Reading Phylogenetic Trees

A phylogenetic tree is a diagram that represents hypotheses about the evolutionary relationships among a set of organisms, genes, or populations, collectively called taxa (singular: taxon). Trees are built from characters — heritable traits, whether morphological features or DNA/protein sequence positions — that vary among the taxa being compared. The central logic is that taxa sharing more recently evolved (derived) characters are more closely related to each other than to taxa lacking those characters, because they inherited them from a more recent common ancestor. The vocabulary of a tree mirrors the vocabulary of a genealogy, and getting it precise matters. A tip (or leaf, or terminal taxon) sits at the end of a branch and represents a taxon in the analysis — typically a living species, a population, a gene sequence, or occasionally a fossil. A node is a branching point; an internal node represents a hypothesized common ancestor, the point at which one ancestral lineage split into two or more descendant lineages. A branch (or edge) connects nodes and represents a lineage persisting through time, sometimes scaled to represent elapsed time or amount of evolutionary change, sometimes drawn with no meaning beyond connectivity (a cladogram). The root is the node representing the most recent common ancestor of all taxa in the tree, and it anchors the direction of time, from root toward tips. A clade — also called a monophyletic group — is any ancestor plus all of its descendants, corresponding to an internal node and everything branching from it. Clades are the fundamental natural units of classification in modern systematics, previewed further in Monophyly and Clades. Critically, a tree is a statement about topology — the pattern of branching — not about the geometric arrangement of ink on a page. Any node can be rotated, swinging its descendant branches left-right, without changing the tree's meaning at all, a fact that trips up nearly every beginner and is addressed at length below.

Step 1: Characters carry the signal. Every tree is reconstructed from characters that vary across taxa — nucleotide identity at homologous genome positions, amino acid states, or discrete morphological traits (number of digits, presence of feathers, shape of a tooth cusp). The key requirement is Homology vs Analogy: characters must be homologous, meaning inherited from a common ancestor, not independently evolved resemblances (analogies/convergences), or the reconstruction will be misled.

Step 2: Distinguish ancestral from derived states. A character state present in the common ancestor of the whole group and retained widely is called plesiomorphic (ancestral); a character state that evolved anew within a lineage is apomorphic (derived). A derived character shared by two or more taxa because they inherited it from their immediate common ancestor is a synapomorphy — the actual evidence used to group taxa into clades. Shared ancestral characters (symplesiomorphies) do not indicate exclusive common ancestry — for example, "has a backbone" is ancestral for vertebrates and groups nothing useful within vertebrates, because it does not distinguish any subgroup from another.

Step 3: Build the topology. Systematists compare character states across taxa and infer the branching pattern that requires the fewest independent origins of derived traits (the parsimony criterion) or that best fits an explicit statistical model of character evolution (likelihood or Bayesian methods). This machinery is developed fully in Cladistics & Parsimony; for reading trees, what matters is that the resulting topology is a hypothesis, always revisable with new data.

Step 4: Root the tree. An unrooted tree only shows how taxa cluster relative to each other; it has no direction of time and no way to say which character states are ancestral. Rooting requires outside information. The standard method uses an outgroup — a taxon or clade known, on independent evidence, to have branched off before the common ancestor of the group under study (the ingroup). Because the outgroup diverged earlier, the point where it attaches becomes the root of the ingroup's tree, and characters shared by the outgroup and part of the ingroup are inferred to be ancestral.

Step 5: Read direction and support. Once rooted, time flows from the root toward the tips. Internal nodes closer to the root are older common ancestors; nodes closer to the tips are more recent splits. Modern trees are usually annotated with support values (commonly bootstrap percentages or Bayesian posterior probabilities) at each internal node, indicating how strongly the data support that particular branching event — a node with 52% bootstrap support is far less trustworthy than one with 98%.

Rotation invariance. Because a tree is a statement about topology alone, any internal node can be freely rotated — swiveled like a mobile hanging from a string — without altering a single relationship the tree conveys. Species 2 and Species 3 could be drawn with Species 3 on top and Species 2 below, and the tree would mean exactly the same thing, because both remain each other's closest relatives, descended from Node C. Only the pattern of connections (who shares which common ancestor with whom) is meaningful; the left-right order of tips at the page's edge is an arbitrary artifact of drawing.

Phylogenetic trees are not drawn by intuition; each one is a tested hypothesis, and the clearest way to see this is through the history of how biologists learned to root trees correctly and to distinguish true common ancestry from misleading resemblance.

Willi Hennig and the founding of cladistics (1950s–1966). Before the mid-twentieth century, classifications often grouped organisms by overall similarity, lumping together taxa that merely looked alike, or separating relatives that had diverged in appearance. The German entomologist Willi Hennig formalized a rigorous alternative in his 1950 work (translated and expanded as Phylogenetic Systematics, 1966): classification should reflect strictly genealogical relationships, and only shared derived characters (synapomorphies) — not shared ancestral characters (symplesiomorphies) or convergent similarities (homoplasies) — should be used to define groups. Hennig's insight that ancestral resemblance is uninformative was the conceptual breakthrough that made rigorous tree-building possible, and it directly explains why, on a real tree, "looking primitive" is not evidence of anything.

The outgroup criterion, tested against known genealogies. The logical method for rooting trees — comparing the ingroup to an outgroup known independently to have diverged earlier — was refined through the 1970s–1980s by systematists including James Farris and others working in the Hennigian tradition. Its validity has been repeatedly confirmed in cases where the true genealogy is independently known, most powerfully in experimental phage and viral phylogenies. In studies growing bacteriophage populations under controlled mutation and passaging (lineages developed from experimental evolution work, including studies of the T7 and φX174 phage systems from the 1990s onward), researchers know the true pedigree of viral lineages because they generated them by hand in the laboratory, sampling and freezing populations at each generation. When DNA sequences from these known lineages were fed into tree-building algorithms blind to the true history, the reconstructed topologies matched the actual, independently recorded genealogies with high fidelity, and outgroup rooting using an ancestral sequence recovered the correct root virtually every time. This gave direct, ground-truthed confirmation that shared-derived-character reasoning and outgroup rooting recover real evolutionary history rather than just convenient-looking patterns.

Molecular sequence trees versus morphology: convergent confirmation. A second, independent line of evidence for the reliability of tree-reading logic comes from cases where morphology-based trees and DNA-based trees, built by entirely different methods on entirely different data, converge on the same topology. The classic case is whales and hippos. Nineteenth- and twentieth-century morphologists, working mostly from fossil bones and teeth, placed whales in their own separate order distant from all living hoofed mammals. Beginning in the 1990s, molecular sequence data (work by groups including Milinkovitch, Gatesy, and colleagues using ribosomal and other gene sequences) instead grouped whales as nested within Artiodactyla, sister to hippos specifically. This was initially controversial precisely because it contradicted the morphological trees. The resolution came from fossil discoveries in Pakistan in the late 1990s and 2000s — including Pakicetus and Indohyus — which revealed transitional anklebones (a double-pulley astragalus) shared uniquely by whales and artiodactyls, a synapomorphy invisible in earlier, less complete fossil material. Independent molecular and (once better fossils were found) morphological evidence converged, and the whale-hippo clade is now one of the best-supported nodes in mammalian phylogeny — a textbook demonstration that tree topologies are falsifiable, testable claims, not arbitrary diagrams.

Example 1 — Reading relatedness from a topology, not from tip order. Suppose a rooted tree shows the following nested structure (written in "Newick" style, read from the innermost parentheses outward): `(Outgroup,((Human,Chimpanzee),Gorilla))` Reading this: the innermost pair, (Human, Chimpanzee), share a more recent common ancestor with each other than either does with Gorilla. That pair, plus Gorilla, forms a larger clade, and the Outgroup diverged before the common ancestor of all three. Now suppose the tree is redrawn with tips reordered left-to-right as: Gorilla, Chimpanzee, Human, Outgroup. A student who reads relatedness by adjacency on the page would wrongly conclude Chimpanzee is most closely related to Gorilla (since they're drawn next to each other) and that Human is somewhat separate. This is incorrect. The topology has not changed — only the arbitrary left-right order of rotation at each node has changed. To find the true closest relative of Human, trace from the Human tip back to its nearest internal node, and ask what else attaches to that same node. That node also holds Chimpanzee, regardless of where Chimpanzee happens to be drawn on the page. Humans and chimpanzees share a common ancestor roughly 6–7 million years ago, more recent than the common ancestor humans/chimps share with gorillas (roughly 8–9 million years ago), and this is a topological fact independent of drawing style.

Example 2 — Polarizing a character using an outgroup. Suppose you are studying four lizard lineages (A, B, C, D) and one trait: presence or absence of limbs. Data:

Independent morphological and molecular evidence establishes the topology `(Outgroup,(A,((B,C),D)))`. Because the outgroup has limbs, and limbs are also present in the ingroup taxon that split off earliest (A) and in D, the most parsimonious inference is that limb-bearing is the ancestral (plesiomorphic) state for this whole group, and limblessness is a derived (apomorphic) state that evolved once, in the common ancestor of the (B, C) clade — not independently in B and in C. This is exactly the logic used in real studies of limb loss in squamates (skinks and relatives), where molecular phylogenies have shown that limblessness evolved more than 25 times independently across different lizard and snake lineages — a striking case of convergent (homoplastic) evolution, detectable only because outgroup-rooted trees let researchers polarize the character and count independent origins, rather than assuming all limbless lineages are simply "closely related because they look similar."

Willi Hennig (1913–1976) was a German entomologist, specializing in dipteran (fly) systematics, who spent much of his career working at the German Entomological Institute in Berlin. Trained in traditional comparative morphology, Hennig grew dissatisfied with classification schemes that grouped organisms by overall resemblance or by subjective judgments of "primitiveness" versus "advancement." Drawing on ideas that had circulated in a more informal way among earlier systematists, he built these ideas into an explicit, rigorous logical system: relationships should be diagrammed strictly according to genealogy, and the only valid evidence for uniting two taxa into a group is a shared character state that evolved after their common ancestor split from other lineages (a synapomorphy). Hennig published his framework first in German in 1950, and it reached a much wider international audience with the 1966 English translation, Phylogenetic Systematics. His method, initially called "phylogenetic systematics" and later widely known as cladistics, transformed taxonomy from a largely intuitive art into an explicit, testable, repeatable procedure, and it underlies essentially every modern tree-building method, from simple hand-drawn cladograms to today's large-scale genomic phylogenies.

Reading trees correctly is not an academic nicety; it underlies practical decisions across biology and medicine. In public health, phylogenetic trees built from viral genome sequences let epidemiologists trace how a pathogen spread, identify superspreading events, and estimate when an outbreak began — this was central to tracking SARS-CoV-2 lineages and identifying variants of concern in near-real time during the COVID-19 pandemic, and to reconstructing HIV transmission chains in criminal and public-health investigations. In conservation biology, trees help identify evolutionarily distinct lineages (species with few close relatives, representing large amounts of unique evolutionary history) for prioritized protection, since losing such a lineage erases more of the tree of life than losing a species with many close relatives. In agriculture, phylogenies of crop wild relatives guide breeders to species carrying useful disease-resistance genes by pointing to close relatives of cultivated crops. In drug discovery and vaccine design, understanding which viral or bacterial strains form a clade (and are therefore likely to share vulnerabilities) shapes which pathogens a vaccine or drug is designed to target. In biodiversity science, correctly rooted trees are the basis for estimating divergence times, testing hypotheses about mass extinctions, and understanding how Monophyly and Clades and Homology vs Analogy structure the classification of all life.

Ask the tutor: If molecular data and morphological data suggest two different topologies for the same group of species, how do biologists decide which tree — if either — to trust?