” If, in fact, everyone is aware of that phylogenetic structure and is using that knowledge in their research (which is, in fact, the case), then the argument for taxonomic splitting collapses.

      2. Kirsten, you misinterpret what I said. I do not believe in the precedence of morphological data over molecular data in generating phylogenetic hypotheses—not at all. Nor do I believe, however, in the precedence of molecular data over phenotypic data, which is arguably the far more popular view these days. When is the last time you heard someone told that their molecular data should be interpreted in light of a morphological phylogeny? I believe that there are times when molecular data consistently retrieve an erroneous result and that this tells us something about how DNA evolves. In the last 5 years or so, for example, more and more studies are demonstrating adaptive convergence in gene sequences in completely unrelated taxa (e.g., echo-locating bats and cetaceans). This is something molecular systematists told us was a big advantage of molecular data over morphological data—the absence of adaptive convergence. Our understanding of these things, at all hierarchical levels, is always changing—hence my distaste for jumping to hasty conclusions and imposing premature interpretations (and taxonomies) on the community. I would never argue, for instance, that morphological data are superior for constructing species-level trees within a genus or family, but for really deep nodes of the sort we are talking about in squamates, first principles argue that morphology should preserve signal long after DNA sequence evolution is saturated. The two types of data are potentially informative at different parts of the tree and are therefore seen as ‘reciprocally illuminating’, as Hennig said.

        So saying, the Anolis issue is very different. This is a question of taxonomy, not phylogeny. I did not suggest, or at least did not mean to, that your findings suggest a LACK of phylogenetic structure—only that it is not necessary to recognize that structure with the taxonomy since it is not accompanied by corresponding phenotypic patterns. My argument is that the historical (and I argue, sensible) use of the ‘genus’ is to recognize phenotypic gaps, not phylogenetic gaps. This is another way that phylogenetic classifications/taxonomies can include information—in this case about phenotypic evolution rather than phylogenetic structure. In my opinion, as someone else pointed out, it is much more useful to know that, despite lots of branching going on, phenotypes tend to remain conservative—that is the ‘purpose’ of generic-level classification. As long as only monophyletic groups are recognized, it becomes a subjective decision about how many branches to give formal names. The mere presence of phylogenetic structure does not, in my mind, justify taxonomic fragmentation without correlated phenotypic divergence. Where would naming clades end if this is the principal justification for supra-specific taxon names? Given the tiny advantage of referencing phylogenetic structure in the phylogeny vs. the huge downside of destabilizing an otherwise very stabile (and phylogenetically faithful) taxonomy, I think it’s better to err on the side of conservatism.

        I apologize for the “molecule jock” reference—it was out of line. I get very exercised about this topic! Some of my best friends are molecular systematists… I agree that many morphologists and even morphological systematists are woefully uninformed about molecular methods and analytics and that in the modern era one needs to master this to some extent, at least, to have an informed opinion. I tend to think, however, that the balance tends to go the other way—there are large numbers of people being trained in molecular systematics whose closest approach to an organism is to access Genbank. The good systematists I know also know the phenotype and recognize when a molecular result is, at least, questionable. I’ve always found it interesting that systematists and morphologists approach morphology in very different ways, but that is a whole ‘nother story…

        I conclude my typically long-winded response by agreeing with cybokat that what we need is MORE information, not less. I think that morphology is going to be critical to generate better hypotheses about squamate evolution. Anolis may be a much more difficult problem owing precisely to its relatively slow rates of phenotypic evolution. As a histologist, however, I can say that there is more variation there than is generally recognized, and that microscopic anatomy may be a useful path to take for anole-heads (something I started doing with Ernest Williams many, many years ago, but never finished).

        1. “This is something molecular systematists told us was a big advantage of molecular data over morphological data—the absence of adaptive convergence.”

          That’s why people use non-coding regions. :-) It’s certainly possible for phylogenies inferred from genetic data to be misled by convergent evolution, but it is far less likely and a lot easier to avoid. Anyone who told you that genetic data would have a -complete absence- of adaptive convergence is full of it, but if you’re dealing with data from multiple non-coding, unlinked loci the likelihood of this being a real problem rapidly drops off into insignificance.

          But, in any case, let’s suppose we just weight phenotypic characters and genotypic characters equally. The signal from the genetic data will almost always win (and the “almost” is probably an unnecessary hedge). Why? There’s more of it available to us. In our age of increasingly large and ever cheaper genetic datasets, in order for morphology to retain any real relevance you have to argue that phenotypic characters should be given dramatically greater weight than genotypic characters.

          Otherwise, it is as simple as this: More data wins.

          1. I don’t see many noncoding markers except 16SrRNA that are used in phylogenetic reconstructions, and fully expect that future discoveries regarding the neutrality of the commonly used, coding markers will make cause systematists having sleepless nights….

          2. cybokat –

            I haven’t kept up with the availability of non-coding loci for herp phylogenetics (my research has been on plants, with herps only a hobby). This is certainly one of the issues with commonly-used mtDNA markers–rapid mutation rate in a coding region means a near-certainty of massive homoplasy. At this point, even if you had to start from scratch (and surely herps aren’t still -that- stagnant a backwater!), finding variable non-coding regions and developing primers is a lot easier than finding an equivalent source of morphological data–so far as I know, there simply isn’t such a thing, for any taxon! It’s nearly impossible to find 100s of morphological characters to start with, never mind worrying about selection.

            Even non-coding loci (and, similarly, third positions in coding loci) probably don’t evolve in a truly neutral fashion. This is an old debate, going back to the 70’s. My understanding is that if you want strict neutrality, you have a problem; if you want “neutral enough that the odds of adaptive convergence are negligible”, it’s a non-issue.

  6. What I am really hoping for in this debate, is more information instead of more argument.

    A genus should be a group that has important evolutionary differences to other groups, which is a pretty subjective definition. “Importance” should be determined on several levels beyond only scalation and some 1500 base pairs of DNA.

    For me the issue at hand is this: Interestingly, Anolis DNA clades do not seem to reflect any “important” differences in morphology (beyond scalation, dentition and some metric characters), or ecology that would be necessary for everyone to immediately agree to the newly proposed genera.
    However one might wish to name them, these clades do seem to agree with the 20kb phylogeny of Alfoldi et al., and with island biogeography, which Alfoldi et al. used to strenghten the argument for convergent evolution. If these island radiation clades furthermore correspond to karyotypic differences (i have not been following these new studies as closely), for me the really interesting question at hand is why, and how morphological and ecological evolution in Anolis exhibits such exceptional stasis over time that island clades have not been defined as multiple genera with more support from the community.
    They seem to have convergently radiated within a relatively constrained portion of eco/morphospace, but not beyond that (if they had, we would maybe have different genera, but no convergence to study). So yeah… why?

  7. I completely agree with Jonathan’s post as well. We in the amphibian world have been dealing with taxonomic revisions that have had a major inhibitory effect on communication. In my view these changes were completely unnecessary and have done nothing to improve our understanding of relationships. It is completely mind blowing to me that there are people still clinging to the notion that our Victorian era hierarchical taxonomy should be revised continually in ways that repeatedly disenfranchise entire audiences. In any sort of system of organization the first rule is to never change the name of the critical subjects. As new information emerges definitely change the position and relationships. But not the name. Read the last chapter of ‘On the Origin of Species.’ Darwin assumed we would be long past this stuff by now. In an era of democratized information, systematists risk losing control of naming or will be ignored if we don’t move on to Phylocode or some other system that focuses on lineages and detaches name from address.

    1. PhyloCode doesn’t have a mechanism for naming species. Hence it does not “detach name from address”; if you don’t have names for species, you can hardly detach them from anything.

      1. Exactly. Phylocode doesn’t change the names people already know so that everything that thousands or millions of people have learned disappears overnight. That is detaching name from address.

        1. Species names remain governed by the ICZN / ICNafp. That means they can still change in all the ways they can change now.

          You are, strictly speaking, correct that PhyloCode doesn’t change species names. That is only because it doesn’t deal with them one way or the other and is hardly an argument in PhyloCode’s favor. It’s like saying I should use ICNafp as a system for the naming lizards, because, since ICNafp doesn’t govern the names of lizards at all, it won’t change any taxon names! I suppose that is a kind of detachment, but… not in a way that is at all helpful.

          Gist being, “we don’t deal with taxon names” ≠ “we provide a good system for dealing with taxon names”.

  8. As a museum curator I would like to add that binomial name changes are a daily problem when it comes to searching databases and locating specimens. In most cases, we need to refer to a species’ synonomy and individually search our database for each possible name. As Jonathan notes, we will one day have the computational infrastructure to avoid this problem when synonyms are part of the databases, but that day is not yet here. I have heard the response that changes to binomials are inevitable and that they are just something we need to deal with. While this may be true, I don’t think that most of the people making this argument are responsible for facilitating access to major biodiversity collections, or are required to deal with the consequences of potentially needless binomial revision on a daily basis. Taxonomic revision is a good thing, but it seems only reasonable to me that we should take simple measures that might help avoid major disruptions in our ability to conduct specimen-based biodiversity research. The suggestion that we use subgeneric designations, which has been made repeatedly here on Anole Annals (in a post back in October of 2012 by Dave Wake and more recently by Peter Uetz) would avoid disruption of our databases while providing anole researchers with formal names to refer to clades within Anolis.

    1. With some specimen databases, that day is here.

      As an example, try try this. The “Include Synonyms from Taxonomic Thesaurus” checkbox turns on precisely the feature you are describing. :-)

      1. Sounds cool. I tried to test it out, but I can’t get this portal to find any specimen records for Arizona native herps whose names have changed in recent years (e.g., Bufo).

        1. Sorry, I should have mentioned that this is a plant specimen database. No lizards. :-)

  9. “Nicholson et al. (2014) don’t even address the point raised by Poe and many others, that such name changes are more than disruptive, but truly damaging to scholarly research.”

    One wonders why public opinion of science is at record lows when the brightest minds continue to argue naming conventions for relationships recognized since the 1990s on a blog and a limited-audience taxonomic journal.

  10. I agree with Dave Wake, Peter Uetz, Rich Glor, and David Cannatella that subgenera are the way to go here. Why not? The argument that the use of subgenera hasn’t “caught on” is weak. (Look at Eleutherodactylus, Bolitoglossa, Batrachoseps, etc.). As Rich said, recognizing subgenera provides all of the benefits and none of the disadvantages of breaking up Anolis.

    1. See also “Misconception 5″ from David Cannatella’s June 13 post “The Battle Over Anole Classification Ignores the War.”

  11. At this stage of the game this is probably a minor comment, but one worth adding. While reading some recent posts about data deposition and the reproducibility of studies issues of nomenclature are already a major problem for repository managers and people on the the computational end of things (similar to the comment Rich made above regarding museum curation, but in a digital world). I read concerns about having different databases losing connectivity and, in turn, leading to an increased potential of losing stored data. As stated on one blog clearly stated, “The benefits of open data will only be realised if we preserve identity of data sets.” This is a problem these databases are already facing, not something that is merely a nebulous, unrealized concern. Considering the wealth of phenotypic and molecular data flowing into online databases it seems like a taxonomic upheaval on Anolis is unwise. So just to reiterate what Jonathan said, “Some day all scholarly resources will be digital and species names will be automatically hyperlinked to their previous identities, but we are a long way from that point. ”

    Considering this and the many other points raised above I don’t yet understand how this revision can be good for the community. The argument feels based on antiquated 20th century taxonomic philosophy rather than modern pragmatism. I don’t say this to simply be disrespectful to the authors, but I would like to hear some argument about how they see these real life issues being efficiently addressed. Furthermore, a lot of effort and discussion went into coming up with a method of gene nomenclature for the anole genome. Sixteen authors from contributed to the paper in BMC where we then asked for further feedback from the community on our recommendations. Considering the vast interest in Anolis taxonomy from a community that probably numbers in the 1000’s someone please tell me why the hell should this situation be any different? With the number of good ideas that have been proposed and well received by the community (sub-genera, clade names, etc.) why are we still working from this stance where there is a line drawn in the sand? Its time to move on with a more productive discussion.

  12. Some taxonomists think that the purpose of biological classification is solely to express evolutionary relationships. This philosophy can and does lead to classifications which are difficult or impossible to use by other biologists for other purposes. One reason that we classify is so we can make meaningful generalizations about groups. A biologist, for example, can make many generalizations about Reptile physiology (meaning Turtles, Crocs, and Squamates) but would be hard pressed to say as much about the physiology of living Archosaurs (Crocs + Birds).

    Thankfully, even the most hard core proponents of a cladistic approach to classification seem to have abandoned one of its most ridiculous tenets. If one really wants to express evolutionary relationships through classification, it forbids any taxon from having more than two subunits. For instance, it the relationship of three closely related species is determined (a tree constructed), putting them in a single genus does not fully describe their relationships; they must be put into two genera in order to group the two most closely related species. I have seen insect classifications based on cladistic analysis which use upwards of fifty diffferent levels of classification (e.g., Supertribes, Tribes, Infratribes, Metatribes, Subtribes, etc.). Of course these schemes while intrinsically valuable never caught on.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Optionally add an image (JPEG only)