Taxacom: On describing new taxa without using phylogenetics: some suggestions

Thu Oct 9 01:36:10 CDT 2025

Dear Kirk,

a comment on total evidence, philosophically and logically.

1. For a given set of taxa, there is only one true tree.
2. We cannot know the true tree, so any reconstructed tree is an approximation with a certain level of confidence.
3. Any subset of the data may result in a different tree.
4. Some subsets of the data results in trees closer to the true tree than others.
5. We cannot know the best subset, i.e. the one that leads to the tree closest to the true tree, but there is one.
6. If we combine two more more subsets, some better, some worse, then the resulting tree will be worse than the one based on the best subsets only.

Ergo: Total evidence (in the meaning of combining different types of characters) is not the best approach to approximate the true tree. Can be shown using simulations. I use this exercise routinely in my classes on integrative taxonomy. Of course I work with fungi, which are a nightmare when it comes to phenotypic classifications. People pay homoplasies in mushroom morphologies with their lives, on an almost daily basis.

Total evidence is a serious issue in phylogenomics. Many tend to think that just throwing hundreds of markers or entire genomes at a problem will solve it ultimately. In reality, different markers in such a huge dataset produce conflicting topologies, but the combinatorics of bootstrap support calculations gives even the shortest branches in such combined trees high support. Phylogenomic trees not rarely conflict with classical multimarker trees and this does not mean phylogenomic trees are necessarily closer to the truth just because they use "total genomic evidence".

Regards
Robert

Von: Kirk Fitzhugh <kfitzhugh at nhm.org>
Gesendet: Mittwoch, 8. Oktober 2025 22:05
An: Marko Mutanen <Marko.Mutanen at oulu.fi>
Cc: Lücking, Robert <R.Luecking at bo.berlin>; John Grehan <calabar.john at gmail.com>; Kuoi Zhang <zhang.guo-yi at outlook.com>; taxacom at lists.ku.edu
Betreff: Re: Taxacom: On describing new taxa without using phylogenetics: some suggestions

Marko,

What I said applies both philosophically and practically. To say, "morphology, or certain other types of data, are so noisy [sic] that they mislead more than they inform," implies that one can readily discern when homologous characters are indeed not homologous, in which case you'd need to rename those as different characters (see Fitzhugh 2022: https://nam10.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdoi.org%2F10.1007%2Fs10441-021-09412-4&data=05%7C02%7Ctaxacom%40lists.ku.edu%7Cbff683ea5c9f4e48599308de06fe27f6%7C3c176536afe643f5b96636feabbe3c1a%7C0%7C0%7C638955885851985025%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=37gLlhzOcknPSDoAxUWWFN%2BQ3dN84VjGr6k%2ByGdukWs%3D&reserved=0) before proceeding with inferring explanatory hypotheses. In the absence of such knowledge, eliminating explanations of homologous characters without any empirical evidence that they are not homologous makes no sense.

The view that "morphological and DNA data often produce highly contradictory phylogenies" is a popular point of view but one that is epistemically specious (see Fitzhugh 2014: https://nam10.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdoi.org%2F10.24199%2Fj.mmv.2014.71.07&data=05%7C02%7Ctaxacom%40lists.ku.edu%7Cbff683ea5c9f4e48599308de06fe27f6%7C3c176536afe643f5b96636feabbe3c1a%7C0%7C0%7C638955885852028024%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=bZLz9cWASYCTltekCvw5WFSuHAmHg%2FcUoC6QrBNjrdQ%3D&reserved=0). Comparing tree topologies has no meaning since the trees imply sets of hypotheses inferred from entirely different sets of premises. Once again, a violation of the requirement of total evidence. Saying "DNA provides far more information and tends to be less biased" carries no weight. Amount of information is irrelevant to what we are attempting to achieve, which is causal understanding. Characters by themselves aren't biased.

"It should also be remembered that all taxonomically relevant morphological information is ultimately encoded in DNA, so combining both raises additional philosophical challenges." That is true, which raises the spectre of downward causation. Something that has yet to be given serious consideration by those claiming sequence data have some priority over morphology (see Fitzhugh 2016: https://nam10.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdoi.org%2F10.1007%2Fs10441-016-9277-0&data=05%7C02%7Ctaxacom%40lists.ku.edu%7Cbff683ea5c9f4e48599308de06fe27f6%7C3c176536afe643f5b96636feabbe3c1a%7C0%7C0%7C638955885852044623%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=7sVAOegvLEX1uCuSy0HtSaWMToR7g%2BsM89euO1BJim4%3D&reserved=0).

Kirk

On Wed, Oct 8, 2025 at 12:47 PM Marko Mutanen <Marko.Mutanen at oulu.fi<mailto:Marko.Mutanen at oulu.fi>> wrote:
Dear Kirk, Thanks for your comment. I agree that this is true at a philosophical level, but it does not necessarily hold true at a practical level. It is quite possible that morphology, or certain other types of data, are so noisy that they
Dear Kirk,

Thanks for your comment.

I agree that this is true at a philosophical level, but it does not necessarily hold true at a practical level. It is quite possible that morphology, or certain other types of data, are so noisy that they mislead more than they inform. In my experience, morphological and DNA data often produce highly contradictory phylogenies. A consensus between the two is not necessarily more reliable than one based solely on large amounts of DNA data.

I recognize the philosophical weaknesses of this reasoning, but as Robert noted, DNA provides far more information and tends to be less biased. It should also be remembered that all taxonomically relevant morphological information is ultimately encoded in DNA, so combining both raises additional philosophical challenges. Homoplasy in morphology often results from natural selection, but this convergence rarely translates into DNA. Although DNA also contains homoplasy, it tends to be random, whereas morphological homoplasy is often systematic and directional.

Sincerely,

Marko M.

From: Kirk Fitzhugh <kfitzhugh at nhm.org<mailto:kfitzhugh at nhm.org>>
Sent: keskiviikko 8. lokakuuta 2025 22.18
To: Marko Mutanen <Marko.Mutanen at oulu.fi<mailto:Marko.Mutanen at oulu.fi>>
Cc: Lücking, Robert <R.Luecking at bo.berlin>; John Grehan <calabar.john at gmail.com<mailto:calabar.john at gmail.com>>; Kuoi Zhang <zhang.guo-yi at outlook.com<mailto:zhang.guo-yi at outlook.com>>; taxacom at lists.ku.edu<mailto:taxacom at lists.ku.edu>
Subject: Re: Taxacom: On describing new taxa without using phylogenetics: some suggestions

Marko Mutanen,

Given that our intent with phylogenetic inferences is to causally account for observed homologous characters, and this occurs via the non-deductive form of reasoning known as abduction, there are no logical arguments supporting the view that sequence data has some priority over other classes of characters. Homoplasy is an ad hoc hypothesis that is the product of inferring hypotheses, so it can't be used as an excuse to exclude morphological characters (indeed, one can just as easily argue that sequence data are just as prone to lead to hypotheses of homoplasy). The only rational approach to abductively inferring phylogenetic hypotheses is to abide by the requirement of total evidence (sensu Carnap 1950) and consider all relevant observed characters.

Kirk Fitzhugh

On Wed, Oct 8, 2025 at 12:02 PM Marko Mutanen via Taxacom <taxacom at lists.ku.edu<mailto:taxacom at lists.ku.edu>> wrote:
Dear Robert, all, I agree with you. Morphological evolution is indeed fascinating, but it should be studied in the context of DNA-based phylogenies rather than by continuing to infer evolutionary relationships solely from morphology, which is

Dear Robert, all,

I agree with you. Morphological evolution is indeed fascinating, but it should be studied in the context of DNA-based phylogenies rather than by continuing to infer evolutionary relationships solely from morphology, which is highly prone to homoplasy. Of course, when dealing with fossils, we must still rely on morphological data, but this inevitably introduces substantial uncertainty into the taxonomy of extinct taxa - likely persisting indefinitely.

Sincerely,

Marko M.

--

--

“Wrong does not cease to be wrong because the majority share in it.”

― Leo Tolstoy

-------------------------------------------------

J. Kirk Fitzhugh, Ph.D.

Curator of Polychaetes

Research & Collections Branch

Natural History Museum of Los Angeles County

900 Exposition Blvd

Los Angeles CA 90007

Phone: 213-763-3233

e-mail: kfitzhugh at nhm.org<mailto:kfitzhugh at nhm.org>

-------------------------------------------------