Twitter Nerd-Fight Reveals a Long, Bizarre Scientific Feud

Yeah, but the point is that they want authors to defend their use of non-parsimony methods on philosophical grounds, rather than scientific grounds. Which, you know, is kinda odd. I mean, like, do they expect researchers to dig up philosophy papers arguing the merits of Occam's razor? It's just weird. Why don't they want authors to defend their use of, say, Bayesian methods on scientific grounds? Why did Cladistics turn into a philosophy journal?And they're adherence to parsimony comes across as kind of dogmatic. Anyone who's worked in molecular phylogenetics or bioinformatics knows that the most appropriate method depends on the data set. Plus, there are good lines of evidence that Bayesian and maximum-likelihood methods are generally superior to parsimony.And we know that parsimony is often an incorrect map of biological reality. According to parsimony based on molecular structure, the type III secretion system and bacterial flagella should be monophyletic. But molecular phylogenetic evidence strongly indicates that type III secretion systems nest within flagellar sequences, violating the notion of parsimony.So for a journal like Cladistics to ask for uniform acceptance of parsimony as the default phylogenetic method is dogmatic and a little creepy. Edited by Genomicus, : No reason given.

To be honest, my head is coming close to exploding. I am presently weirded out by the editorial board of Cladistics. Just a bunch of fervent cladists. Edited by Genomicus, : No reason given.

There are two approaches to defending the superiority of a given phylogenetic method: philosophical and scientific. As an example of how these are fundamentally different: one can debate the merits of Neo-Darwinian evolution vs. creationism on epistemological grounds, but that belongs in a philosophy journal, not a technical scientific journal.So the real question is this: why did Cladistics, a technical journal, decide that authors needed to articulate a philosophical defense of their phylogenetic approach instead of outlining the scientific rationale for their method of choice? Yeah, but:1. A policy can be grounded in dogmatism.2. Their preference for parsimony -- asking all phylogenetic research submitted to their journal to always use parsimony as a method of tree construction ("Phylogenetic data sets submitted to this journal should be analysed using parsimony") -- indicates a dogmatic, parochial bias towards a particular technical method. And it's dogmatic because they don't actually give very good reasons for this preference for parsimony. Science is, of course, based on philosophy. It's unfortunate that many scientists don't have a solid grounding in the philosophy of science. But Cladistics isn't a philosophy journal in the colloquial sense; it's a technical scientific journal. They're essentially asking scientists to become expert philosophers who can appropriately analyze the epistemic merits of Occam's razor vs. probabilistic approaches. Sure, because everything ultimately traces down to philosophy. But you can use this argument for just about everything: "If you reject creationism, you'll be doing so based on philosophical grounds.""If you think vaccines are safe, it's because of philosophical reasons.""Oh, you think the anthropogenic global warming has merit? Well, that's just your philosophical perspective."See how easy it becomes to make anything philosophical? So let me rephrase my original comment:There are good lines of scientific (that is, falsifiable) evidence that Bayesian and maximum-likelihood methods are pragmatically superior to parsimony.

You do realize that the statement "There are two approaches to defending the superiority of a given phylogenetic method: philosophical and scientific" is not incompatible with your statement that science is better at generating hard predictions? You're not really disagreeing with me with your above comment.

See Gadagkar and Kumar (2005) for an example of datasets where maximum likelihood is more appropriate than maximum parsimony. For phylogenetics as an overarching field? Yes, parsimony has long dominated phylogenetic constructions based on, e.g, morphological characters. But the application of parsimony to molecular sequences occurred at about the same time as maximum likelihood methods and distance matrices were use for constructions of molecular phylogenies (in the 60s and 70s). So parsimony as a method for molecular phylogenetics has never been a gold standard per se, with the caveat that sophisticated statistical methods took longer to develop than parsimony methods. Well, different phylogenetic methods are more appropriate depending on the data set. I hear you, Taq, but when it comes to molecular phylogenetics it's pretty hard to have an absolute standard given the diversity of sequences involved. So, for example, I've published molecular phylogenies of certain gram-positive protein systems, but opted to use more than one phylogenetic method (ML and Bayesian analysis) to see the extent to which the phylogenies were congruent.It'd be inappropriate, IMHO, to always adopt a given phylogenetic method as a standard -- particularly given that molecular phylogenetic analysis is based on either a method that is philosophically grounded or statistically grounded.ReferenceGadagkar, S., Kumar, S., 2005. Maximum Likelihood Outperforms Maximum Parsimony Even When Evolutionary Rates Are Heterotachous. Mol Bio Evo, 22(11): 2139-2141.

You can defend the scientific superiority of ML vs. parsimony through the use of, e.g., statistical evidence. As an example of this, see Gadagkar and Kumar (2005), referenced in a prior post.Now, you could respond that the choice to use statistical evidence reveals a prior philosophical commitment. And you would be right. However, statistical evidence based on simulations of sequence evolution is falsifiable (that is, scientific). The philosophical perspective is not falsifiable. So why would Cladistics, a technical journal, ask for authors to philosophically defend their choices instead of asking them to defend their method of choice through falsifiable evidence? Think about it.The actual reason Cladistics made that decision -- for authors to articulate a philosophical defense -- is because that journal has an ideological commitment to parsimony-based phylogenetics (as revealed by their history). Parsimony-based phylogenetics doesn't fare so well when it comes to falsifiable evidence; it does better when philosophy a la William of Ockham is invoked. What's the difference between a philosophical defense of Neo-Darwinian evolution vs. a scientific defense? I'm genuinely curious how you approach this demarcation. Are you familiar with the history of the cladistics vs. frequentists debate, and the role that the Willi Hennig Society played in this? An examination of this history will reveal that the Willi Hennig Society has long had a dogmatic commitment to parsimony over statistical approaches to phylogenetics. You have a rather dogmatic definition of dogmatism. I think "positiveness in assertion of opinion especially when unwarranted or arrogant" suffices in this case as an adequate definition of "dogmatism." Cladistics made no mention of why exactly they consider parsimony to be so much superior to other methods that it should be adopted as the default method by authors.

Yes, there has been some work in this area. Baele et al. (2006) used neighbor joining trees (not ML or Bayesian analysis) to detect heterotachy in nucleotide sequences. That being said, a lot more work needs to be done in this area. Now I'm curious, actually. I'm not that familiar with protein assaying. How do you know if you have a nearly pure sample if you still don't know the concentration of the protein? Hopefully that question makes some sense. Actually, by congruence I mean:1. Congruence among the phylogenetic trees generated through ML and Bayesian analysis. Specifically, I was testing for monophyly among distantly related protein systems, so if both methods pointed to monophyly (with strong bootstrap support) -- then these trees probably accurately describe the overall relation among the sequences. The internal branches were less congruent, with weak bootstrap support.2. Congruence to the accepted species tree (which is hard in the case of distantly related prokaryotes, since there is disagreement on the exact phylogeny for different prokaryotic phyla). I tested for horizontal gene transfer and/or gene duplication + loss using the Notung program, and indeed there were instances of horizontal gene transfer.ReferencesTired and my eyes are droopy, so here's the link instead of a journal-esque citation: Improved Statistical Method for Detecting Heterotachy in Nucleotide Sequences | Molecular Biology and Evolution | Oxford Academic