Luskin makes more mistakes on the Cambrian and Cladistics

By Nick Matzke

(edited to add a point on Aegirocassis and Parapeytoia)

This week, the Discovery Institute Press put out another book called Debating Darwin’s Doubt. I took one for the team and bought it, in part because a a decent chunk of the book is responding to me. I’m pretty sure I’ve never been mentioned so much in a book!

Sadly, though, looking through it, almost all of it is material re-hashed from the DI “Evolution News and Views” blog and is no better than it was the first time. There is, however, a new chapter (I think it is new) by Casey Luskin, chapter 9, “Cladistics to the Rescue?” responding to me. If you don’t want to buy the book, there is a free podcast at ID the Future (heh), “Debating Darwin’s Doubt: Casey Luskin on Classification of Organisms” that interviews Luskin (although I think he wrote the questions). It has mostly the same material.

Unfortunately, I do not have time at the moment to write the introductory-level-tutorial-from-square-one that would be required to really explain the basics of cladistics and phylogenetics to Luskin et al. I have literally just moved to Australia to start as a Discovery Early Career Researcher Award (DECRA) Fellow in the Division of Ecology, Evolution, and Genetics, Research School of Biology, at The Australian National University in Canberra. Once I have a bed and a computer in my office I may be in better shape to do things more thoroughly – I have a bit of a fantasy about writing an R vignette or R package called something like BasicPhylogeneticsForCreationistsEspeciallyLuskin (I’ll take suggestions on a better name/acronym).

However, below, I can briefly hit the high points on the small bit of Luskin’s chapter that was new.

Except for chapter 9, everything else I’ve seen pertaining to me in Debating Darwin’s Doubt has already been addressed in previous posts:

Background – previous posts on the Cambrian/Meyer

Alan Gishlick, Nick Matzke, and Wesley R. Elsberry (2004). “Meyer’s Hopeless Monster.” Panda’s Thumb post, August 24, 2004. http://pandasthumb.org/archives/2004/08/meyers-hopeless-1.html

The “Meyer 2004” Medley - The Panda’s Thumb – the complete history of the Meyer 2004 craziness.

Matzke, Nicholas (2005). Down with phyla! - The Panda’s Thumb, which reviewed:

Matzke, Nicholas (2005). Down with phyla! (episode II) - The Panda’s Thumb

Matzke, Nicholas (2007). Meet Orthrozanclus (down with phyla!) - The Panda’s Thumb

Matzke, Nicholas (2013). “Meyer’s Hopeless Monster, Part II.” The Panda’s Thumb.

Matzke, Nicholas (2013). “Luskin’s Hopeless Monster.” The Panda’s Thumb.

Matzke, Nicholas (2013). “Meyer on Medved: the blind leading the blind.” The Panda’s Thumb.

Matzke, Nicholas (2013). “A Very Darwinian Halloween.” The Panda’s Thumb.

Matzke, Nicholas (2014). “Meyer’s Hopeless Monster, Part III.” The Panda’s Thumb.

Here is my main comment responding to Luskin’s chapter, “Cladistics to the Rescue?” This is modified from a comment on Larry Moran’s Sandwalk post.

Key flaws in Luskin’s chapter 9, “Cladistics to the rescue?”

(1) Lobopods aren’t a natural group. Luskin continues to think of “lobopods” as a coherent group, which results in pointless arguments about e.g. whether lobopods are “closer” to arthropods than anomalocarids. Earth to Luskin: lobopods are a paraphyletic grab-bag. Living arthropods, onychophorans, and tardigrades all descend from lobopods, as do anomalocarids. Phylogenetically speaking, then, all of these groups are *within* the lobopod group.

When Luskin quotes certain authorities (mostly the Erwin et al. Science paper) calling lobopods a “phylum” (if memory serves, Erwin et al. actually create two phyla out of lobopods, phyla which no one else recognizes IIRC, further illustrating the fundamental conceptual problems with applying ranked Linnaean taxonomy to fossils), he is ignoring what I explained previously about old-fashioned Linnaean taxonomy and the confusions it causes. Some (typically older) authors accept paraphyletic phyla. But it has long been clear that purely phylogenetic classification is taking over. It’s already dominant in most areas of biology/paleontology, and it’s just taken a bit longer with fossil invertebrates, although that is now clearly the leading edge in the field, even in the Cambrian Explosion research (see the work of Graham Budd, David Legg, etc.) You can’t quote authors talking about “phyla” from different taxonomic schools of thought, writing in different decades, etc. without understanding the differences in what they are referring to.

Sometimes it appears that Luskin actually means “onychophorans” or “crown + stem onychophorans” when he says “lobopods”. Some of what Luskin says would seem less crazy on that hypothesis. More on taxonomy below.

(2) Panarthropods and the importance of stem/crown terminology. In a similar vein, the new Luskin chapter contains a fair bit of discussion about how different the anomalocarids are from true arthropods, contradicting Meyer/Luskin’s previous arguments (repeated in this same book, in other chapters!) about how it was totally OK to lump anomalocarids in as just another thing in the arthropod group! (Another argument for phyla-schmyla! I thought phyla were supposed to be super-duper-distinct!)

Throughout the DI commentary (mostly Meyer/Luskin) discussing arthropods, we see confusion over “arthropods”, often evidenced in how Luskin deploys quotes. Sometimes by “arthropods” Luskin means panarthropods, sometimes he means crown-group arthropods, sometimes he means critters sharing “enough” arthropod traits. It is true that all of these usages can be found in the literature over the decades, and it’s true that it can be confusing, but THIS IS PRECISELY WHY RIGOROUS PHYLOGENETIC CLASSIFICATION, AND THE STEM/CROWN DISTINCTION, WAS INVENTED IN THE FIRST PLACE. It’s meaningless to quote-mine a bunch of quotes with people using slightly different terminology about, say, the position of Anomalocaris, and to pretend they all mean hugely different things. Typically they are just different ways of saying “on the arthropod stem”.

I’ve taken one position from the beginning, which is that anomalocarids are below the arthropod crown group (exactly how far below can be debated, but these are details), and this means they have transitional morphology. Everything else on the stem (dozens of fossil taxa, at the very least) also has transitional morphology. The transitional morphology is what places them on the stems, instead of inside crown arthropods or crown onychophorans. Thus there are many fossils with transitional morphology between crown-group phyla, and many of these fossils are in the early Cambrian. No, they don’t all have to appear in exact chronological order, like a children’s-cartoon version of evolution, because fossil sampling is a stochastic process, like taking a phylogenetic tree and sampling it by throwing darts at it, or (for lagerstatten) taking a number of samples at one time-point. There will be an overall correlation between phylogeny and fossil dates, which has been demonstrated in publications in many cases, but it won’t be exact. The fact that stem groups are so prevalent compared to crown groups in the early Cambrian is evidence of this time-phylogeny correlation in the Cambrian taxa specifically. But I’ve said all of this before, not sure why I am saying it again. In Darwin’s Doubt, Meyer missed this crucial transitional fossil data in epic fashion, and all of the subsequent discussion of this point by Meyer, Luskin et al. has been an attempt to avoid the key point: many transitional fossils are known from the early Cambrian.

(3) Anomalocarids and legs. Luskin makes much use of an argument along the lines of “lobopods have arthropod-like legs but no complex head; anomalocarids have an arthropod-like head but no legs (they have swimming flaps); this means the data conflict with the tree and therefore the whole thing is bunk and special creation is a better alternative” (I am paraphrasing, obviously. But that’s his argument.) Actually, if that were the data, only one extra change would necessarily need to be postulated, namely loss of legs in the anomalocarids, and adding one character change step to a cladistic reconstruction does no great violence to the data. IDists/creationists, being almost always hopeless amateurs who can’t be bothered to get to the library and really learn a topic, basically always think about the evolution of just a few characters, and judge scenarios on that basis. But in real life, cladistic analyses are typically done on hundreds of characters, and cladistic reconstructions will have hundreds or thousands of character steps in the most parsimonious tree. Having a step for leg loss is perfectly justified, if other characters support the tree topology in that region.

As if that weren’t enough, we have lots of evidence from all kinds of sources that events like limb loss happen occasionally, undoubtedly much more commonly than limb gain.

But – this whole discussion is pointless, because the anomalocarid or near-anomalocarid Parapeytoia had friggin’ legs! And it’s early Cambrian (530 Ma)! Hello transitional form!

And, on top of that, earlier this year, to international acclaim, Aegirocassis was published. This guy is Ordovician (480 Ma), and is an anomalocarid, but the specimen is huge (2 meters), and some combination of the preservation and the size allowed the authors to notice that the flaps can actually be separated into dorsal and ventral flaps, and resolve other features of flap anatomy.

The title and the abstract of the Nature paper tell the story:

Anomalocaridid trunk limb homology revealed by a giant filter-feeder with paired flaps

Exceptionally preserved fossils from the Palaeozoic era provide crucial insights into arthropod evolution, with recent discoveries bringing phylogeny and character homology into sharp focus. Integral to such studies are anomalocaridids, a clade of stem arthropods whose remarkable morphology illuminates early arthropod relationships and Cambrian ecology. Although recent work has focused on the anomalocaridid head, the nature of their trunk has been debated widely. Here we describe new anomalocaridid specimens from the Early Ordovician Fezouata Biota of Morocco, which not only show well-preserved head appendages providing key ecological data, but also elucidate the nature of anomalocaridid trunk flaps, resolving their homology with arthropod trunk limbs. The new material shows that each trunk segment bears a separate dorsal and ventral pair of flaps, with a series of setal blades attached at the base of the dorsal flaps. Comparisons with other stem lineage arthropods indicate that anomalocaridid ventral flaps are homologous with lobopodous walking limbs and the endopod of the euarthropod biramous limb, whereas the dorsal flaps and associated setal blades are homologous with the flaps of gilled lobopodians (for example, Kerygmachela kierkegaardi, Pambdelurion whittingtoni) and exites of the ‘Cambrian biramous limb’. This evidence shows that anomalocaridids represent a stage before the fusion of exite and endopod into the ‘Cambrian biramous limb’, confirming their basal placement in the euarthropod stem, rather than in the arthropod crown or with cycloneuralian worms.

For more commentary, see Edgecomb in Current Biology last month: “In a Flap About Flaps.” He appears to basically agree with the Nature authors, although he adds some more considerations about another specimen that may further illustrate the transition.

(4) The Consistency Index (CI) and null distributions Luskin has finally discovered the concept of a null distribution for the Consistency Index (CI)! It’s only taken him about 2 years! Now, finally, having learned about it, he can dimly see the problem with his/Meyer’s old tactic of squinting at some published CI value and declaring it “high” or “low” without any consideration of what the null distribution is. So, his new argument is that the null hypothesis of random distribution of characters is silly. To that I say – why? That is precisely what one is claiming if one claims the data have no cladistic tree structure, which is precisely what these turkeys have been telling their readers for years now. (Except Berlinski; he admitted at one point that there is tree structure in the data, which is not made up.)

Luskin then raises the idea that intelligent design could correlate some characters, and this could cause above-null CIs. This is true enough, but such structure in the data, when the designers are humans, is very limited – all of this was thoroughly discussed years ago by Doug Theobald in his discussion of natural versus artificial hierarchies, in his 29+ Evidences for Common Ancestry FAQ, a resource which Luskin, Meyer et al. still lack the courage to engage with in any detail: http://www.talkorigins.org/faqs/comdesc/section1.html#nested_hierarchy

If Luskin specified a quantifiable model for ID that specified what parameters are to be learned from the data, and generated distributions of data (or CI or other statistics) from the model, then he’d have some shot at progressing in an anti-frequentist direction. But good luck with that – IDists rarely say anything specific enough about their designer to be subject to empirical test.

As for going beyond frequentist null-hypothesis rejection, us phylogeneticists got there years ago. Maximum likelihood and Bayesian methods are now dominant in the field (although only just starting in fossil invertebrate studies like the Cambrian; almost the last frontier for this area). For the sake of simplicity, I focused on parsimony/cladistic methods in my critiques of Darwin’s Doubt, and as a result of that, plus the IDists’ systematic naivety and amateurism, cladistics is almost all that gets talked about in the IDists’ replies. But if they would like tests of common ancestry in a fully likelihoodist or Bayesian framework, where null hypotheses do not have to be assumed at all, we’ve got that covered. Doug Theobald did that already (also), in his 2010 Nature paper testing common ancestry. We could do it for Cambrian morphology data matrices too, although it would take a couple of weeks of full-time work and thus a grant or a graduate student. Of course, the IDists just summarily rejected that work as well, so I’m not sure what the point would be.

Other loose points:

(5) Character states evolving twice. Luskin seems to think that any character state that evolves twice constitutes a contradiction of evolutionary expectations. So, under his view, under evolution, all character states of any sort, whether complex (wings, eyes) or simple (a bump on a bone, a bend in an exoskeleton plate) have to evolve once and only once, or else common ancestry is wrong and we should just conclude special creation instead. He seems to think that, under evolution, if a character can change states once, there must be some magical force that prevents a similar event from happening elsewhere somewhere else in the tree.

(6) Estimating phylogenies from characters that change more than once. And scotch. Luskin also seems to think that, if characters change states more than once, then the whole enterprise of estimating phylogenetic trees is hopeless and subjective. He ignores a point I made before, which is totally obvious when you think about it, that on any tree that is not tiny, characters that change twice on a tree will still have plenty of phylogenetic signal. In fact, we could easily simulate characters on a tree, give them an evolutionary rate high enough so that the expected amount of change is 2 changes per character on the tree, and then see if we can infer the tree given only the character states at the tips as data. It’s not even that interesting to run the experiment, because I know what the result would be: this would be an easy phylogenetic inference problem, given a reasonable number of taxa and reasonable number of characters. I’d even wager a bottle of single-malt scotch on it.

The real puzzle is why Luskin thinks that anyone who actually knows phylogenetics, and knows facts like the above, will take him seriously.

Various things I wish the IDists would get clear in their heads:

(7) Characters versus character states. They don’t get that characters can be homologous, even while character states can evolve convergently.

(8) Changing dating of the Cambrian. The dating of the beginning of the Cambrian, and key (although now outdated) subsets like the Tommotian, has been some of the least certain dating in the Phaenerozoic timescale. However, the IDists love to cite dating estimates from the mid-1990s that put the “appearance of ‘phyla’” (this whole phrase relies on ignoring the stem/crown distinction) in a particularly narrow time window. They also LOVE to conflate this period with the beginning of the Cambrian, and pretend that the history is basically: unicells, Ediacarans, boom-modern-phyla. This depiction of history is the dominant picture presented in Darwin’s Doubt. But, first, the estimates have broadened in the last two decades; second, relevant stuff was going on before the “appearance of ‘phyla’”, namely the diversification of the small shellies, still completely inadequately dealt with in Debating Darwin’s Doubt (there is no new material on them in the book); third, the small shellies go back into the late Precambrian, and do not themselves constitute the beginning of the Cambrian; and fourth, if one is phylogenetically rigorous about the crown/stem distinction, many “phyla” originate well after the “Explosion” – what you have in the Cambrian Explosion is mostly stem groups to the classic phyla, or groups on the stems of classes, etc.

(9) Cladistics isn’t the beginning and ending of phylogenetics. It’s more the beginning.Many of the limitations of “classic” cladistics (no direct ancestors, no consideration of time, equal weighting of characters etc.) don’t apply to statistical phylogenetics.

(10) And, pattern cladistics isn’t cladistics, it’s an almost-extinct subset of cladistics that was probably always in the minority. Quoting a stale bit of pattern cladistic dogma, whether from an alleged authority or something recycled in a textbook, does not prove anything except one’s quote-mining ability. In contrast to pattern cladistics, most modern researchers in phylogenetics think that cladograms and phylograms (and in the best case, chronograms) are an estimate of evolutionary history, not just a description of pattern and not just a method of classification. Most researchers similarly accept that having an estimate of the phylogenetic tree, which automatically includes an estimate of character evolution histories, also tells us many important things about the *processes* involved – speciation and extinction rates, correlations between characters and these rates, rates of change in character evolution and how those rates change through time (for example, major radiations into empty ecological niches, versus evolution in stable “filled” ecosystems), etc.

(11) It’s true that cladistics, or better, phylogenetics, doesn’t answer all questions of interest about the Cambrian Explosion, or anything else. BUT, FOR THE LOVE OF THE DESIGNER, THAT IS THE CASE FOR EVERY INVESTIGATIVE METHOD IN SCIENCE. Carbon dating tells you the age of organic material that is less than 50,000 years old, not the age of the Earth. Stratigraphy tells you relative age, not absolute age. Light microscopy can tell you what chromosomes are doing, but not the DNA sequence. Sequencing a genome tells you what the DNA sequence is, but doesn’t tell you what is functional, nor how your functional assessment would change if you looked at the highly-different genome sizes in related organisms (take note, ENCODE). Cladistics and phylogenetics, as I have said, give the big picture: in what order did character states change? How did the collection of character states that we now take to mean “phylum” assemble, character-change-by-character change? Other disciplines (evo-devo, molecular biology, population genetics) can then examine how individual characters change. When Luskin et al. whine that “cladistics doesn’t explain the origin of information”, the major response is: “No, you moron, molecular biology and population genetics explain the origin of new genetic information, especially gene duplication and modification of duplicates by mutation, drift, and selection.”

(Behe and Berlinski have already basically admitted that evolutionary biologists have reasonably and successfully explained the origin of at least some new genes with these processes, fatally sinking much of the Luskin/Meyer’s core arguments on this topic. This conflict is unaddressed in Debating Darwin’s Doubt.)