Science Wiki

Tree of life (biology) (commonly referred to as the tree of life) is a metaphor, a model, and a research program for the evolutionary relationships among all organisms. Its central idea is that living species descend from common ancestors and that their history of divergence can be represented as a branching diagram, a phylogenetic tree. The idea has a long history. Charles Darwin drew a single tree diagram in On the Origin of Species (1859), and Ernst Haeckel drew elaborate trees in the 1860s. Modern trees are inferred mainly from molecular phylogenetics, which compares DNA, RNA, and protein sequences, together with comparative genomics, morphology, and the fossil record.

A single bifurcating tree is an accurate summary for most animals, plants, and fungi, where inheritance is overwhelmingly from parent to offspring. It is less complete as a description of microbial life, where horizontal gene transfer (HGT), endosymbiosis, and hybridization add connections between lineages. Many biologists therefore describe the history of life as a tree with some reticulation, and some describe the deepest parts of it as a network. The disagreement concerns how much of each genome follows the tree, and not whether common descent occurred.

Early trees and kingdoms

Before molecular data, classification relied on morphology and development. Early systems divided life into two kingdoms (animals and plants). Haeckel added a third kingdom, Protista, in 1866. Robert Whittaker proposed five kingdoms in 1969 (Animalia, Plantae, Fungi, Protista, and Monera), a scheme that was developed further by Lynn Margulis. These systems grouped all prokaryotes together as Monera, and they grouped many unrelated single-celled eukaryotes as Protista. Morphological characters can mislead over long times, because unrelated lineages often evolve similar features (convergent evolution).

Ribosomal RNA and the three domains

In 1977, Carl Woese and George Fox reported that comparisons of ribosomal RNA sequences, obtained by cataloging oligonucleotides of the small-subunit rRNA, separated the prokaryotes into two distinct groups, the bacteria and a group that they first named archaebacteria (now Archaea). Woese, Otto Kandler, and Mark Wheelis formally proposed the three-domain system in 1990, with the domains Bacteria, Archaea, and Eukarya (Eukaryota).

The small-subunit rRNA (16S in prokaryotes and 18S in eukaryotes) was suitable for this purpose for several reasons. All cellular life has ribosomes, so the gene can be compared across the whole tree. Its function in translation constrains its sequence, so that changes accumulate slowly in the most important regions, while more variable regions provide information on recent divergence. Ribosomal RNA genes are also rarely transferred between species, though exceptions occur. It is not a perfect "molecular clock". Rates of change differ among lineages and among sites, and models of sequence evolution are needed to correct for this.

Sequence divergence

The simplest measure of divergence between two aligned sequences is the proportion p of sites that differ. Because a site can change more than once, p underestimates the number of substitutions per site. For the Jukes-Cantor model, in which all four nucleotides are equally likely to replace each other, the corrected distance is:

For example, for two sequences that differ at 20 percent of sites (p = 0.20), d = −0.75 × ln(0.733) ≈ 0.75 × 0.310 ≈ 0.23 substitutions per site. The correction grows rapidly as p approaches 0.75, at which point the sequences are saturated and carry no information about the time of divergence. This is why deep relationships are difficult to resolve, and why more realistic models, with different rates among sites and among types of change, are used.

Tree-building methods

Methods include distance methods (such as neighbor-joining), maximum parsimony, maximum likelihood, and Bayesian inference. The number of possible trees increases very rapidly with the number of taxa. For n taxa, the number of distinct unrooted bifurcating trees is the double factorial (2_n_ − 5)!!, and the number of rooted trees is (2_n_ − 3)!!. For 10 taxa there are 2,027,025 unrooted trees and 34,459,425 rooted trees, and for 50 taxa the number is about 3 × 10⁷⁴. Search heuristics, and statistical support measures such as the bootstrap and posterior probabilities, are therefore necessary. Systematic errors, such as long-branch attraction, in which fast-evolving lineages group together artificially, can give strongly supported but incorrect trees. Rooting a tree of all life requires an outgroup, and it is difficult because there is no known outgroup for cellular life. Rooting is often done with ancient gene duplications (paralogs) that predate the last universal common ancestor.

The traditional molecular tree divides cellular organisms into three domains. They differ in membrane lipids, in the structure of the cell envelope, and in their machinery for transcription and translation.

Bacteria

Bacteria are prokaryotes without a membrane-bound nucleus. Most have cell walls of peptidoglycan, and their membranes are made of ester-linked fatty acids. They show the greatest metabolic diversity of any group, and they use a wide range of electron donors and acceptors. Major lineages include the Cyanobacteria (also called Cyanobacteriota), in which oxygenic photosynthesis evolved and which changed the atmosphere in the Great Oxidation Event about 2.4 billion years ago. Another is the Pseudomonadota (formerly Proteobacteria), which includes the class Alphaproteobacteria, from which the mitochondrion descends. The mitochondrial ancestor was therefore an alphaproteobacterium, and not a member of the whole phylum. Genome-based classifications, such as the Genome Taxonomy Database, recognize well over 100 bacterial phyla. Many of them have no cultured members, and they were found by metagenomics. Among them is a large radiation of small-celled bacteria with reduced genomes, the Candidate Phyla Radiation, which is thought to depend on other organisms.

Archaea

Archaea were first thought to be unusual bacteria. Their cell membranes are made of isoprenoid chains joined by ether bonds to glycerol-1-phosphate, unlike the ester-linked fatty acids on glycerol-3-phosphate of bacteria and eukaryotes. Their machinery for transcription (a multisubunit RNA polymerase and TATA-binding protein), for replication, and for translation is more similar to that of eukaryotes than to that of bacteria, and many archaea have histones. They include many extremophiles, such as hyperthermophiles at hydrothermal vents, halophiles, and acidophiles, but they are also abundant in ordinary soils and oceans. Microbial methanogenesis, a source of biogenic methane, is carried out by archaea. It is not completely exclusive to them, because small amounts of methane are made by other processes, such as aerobic production by some plants and phytoplankton, so it is the main biological route and not the only one. Major groups include the Thermoproteota (formerly Crenarchaeota), the Euryarchaeota, the DPANN group of small-celled archaea, and the Asgard archaea.

Eukaryotes

Eukaryotes have a nucleus, an endomembrane system, a cytoskeleton, and mitochondria (or their reduced relatives). Their origin, eukaryogenesis, is described below. It is not the result of a "mathematically modeled" improbable event. It is inferred from genomic and cell biological evidence, and several scenarios are consistent with the data. The last eukaryotic common ancestor (LECA) already had mitochondria, a nucleus, sex, and many other features. Molecular clock estimates for its age range from about 1.2 to 1.8 billion years or older, with large uncertainty, and the first eukaryotic fossils may be older than that.

Major Eukaryotic Supergroups

Eukaryotic diversity is mostly microbial. The old kingdom Protista is not a natural group, and it has been abandoned. Phylogenomic analyses of many genes, together with studies of ultrastructure, divide eukaryotes into a small number of large groups, often called supergroups. Their number and membership have changed with new data and are still under discussion. Groups that are widely recognized include:

• Opisthokonta (animals, fungi, and their single-celled relatives) and Amoebozoa, which are often joined as Amorphea. The Obazoa clade contains the Opisthokonta and related groups.
• TSAR, which contains the Telonemia, the Stramenopiles (such as diatoms and brown algae), the Alveolates (including ciliates, dinoflagellates, and apicomplexans such as Plasmodium), and the Rhizaria.
• Archaeplastida (also called Plantae in a broad sense), which contains the glaucophytes, red algae, and green algae and land plants, which have plastids from a primary endosymbiosis of a cyanobacterium.
• Excavata (for example the euglenids and kinetoplastids, and the diplomonads and parabasalids), whose monophyly is uncertain.
• Other groups, such as the Haptista, the Cryptista, and the Hemimastigophora.

Newly described deep lineages

Sequencing of environmental DNA and the culture of rare protists have revealed several lineages that do not group clearly with the main supergroups. Among them are the Hemimastigophora, predatory flagellates with an unusual cell structure whose phylogenomic placement as a deep branch of the tree was shown in 2018, and the Provora, a group of predatory protists that was described in 2022, and the Meteora and related protists. The name Disparia has been proposed for a possible grouping of some of these lineages, but the membership of this group and its position in the tree are still being studied, and they have not been settled. A statement that a particular new phylum, Caelestes, and a protist Solarion arienae, were discovered in late 2025 could not be verified, and it was removed from this article. Such claims should be added only when they are supported by the published description of the organism. Studies of deep-branching protists matter for the reconstruction of the features of LECA, but statements that specific lineages preserve "proto-mitochondrial" physiology should be made with care, since modified mitochondria, such as hydrogenosomes and mitosomes, evolved several times from the mitochondrion.

The eocyte hypothesis and the two-domain tree

In 1984, James Lake and colleagues proposed, on the basis of ribosome structure, that eukaryotes are related to a group of archaea that then were called eocytes (the Crenarchaeota). This is the eocyte hypothesis. For decades, the three-domain tree, in which Archaea and Eukarya are sister groups, was more widely accepted. The sequencing of Asgard archaea changed this. Genomes assembled from metagenomes of marine sediments and hot springs, beginning with Lokiarchaeota in 2015 (the group was named after the Loki's Castle hydrothermal field), and with the other members, which include Thorarchaeota, Odinarchaeota, and Heimdallarchaeota, showed that they encode many proteins that were thought to be unique to eukaryotes. These "eukaryotic signature proteins" include homologs of actin, small GTPases, components of the ESCRT machinery, ubiquitin-like modifiers, and proteins of membrane trafficking. Phylogenomic analyses place the eukaryotes within the Asgard archaea, most closely related to the Heimdallarchaeia or to a related lineage. In this "two-domain" tree, the only primary domains are Bacteria and Archaea, with eukaryotes as a derived branch of the Archaea. Most recent phylogenomic studies support such a position, but the exact relationship is under study, and some analyses are sensitive to the models and the gene sets that are used. The claim that the issue is settled, or that Asgard archaea encode the "complete" cytoskeleton and trafficking system, is too strong. They have homologs of parts of these systems, and the eukaryotic versions are more complex.

The cultivation of an Asgard archaeon, Prometheoarchaeum syntrophicum, from marine sediment, which was reported in 2020 by Imachi and colleagues, showed a slow-growing cell that grows in a syntrophic association with bacteria and that has long, branching protrusions. It does not have a nucleus or complex organelles, and it supports models in which the host cell lived in close association with bacteria before the origin of the mitochondrion.

Models of eukaryogenesis

The mitochondrion descends from an alphaproteobacterial endosymbiont, according to the endosymbiotic theory developed by Lynn Margulis (1967), with earlier ideas by Konstantin Mereschkowski. Whether the host was a phagocytosing cell that "engulfed" the bacterium is debated. Alternative models propose that the symbiosis arose by syntrophy (the exchange of hydrogen or other metabolites), with the bacterium enclosed by protrusions from the host, and that the nucleus and the endomembrane system arose later. Models also differ in whether the mitochondrion was acquired early or late in the origin of eukaryotic complexity. A transfer of many genes from the endosymbiont to the host nucleus (endosymbiotic gene transfer) occurred, and the ancestor of plants and algae acquired plastids by a later primary endosymbiosis of a cyanobacterium.

Mechanisms

Bacteria and archaea exchange DNA by transformation (uptake of free DNA), by transduction (transfer by bacteriophages), by conjugation (transfer from cell to cell, often on plasmids), and, in some groups, by gene transfer agents. Eukaryotes also acquire genes from other organisms, but less frequently, mainly from endosymbionts, from viruses, and from other sources.

Consequences for the tree

The "tree" metaphor is based on vertical descent. In prokaryotes, the genes of one genome can have different histories. A gene for ribosomal RNA may be inherited vertically, a metabolic gene may come from a distant lineage, and a gene for antibiotic resistance on a plasmid may spread widely. The set of genes of a species (its pan-genome) is therefore not a single unit in the tree. Different genes give different trees, and the "tree of life" is better seen as the central tendency of many gene histories. Evidence supports a "tree-like" signal in a core of genes, especially the ones involved in information processing and in the ribosome, although this is a small part of each genome. Doolittle and others have argued that the base of the tree is better described as a network. Charles Darwin's tree is then an approximation, and the "web of life" is a better description of microbial genomes. It is also not correct to say that HTLV gene transfer "completely obscures" vertical descent in all prokaryotes. Rates of transfer differ among genes and among lineages, and an estimated fraction of genes in a typical bacterial genome has been transferred at some time, with a large range among estimates. Reticulate evolution also occurs in eukaryotes by hybridization, as in many plants, and by the transfer of genes between organelles and nuclei.

Polyphyletic Origins and Absence of Ribosomes

Viruses are not included in the tree of cellular life, because they do not have a single origin, they do not share a universal gene, and they have no ribosomes. The classification of viruses is a separate system. The relationship of viruses to cellular life, and the possibility that viruses took part in the origin of eukaryotic features, such as the nucleus, are debated.

Metagenomic Dark Matter and the Open Tree of Life

Large projects assemble the tree for all species. The Open Tree of Life combines published trees, and the Genome Taxonomy Database provides a genome-based classification of bacteria and archaea. About 2 million species have been formally described, of an estimated total of several million to hundreds of millions for all kinds of life, the larger numbers being driven by estimates for prokaryotes. A 2016 tree by Hug and colleagues, which was based on 16 ribosomal proteins from more than 1,000 new genomes, showed that bacteria make up most of the diversity, and that much of it is in lineages without cultured members. Such trees are inferences, and they depend on the genes, the models, and the sampling.

How can biologists root the universal tree of life without an external cellular outgroup?

Because all known cellular organisms descend from the Last Universal Common Ancestor (LUCA), no living cellular species can serve as an external outgroup to root the entire tree. In 1989, Naoyuki Iwabe and J. Peter Gogarten independently solved this problem by using ancient paralogous gene duplications (such as elongation factors EF-Tu/EF-G and F-/V-type ATPase subunits) that duplicated prior to LUCA, using one paralog tree to root the other and placing the root between Bacteria and the Archaea-Eukarya clade.

What is long-branch attraction (LBA) in molecular phylogenetics and how is it corrected?

Long-branch attraction is a systematic statistical artifact in which two rapidly evolving lineages accumulate convergent nucleotide or amino acid substitutions by chance and are erroneously grouped together as sister clades, particularly under maximum parsimony or oversimplified substitution models. Phylogenomicists mitigate LBA by using site-heterogeneous Bayesian mixture models (such as the CAT model), dense taxonomic sampling to break up long branches, and removing saturated fast-evolving alignment positions.

Why are informational genes less prone to horizontal gene transfer than operational metabolic genes?

According to the "complexity hypothesis" formulated by Jain, Rivera, and Lake (1999), informational genes—such as those encoding ribosomal rRNAs, core ribosomal proteins, and RNA polymerase subunits—participate in large, tightly coevolved macromolecular complexes with hundreds of physical interaction interfaces, making foreign orthologous replacements functionally incompatible. Conversely, operational metabolic enzymes typically belong to modular linear pathways and can be transferred horizontally on plasmids with immediate selective benefit.

• Phylogenetic tree
• Three-domain system
• Eocyte hypothesis
• Horizontal gene transfer
• Last universal common ancestor
• Asgard (archaea)
• Eukaryogenesis
• Endosymbiotic theory