Biology · Phylogeny, Classification & the Tree of Life
Homology, convergent similarity and character evidence
A bird and a bat both fly. A bird and a human both have forelimbs with corresponding bones. Which resemblance should guide a tree? First specify the feature and the historical claim: the answer need not be the same for the whole limb and its use in flight.
Heredity, Variation & Evolution introduces inherited differences and common ancestry; Evolutionary Processes & Adaptation distinguishes observations from evolutionary mechanisms. Species, Speciation & Macroevolution provides useful context for species concepts and reticulate histories, though its detailed cases are optional here. You need to read small tables, add branch values and compare fractions. Each tree and probability rule is explained before it is used; no programming or previous taxonomic vocabulary is assumed.
Back to the Biology mapBy the end, you can
Define the character being compared and distinguish shared ancestry from independently acquired similarity.
A resemblance needs a precise comparison
A character is a feature selected for comparison; character states are its scored versions. “A limb element is present” is a different character from “powered flight is present.” Anatomy, behaviour, development and sequence positions can all provide characters, but each comparison needs an operational description.
Homology describes correspondence through common ancestry. Vertebrate forelimbs can retain corresponding skeletal elements despite differing shapes and functions. Our schematic marks a humerus connected with a radius and ulna in two modified limbs. It shows those relationships rather than actual bone proportions or every joint.
Bird and bat forelimbs have this deeper shared history, while their modifications for powered flight arose independently in the two lineages. The flight resemblance is therefore convergent at that level of comparison. Calling the limbs homologous does not make every detail of the wings a single ancestral invention. Conversely, convergence in flight does not erase the inherited forelimb relationships.
Researchers evaluate a proposed correspondence using anatomical position and connections, development, fossils and molecular evidence. Similar performance alone is insufficient. A character’s historical interpretation is a hypothesis that can change when better evidence shows a mistaken correspondence or a different origin.
Inherited limb organisation and independently modified flight surfaces
Read the diagram in words
Humerus, radius and ulna correspond in inherited forelimb organisation; simplified feather versus membrane symbols identify distinct wing modifications.
Bird forelimb
- Humerus
- Radius
- Ulna
| Description | Relationship |
|---|---|
| Humerus → Radius | Corresponding skeletal elements |
| Humerus → Ulna | Corresponding skeletal elements |
Feather-based flight surface; simplified
Count: 5
Bat forelimb
- Humerus
- Radius
- Ulna
| Description | Relationship |
|---|---|
| Humerus → Radius | Corresponding skeletal elements |
| Humerus → Ulna | Corresponding skeletal elements |
Membrane flight surface; hand supports omitted
| Bird forelimb | Bat forelimb | Relationship |
|---|---|---|
| Humerus | Humerus | Inherited humerus correspondence |
| Radius | Radius | Inherited radius correspondence |
| Ulna | Ulna | Inherited ulna correspondence |
Conceptual bird/bat forelimb correspondence only. No actual proportions, complete digit counts, joint mechanics or ancestral wing reconstruction.
Count evidence rather than repeated descriptions
Several measurements of one tightly linked structure may record the same underlying change. Treating them as fully independent votes can overstate the evidence. Likewise, adjacent sites, duplicated copies and selected gene regions can have relationships that a simple independent-character model ignores.
Fossils can provide anatomical combinations and time constraints even when recoverable DNA is absent. Missing information should remain missing rather than being scored as absence. An incomplete specimen cannot be said to lack a bone simply because the relevant region was not preserved.
Useful comparisons make their sampling and scoring choices visible. Researchers can examine whether conclusions survive alternative character definitions, added taxa and different models. A tree supported by a broad, well-characterised comparison is more informative than a grouping chosen because one familiar resemblance fits it.
Worked example
Locate evidence before claiming an origin
On the assigned rooted tree, O branches before the focal A/B and C/D lineages. X = 1 appears in C and D, whose shared stem can carry one gain. If you group all Y = 1 taxa instead, you select A and C. Their common ancestor in this tree also has B and D as descendants, so the selected pair is not a clade.
Two Y gains, one on A’s terminal branch and one on C’s, fit the scored observations. This demonstrates a possible convergent history under the assigned tree and equal-cost binary rule. It does not prove that the real history of an unknown trait contained exactly those two events.
Now suppose C’s Y state could not be observed. Record it as unknown, not 0. The available observations would no longer justify the same complete mapping. Missing data and a measured absence are different evidence, even when a blank table cell makes them look similar.
Check your understanding
Can bird and bat wings be homologous in one comparison and convergent in another?
Show reasoning
Yes. Their underlying vertebrate forelimbs include inherited corresponding structures, while modifications for powered flight arose independently. State whether you mean the forelimb organisation, a specific component or the flight adaptation.
Does state X = 1 in C/D prove that 1 is the ancestral state for all five taxa?
Show reasoning
No. In the assigned example the outgroup and A/B have 0, and a gain on the C/D stem explains 1. Ancestral-state inference depends on the tree, comparison and change model; “present in two taxa” does not settle it.
A fossil preserves no wrist region. Why is scoring “wrist element absent” unsafe?
Show reasoning
The element might simply be unobservable because that region is missing. Score unknown unless preservation or other evidence permits a reliable absence claim. Converting unknowns to absences invents character evidence.
What to carry forward
Specify the character, test its correspondence and keep observed states separate from reconstructed changes.
References
Resources supporting this lesson’s explanations and examples.
- Developmental homoplasy: convergence in cellular differentiation (Journal of Anatomy, 2010)Anthony GrahamAuthor research review — selected text
- The Tree Room — Reading trees and comparing evidence (Live teaching text; inspected October 2026)Understanding Evolution team · UC Museum of Paleontology, University of California, BerkeleyInstitutional scientific educationThe Tree Room — Reading trees and comparing evidence (Live teaching text; inspected October 2026)Common ancestry and relationshipsTip order, rotation and progress misconceptionsComparing charactersParsimony and its limitsResolution and branch support
- Foundational likelihood and bootstrap methods (Journal of Molecular Evolution 17:368–376, 1981; Evolution 39:783–791, 1985)Joseph FelsensteinResearch abstractEvolutionary trees from DNA sequences: a maximum likelihood approachConfidence limits on phylogenies: an approach using the bootstrap