In response to: “On Being a Fish” by Bret Weinstein (Vol. 2, No. 3).


To the editors:

Dr. Weinstein’s essay on the importance of phylogenetic classification of biological diversity is largely well-taken, and I look forwards to subsequent contributions to his announced series on biological nomenclature. His main thesis that taxonomic groups should reflect ancestry rather than functional similarity is correct. Modern systematists should, of course, endeavor to identify and name only those groups of organisms that are monophyletic, which is to say those taxa inferred to include all the descendants of a single common ancestral species. However, the crucial word here is “inferred,” and when Weinstein argues that taxa are defined by their ancestry and not by their characters, he makes a metaphysical leap that, in my view, sells short the taxonomic work that allows us to make claims about ancestors and evolution at all.

There is a school of thought, “tree-thinking,” whose proponents have argued that it is not necessary to understand how phylogenetic trees are constructed in order to be able to interpret them as pictures of evolutionary relatedness.1 Under that view, students and other users of phylogenies are supposed to accept the tree at face value, without knowing what evidence it is based on or how that evidence was interpreted. I have referred to this as evolutionary fundamentalism, which I described as “‘knowledge’ based on belief without the critical capacity to trace its empirical source or understand why we believe what we believe.”2 In short, rather than naïvely accepting a phylogenetic tree as a picture of how evolution went, we need to acknowledge that phylogenetic trees are evolutionary hypotheses based on characters (morphology, behavior, DNA or whatever). The idea of common ancestors is an inference from the pattern of shared character states among observed taxa, estimated by parsimony or some other optimality criterion. Groups may be “defined” however one pleases, but they are recognized by synapomorphies (shared derived character states).

To address Weinstein’s rhetorical questions about the criteria for penguins being birds: we are convinced of this not because we have any direct knowledge of ancestor-descendant relationships between them and some ur-bird, but because they share with other modern birds feathers, as well as a diapsid skull, a beak with no teeth, a pygostyle, forelimbs modified into wings, and other derived characteristics, or synapomorphies. Although the metaphysical cause of these shared features may be that the species bearing them have descended from a common ancestor, all of the evidence supporting that notion comes from comparison of homologous features. While individual characters may be gained more than once (such as endothermy in birds and mammals) or lost separately among different lineages (such as loss of legs in snakes and amphisbaenians), the preponderance of evidence from other characters still provides our best inference of phylogenetic relationships.

Thus, some characters, such as backbones, mammary glands or feathers, are unique to particular clades and actually do diagnose or “define” the taxa that possess them. Posing counterfactual hypotheticals such as milkless mammals or featherless birds is fine, in theory, but such creatures would either need to be recognized on the basis of other congruent characters, such as heterodont dentition (for the mammal) or some combination of the bird features mentioned above, or they would not be classifiable.

Another potentially confusing issue is Weinstein’s definitions of clade and monophyletic group. He says, “to be a clade, a group of creatures must include an ancestor and all of its descendants.” So far, so good. But then he suggests that, “your mother, and all of her offspring, comprise a monophyletic group.” This is another misleading idea promulgated by the tree-thinking folks: it represents an epistemological category error. Human maternal or paternal pedigrees represent halves of a network that is connected by meiosis and syngamy, not cladogenesis. No person is the sole ancestor of another person, and no person can give rise to descendants without a genetic contribution from another person. Clades or monophyletic groups are groups of taxa, not groups of individual organisms. As clearly described by the father of cladistics, Willi Hennig, groups of sexually-reproducing organisms connected by interbreeding, such as human families, exhibit tokogenetic relationships that form non-hierarchical networks of shared character states.3 Monophyly is a property of hierarchically-arranged groups of species. Molecular systematists also use these terms in describing gene genealogies, but usually only when there is a predominantly hierarchical pattern among the alleles sampled that is not obscured by recombination (tokogeny).

Just to reiterate the crucial point: common ancestry is an inference, not an observation, and hypotheses about common ancestry are results of systematics, not assumptions that justify it.

Andrew Brower


Bret Weinstein replies:

Dr. Brower and I are mostly in agreement about the optimal interaction between systematic inference and taxonomic nomenclature. Clades are monophyletic groups, and taxa should be defined based on evidence of monophyly, no matter how surprising the conclusion. In my essay I went one step further, arguing that surprising cladistic relationships are a feature and not a bug of this system. When children learn in grade school that they, bats, and whales are all fish, they will grow smarter, faster, because those conclusions imply a deep truth that is otherwise obscured by pre-Darwinian convention.

Brower and I do, however, differ over a few important points. For example, he correctly argues that phylogenetic trees are inferred from evidence, and wherever the evidence is noisy, the topology of the resulting tree may be dependent on the method of inference. But he incorrectly extrapolates that my proposal is therefore invalid. His argument is, approximately, this: if multiple methods generate distinct trees, then taxonomists will be forced to choose a cladistic methodology to resolve any given dispute. This then robs my argument of its central tenet: subjectivity can and should be removed from taxonomy so that groupings and nomenclature reflect our scientific understanding of the actual history.

My argument is not specific to any particular method, however, nor is it troubled by disagreement between methods, but the reason why was implied, rather than stated. We should not uproot cladistic nomenclature if a slight preponderance of evidence newly tips a given balance in favor of one interpretation of history over another. It would be unwise, and entirely impractical, for taxonomists and those who depend on their work to pivot around the latest published phylogeny, across the entire tree of life. My argument applies only when there is enough data that the cladistic interpretations generated from all valid methods converge. There is broad agreement, for example, that whales are a subclade of what we used to call artiodactyls, and that clade has now been redefined as the cetartiodactyls such that this taxon—dominated by functionally even-toed ungulates, and presumably derived from an ancestor that looked more like a goat than a tuna—correctly reflects the membership of cetaceans. When cladistic interpretations settle into a stable state, our taxonomy should reflect the conclusion, no matter how strange.

Brower goes on to argue that trees are only “hypotheses of relationship,” and he is right that cladists treat phylogenies this way in the scientific present, but the fact that they do so is a problem. I address this very issue in an upcoming piece about the ways different sciences use shared terms.

Brower’s point about my familial illustration of an ancestor’s relationship to the clade he or she defines contains an important truth that I should have stated explicitly: your mother and your father are equally the most recent common ancestor of the clade that derives from their union, and comprises all your full siblings, plus all their descendants. The reason I did not is that, one way or another, your father is a problem. For better or worse, he may not be who you think he is. And whoever he is, he may have produced some half-siblings, the consequence of an encounter with another woman before, or after the coupling that produced you. So, although your mother and genetic father, whoever he may be, are equally your ancestors, they may define different clades. To take an extreme case, Genghis Khan is the most recent common ancestor of a large monophyletic lineage, while each of the women with whom he reproduced was ancestor to a small monophyletic clade, most such clades having gone extinct by now.

Your mother too may have produced half-siblings, and may even have misrepresented the facts to you, your siblings and your father. But, by focusing on your mother’s offspring, and keeping your father out of it, I dodged the issue. It is unlikely that your mother, being a placental mammal, has produced any offspring that she does not know about, and even if she misled your dad into believing he was the genetic parent of some or all of your siblings, your sibling relationship, defined by your maternal parentage, is secure. Your mother and father each define a clade of descendants, but your mother is better at it.

In his letter, Brower argues that clades are defined and diagnosed by shared, derived characteristics (i.e. synapomorphies) such as the mammary glands, and single-bone dentary newly arising in mammals, and present in every species. But although it is almost always true that a clade can be diagnosed by such a trait, it is simply a convenient fact, not a logical requirement of nature. There is no trait, in any lineage, that could not be erased or reversed given enough time, and the right selective regime. Therefore, there can be no requirement that any trait be present for a given creature to exist within a monophyletic group.

The ability to say such things, succinctly, without caveats, is an excellent demonstration of the power of evolutionary logic when coupled with precise terminology. Why on earth would we hesitate to fully breathe that same Darwinian power into the taxonomic system?


Andrew Brower is professor of Biology at Middle Tennessee State University.

Bret Weinstein is a theoretical evolutionary biologist at The Evergreen State College in Olympia, Washington.

  1. David Baum and Susan Offner, “Phylogenies and Tree-Thinking,” The American Biology Teacher 70 (2008): 222-229. 
  2. Andrew Brower, “Rethinking Tree-Thinking: Cladograms, Ancestors and Evidence,” The American Biology Teacher 78 (2016): 380-84. 
  3. Willi Hennig, Phylogenetic Systematics (Urbana, IL: University of Illinois Press, 1966).