Phylogeny is the evolutionary history and relationships among organisms, usually represented as a branching diagram called a phylogenetic tree. Think of it as a family tree for life itself, where every fork in the diagram represents an ancestor that split into two separate lineages. The concept is straightforward, but reconstructing those relationships from the evidence available today is one of the most challenging problems in biology. Molecular data have transformed the field in the past few decades, yet even with whole genomes in hand, some parts of the tree of life remain genuinely contentious.
What a Phylogenetic Tree Actually Shows
A phylogenetic tree is a diagram of inferred ancestor-descendant relationships. Each tip of the tree represents a living or extinct group. Each internal branch point, called a node, represents a hypothetical common ancestor shared by everything downstream of it. The pattern of branching, rather than the order of the tips from left to right, carries the information. Two species sitting next to each other on the page are not necessarily closer relatives than two species on opposite ends of the diagram; what matters is how recently they share a common ancestor.
One of the most persistent misunderstandings in biology is treating a tree as a ladder of progress, reading it from left to right and assuming that lineages appearing on one side are more “primitive” than those on the other. Researchers have called this the “primitive lineage fallacy,” and it leads to genuinely misleading conclusions about how traits evolved. A lineage that branched off early is not frozen in time; it has been evolving for exactly as long as every other living lineage. Calling sponges or bacteria “primitive” because they sit near the base of a tree misrepresents what the tree is saying.1PubMed. Tree thinking for all biology: the problem with reading phylogenies as ladders of progress
Another common confusion involves the terms used to describe groups on a tree. A group that includes an ancestor and all of its descendants is considered monophyletic, and most modern classification systems insist on recognizing only these kinds of groups. A group that includes an ancestor and some but not all of its descendants is called paraphyletic. Reptiles, for instance, are paraphyletic in the traditional sense because they exclude birds, even though birds descend from the same ancestor. Some researchers argue that paraphyletic groups are a natural transitional stage in evolution and should still be recognized in classification, but strict phylogenetic classification rejects them.2TAXON. Paraphyletic groups as natural units of biological classification
Building a Tree From Molecules and Morphology
Historically, phylogenetic trees were built from physical characteristics: bone shapes, leaf structures, shell patterns, tooth counts. Researchers would identify shared features thought to be inherited from a common ancestor and use them to group organisms. This morphological approach still has value, but it can be misled when unrelated species evolve similar-looking traits independently. A dolphin’s streamlined body and a shark’s streamlined body look alike but arose separately.
Since the 1960s, DNA and protein sequences have increasingly become the primary data for tree-building. Molecular data offer vastly more characters to compare, and they can be gathered from organisms that leave few or no fossils. A study comparing over two hundred molecular and morphological trees found that both approaches generally track a shared underlying evolutionary history: the level of disagreement between them, while real, was far lower than you would expect if either type were producing random noise. Molecular trees showed slightly higher internal consistency on average, but the difference was not large enough to be statistically meaningful, and the authors cautioned against automatically favoring molecules when the two conflict.3PubMed. Congruence of morphological and molecular phylogenies
In practice, most modern phylogenetics is molecular. Researchers extract DNA or RNA sequences from the organisms of interest, align those sequences to identify corresponding positions, and then use computational methods to find the tree that best explains the observed similarities and differences.
How Computational Methods Differ
Two broad families of methods dominate phylogenetic inference. Maximum likelihood asks: given a mathematical model of how DNA evolves, which tree makes the observed sequences most probable? Bayesian inference flips the question: given the data and a prior expectation about what the tree might look like, what is the probability of each possible tree? Both methods rely on models of sequence evolution and can produce subtly different results depending on the assumptions fed in.
One well-known issue is that Bayesian analysis tends to express higher confidence in its results than bootstrap-based likelihood analysis does. Bayesian posterior probabilities can reach 100% for branches that maximum likelihood bootstrapping supports at only about 80%.4PubMed Central. Bayesian and maximum likelihood phylogenetic analyses of protein sequence data under relative branch-length differences and model violation Whether this means Bayesian methods are overconfident or bootstrapping is too conservative has been debated for years. Detailed comparisons show that the two methods’ bootstrap-based measures track each other closely, but Bayesian posterior probabilities and likelihood bootstraps are only moderately correlated, suggesting they are measuring somewhat different things.5Molecular Biology and Evolution. Comparison of Bayesian and Maximum Likelihood Bootstrap Measures of Phylogenetic Reliability
Both frameworks can also go wrong when the data contain ambiguous or missing positions. Variation in how fast different sites in a gene evolve can interact with gaps in the data to produce strongly misleading results, and in Bayesian analyses, the choice of prior assumptions about branch lengths and rate variation can make the problem worse.6Systematic Biology. The Effect of Ambiguous Data on Phylogenetic Estimates Obtained by Maximum Likelihood and Bayesian Inference The practical takeaway is that no single method is foolproof, and researchers routinely run multiple analyses and check whether the results agree.
Long-Branch Attraction and Other Systematic Errors
Even with enormous datasets, phylogenetic reconstruction can be tripped up by predictable patterns of error. The most notorious is long-branch attraction. Because DNA has only four nucleotide letters, two unrelated lineages that evolve quickly will, by chance alone, accumulate many of the same mutations at the same sites. A tree-building algorithm can mistake this coincidence for evidence of shared ancestry, pulling those fast-evolving lineages together on the tree even though they are not closely related.7Current Biology. Systematic errors in phylogenetic trees
Long-branch attraction is especially dangerous because adding more data does not fix it. With a small dataset, random noise can group species incorrectly, but adding more genes will usually wash that noise out. Long-branch attraction, by contrast, gets stronger as you add data, because the misleading pattern is consistent across genes. Correcting for it requires better models of evolution that account for differences in rates among lineages, or choosing genes that evolve at a pace suited to the depth of the question.
This is one reason the choice of genetic markers matters so much. Ultraconserved elements, stretches of DNA that barely change across vast evolutionary timescales, have become popular for resolving deep branches in the tree of life. Combined with species-tree methods, they proved effective at resolving relationships among the major groups of placental mammals.8PubMed Central. Ultraconserved elements are novel phylogenomic markers that resolve placental mammal phylogeny when combined with species-tree analysis But ultraconserved elements have their own weakness: because they evolve slowly, they carry little signal for groups that split apart in rapid bursts of diversification, and their short alignment lengths can introduce random error into individual gene trees.9Molecular Biology and Evolution. Optimizing Phylogenomics with Rapidly Evolving Long Exons: Comparison with Anchored Hybrid Enrichment and Ultraconserved Elements
When Different Genes Tell Different Stories
Perhaps the most humbling realization of the genomic era is that different genes in the same organism can have genuinely different evolutionary histories. You might expect that if you build a tree from gene A and a tree from gene B, both sampled from the same set of species, they would agree. They often do not. This gene-tree discordance is not a failure of the methods; it reflects real biological processes.
The most common cause is incomplete lineage sorting. When a species splits into two, and then one of those daughter species splits again shortly afterward, the ancestral population may not have had time to “sort” its genetic variation. As a result, some genes track one branching pattern while others track a different one, and neither is wrong from the gene’s perspective. In extreme cases, species can sit in what researchers call the anomaly zone, where the most common gene tree is actually different from the true species tree. A study of the plant family Amaranthaceae found exactly this situation: rapid ancient radiations left the backbone of the family deep in the anomaly zone, with the species tree failing to be the most frequent gene tree.10PubMed Central. Disentangling Sources of Gene Tree Discordance in Phylogenomic Data Sets: Testing Ancient Hybridizations in Amaranthaceae s.l. Similar findings emerged in eared seals, where explosive diversification in the Southern Hemisphere produced high levels of genealogical conflict best explained by incomplete lineage sorting rather than hybridization.11PubMed. Introgression across evolutionary scales suggests reticulation contributes to Amazonian tree diversity
Hybridization and introgression are the other major source of discordance. When species interbreed, genes from one lineage get woven into the genome of another, creating a web of relationships rather than a clean branching tree. This reticulate evolution is common in plants. Studies of the genus Epimedium in Japan, for example, found complex reticulate relationships among recently diversified species, with frequent signals of historical interspecific introgression that likely contributed to their diversification across different environments.12PubMed. Reticulate evolution, introgression, and recent diversification in Epimedium sect. Macroceras Similar patterns appeared among Amazonian trees in the genus Brownea, where gene tree incongruence was best explained by reticulation rather than incomplete lineage sorting alone.13PubMed. Introgression across evolutionary scales suggests reticulation contributes to Amazonian tree diversity
In microorganisms, the problem is even more severe. Horizontal gene transfer, where genes move between completely unrelated organisms rather than being passed from parent to offspring, is rampant in bacteria and archaea. Different genes within the same bacterial cell can have wildly different histories, to the point that some researchers have questioned whether a single branching tree can meaningfully describe prokaryotic evolution at all.14PubMed Central. Horizontal Gene Transfer and the History of Life The idea of a neatly bifurcating “tree of life” gives way, in these organisms, to something more like a tangled web or network.15Nature Reviews Microbiology. Horizontal gene transfer, genome innovation and evolution Some have gone so far as to propose “lateral genomics” as a framework, questioning the relevance of tree-like thinking for prokaryotic evolution entirely.16PubMed Central. Horizontal gene transfer: essentiality and evolvability in prokaryotes, and roles in evolutionary transitions
Putting Dates on the Branches
A tree tells you who is related to whom, but it does not automatically tell you when those lineages split. Turning a phylogeny into a timetree requires a molecular clock: the idea that mutations accumulate in DNA at a roughly steady rate over time, so the number of differences between two sequences reflects how long ago they diverged. The reality is messier, because rates vary across lineages and across genes, but relaxed-clock models can accommodate that variation.
The clock still needs calibration, and fossils are the main source. A fossil places a minimum age on a split: if a 66-million-year-old fossil clearly belongs to one branch of a tree, the split that created that branch must have happened at least 66 million years ago. Researchers translate these minimum and maximum age bounds into statistical distributions that feed into the analysis. The choice of how to set these distributions matters enormously. Different strategies for borrowing information across fossil calibration points can shift age estimates by tens of millions of years for deep nodes in the tree.17PubMed Central. Comparison of different strategies for using fossil calibrations to generate the time prior in Bayesian molecular clock dating
Modern approaches typically use “soft bounds,” meaning that the probability of the true divergence time falling outside the fossil-defined window is small but not zero. This flexibility accounts for the fact that fossils only record organisms that happened to be preserved and discovered; the actual split almost certainly predates the oldest known fossil of any group.18Molecular Biology and Evolution. Bayesian Estimation of Species Divergence Times Under a Molecular Clock Using Multiple Fossil Calibrations with Soft Bounds
Unresolved Branches and Open Frontiers
Some parts of the tree of life remain stubbornly unresolved despite decades of work. One of the most famous disputes concerns the base of the animal tree. Traditional thinking placed sponges as the earliest-branching animal lineage, but several genome-scale analyses have instead placed comb jellies, or ctenophores, in that position.19PubMed. The ctenophore lineage is older than sponges? That cannot be right! Or can it? If ctenophores really are the sister group to all other animals, it would mean that nervous systems and other complex features either evolved more than once or were lost in sponges and placozoans. The debate remains active, and it illustrates how even massive datasets cannot always overcome the challenges of ancient, rapid divergences compounded by long-branch attraction.
At the microbial scale, the unknowns are staggering. In just the past four years, the number of identified bacterial families and orders in major genomic databases has doubled, and entirely new high-level groups continue to be discovered. A decade ago, no genomic data existed for the Asgard archaea, a group now considered the closest living relatives of eukaryotic cells. Other vast groups, like the Candidate Phyla Radiation bacteria, were known only through a handful of representatives until metagenomics began pulling their DNA directly from environmental samples.20Genome Biology and Evolution. Microbial Diversity and Open Questions about the Deep Tree of Life The deep tree of life is being redrawn in real time.
How Phylogeny Reshapes Classification
The flowering plant family tree is a good case study for how phylogenetics changes the names and groupings you encounter in a field guide. The Angiosperm Phylogeny Group, an international collaboration, has issued successive classifications of flowering plant families based on molecular phylogenies. The second edition alone introduced several new orders and rearranged many families to reflect their true evolutionary relationships rather than superficial resemblance.21Botanical Journal of the Linnean Society. An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG II Early rounds of reclassification were dramatic, shuffling hundreds of genera into new families. But the system has matured. A recent analysis of New Caledonian flora found that family-level changes have dropped to nearly zero since the latest APG revisions, suggesting the molecular classification has stabilized enough to be applied confidently outside specialist journals.22PubMed Central. The family classification of the New Caledonian angiosperms has become more stable with the application of the APG system
Practical Applications Beyond Classification
Phylogenies are not just academic exercises in drawing trees. They have become practical tools across fields that might surprise you.
In epidemiology, the shape of a pathogen’s phylogenetic tree carries information about how a disease is spreading. Trees built from virus sequences sampled during an outbreak have distinct shapes depending on whether transmission is happening through a single superspreader, broad community spread, or environmental exposure. Researchers have shown that computational classifiers trained on these topological features can predict an outbreak’s transmission dynamics from the tree shape alone, giving public health teams information they would otherwise need intensive contact tracing to obtain.23PubMed Central. Phylogenetic tree shapes resolve disease transmission patterns
In conservation, phylogenies help decide which species to prioritize when resources are limited. The logic is that a species sitting alone on its own long branch of the tree of life represents more unique evolutionary history than a species surrounded by close relatives. Lose it, and you lose a disproportionate share of biological diversity. The EDGE metric, which combines a species’ evolutionary distinctiveness with its extinction risk, was applied to a near-complete mammal phylogeny to produce a global priority list. Some of the species ranking highest, like the long-beaked echidnas and the aye-aye, are ones that might otherwise be overlooked in favor of more charismatic animals that happen to have many close living relatives.24PubMed Central. Mammals on the EDGE: conservation priorities based on threat and phylogeny Similar phylogenetic weighting has been applied to threatened species in New Zealand and elsewhere, where prioritization protocols increasingly incorporate evolutionary distinctiveness alongside raw species counts.25Biological Conservation. Balancing phylogenetic diversity and species numbers in conservation prioritization, using a case study of threatened species in New Zealand
In drug discovery, phylogenies can help find promising new medicines by pointing toward groups of plants that are unusually rich in bioactive compounds. If traditional medicine from multiple independent cultures uses closely related species for the same purpose, that clustering on the tree suggests the medicinal activity is real and inherited. Cross-cultural phylogenetic comparisons of traditionally used plants have been shown to focus bioprospecting efforts on a subset of species richer in bioactive molecules than you would find by random screening.26PubMed Central. Phylogenies reveal predictive power of traditional medicine in bioprospecting Researchers have extended this approach specifically to plants used in traditional cancer management, using a genus-level angiosperm phylogeny to predict which lineages have elevated potential for drug discovery.27PLANTS, PEOPLE, PLANET. Phylogeny and bioprospecting: The diversity of medicinal plants used in cancer management
Phylogenetic Comparative Methods
Beyond sorting organisms into groups, phylogenies underpin an entire statistical toolkit for studying how traits evolve. The core insight is simple: closely related species tend to resemble each other, which means that traits measured on different species are not statistically independent. If you want to test whether, say, brain size correlates with diet across mammals, you cannot just plot every species as an independent data point, because all the primates will cluster together and all the rodents will cluster together, inflating your apparent sample size. Phylogenetic comparative methods correct for this shared ancestry, treating the phylogeny as a kind of statistical structure that accounts for which data points are expected to resemble each other and by how much.28PubMed Central. Controlling for non-independence in comparative analysis of patterns across populations within species
These methods have been standard practice in evolutionary biology since the 1980s, and their sophistication has grown enormously. Modern approaches do not just control for phylogenetic relatedness; they model the tempo and mode of trait evolution across the tree, testing hypotheses about whether a trait evolved gradually, in bursts, or in response to ecological opportunity.29Systematic Biology. Rethinking phylogenetic comparative methods The accuracy of these analyses depends directly on having a reliable phylogeny. Get the tree wrong, and your statistical conclusions about how traits evolved are built on a faulty foundation.
Coevolution and Tangled Trees
One of the more elegant uses of phylogeny is testing whether two groups of organisms have evolved in lockstep. Parasites and their hosts are a classic case. If a parasite lineage speciates every time its host lineage speciates, the parasite tree and the host tree should be mirror images of each other. In practice, perfect mirroring is rare. A study of feather lice on pelicans and their relatives found extensive evidence of cospeciation: the louse phylogeny and the bird phylogeny showed significant congruence across multiple analytical methods. But the match was not perfect, with occasional host-switching events disrupting the pattern, and the strength of the cospeciation signal varied depending on which analytical method and which version of the host phylogeny was used.30Systematic Biology. Multiple Cophylogenetic Analyses Reveal Frequent Cospeciation between Pelecaniform Birds and Pectinopygus Lice
Coevolutionary studies like these are not just curiosities. They help reconstruct the history of host-parasite associations, predict which parasites might jump to new hosts, and even calibrate divergence times when fossils are lacking for one of the two groups. When two phylogenies agree, each serves as an independent check on the other, strengthening confidence in both.