Data table
| Group | Described living species | Undescribed species (conservative to liberal) | Alive now (published) | Biomass, Gt C (published) |
|---|
How it's built
Branching. The tree is the Open Tree of Life synthetic tree, cut at family level: about ten thousand families of living organisms, joined the way roughly two thousand published phylogenies agree they're related. Where Open Tree can't keep a family together as one branch, the family is drawn where two thirds of its sampled genera sit. A family Open Tree leaves unresolved at the base of a whole domain is placed with the rest of its classification group instead, and its hover card says so. The tree is only as resolved as Open Tree's synthesis: among bacteria especially, some textbook groups don't hold together there (the gammaproteobacteria, for one, so Pseudomonas meets E. coli deeper than it should). Species that belong to no family Open Tree recognises are grouped as Other or Unplaced.
Extinct life. About four thousand extinct families from the Paleobiology Database are drawn beside the living, each from its first fossil to its last, read conservatively (the young end of each dated interval, or the old end of a last record dated only to a whole epoch: Tyrannosaurus 72 to 66 million years ago). A first appearance is checked against the oldest records of the genera in the family first: one far older than all of them rests on lumped or misfiled records, and so does a single genus far older than the next (the reef corals Merulinidae are listed from 347 million years ago, from one genus; the next is 234). Those are set aside, and the hover card says so. Before its first fossil, a lineage is dotted: it must have existed from the moment it split off, but nothing of that age has been found. A few hundred are placed by published phylogenies; most hang from their order or class, since no study places them more exactly, and the hover card says which. Where the synthetic tree can't relate a fossil group to the living (no one study holds both tyrannosaurs and birds), its classification places it on the stem of its nearest living relatives: tyrannosaurs beside the birds, inside Coelurosauria. The Paleobiology Database's classification, finer than Open Tree's for fossils, nests them (Massospondylus in Sauropodomorpha, in Saurischia, in Dinosauria) and takes out those it files elsewhere (Morganucodon is a mammaliaform, on the mammals' stem, not among them). A fossil filed in a living group but older than that group's first split, by more than the molecular date's uncertainty, is a stem member, and sits on the group's stem. Switch to Living to hide them. The fossil record is far from complete: soft-bodied life and microbes barely fossilise, which is what the pencil lines are for.
Still to be found. Like a periodic table drawn with its gaps left in, the chart marks what should exist but hasn't been found. The cells beside each group's name are its species: filled for those named, hollow for those thought to exist and not yet described (from the estimates under The unknown). A dotted stretch on a living family's stem is a ghost lineage: it split off from its relatives at that point, so it existed from then on, but its oldest known fossil is younger; fossils of that age should be out there. And for the extinct, the known species are a small fraction of those that lived: perhaps 250,000 fossil species are described, of the one to four billion species thought to have ever existed (Raup 1991), since most leave nothing behind.
The root. Life is drawn as a three-way split between Bacteria, Archaea and eukaryotes, as Open Tree leaves it. Most recent studies instead place eukaryotes inside the Archaea, as relatives of the Asgard archaea (Eme et al. 2023, Nature). If they're right, your last common ancestor with any archaeon lived around two billion years ago, when eukaryotes began, not at the root. Where that matters, the page says so.
Groups and colours. Bacteria and Archaea are domains; fungi, animals and green plants (with the green algae) are kingdoms or near it. “Other eukaryotes” isn't a natural group: it's everything left over, from diatoms and amoebae to ciliates, and it includes the red algae, which are relatives of plants. Each group is drawn in an ink standing for a pigment it's known for: the blue-green phycocyanin of cyanobacteria, the purple membrane of salt-loving archaea, the gold-brown of diatoms, chlorophyll, a mushroom brown, haemoglobin red. They're reminders, not claims: most bacteria make no phycocyanin, and globins turn up across all of life.
Dates. Each branch point is dated with TimeTree, which combines thousands of molecular-clock studies. Where TimeTree has no estimate, the branch point is spaced evenly between the dated points above and below it (the BLADJ method). Fossils are minimum ages: a lineage is at least as old as its oldest checked fossil, so a branch point younger than one is moved just older than it (fossil calibration: Dinosauria to about 232 million years ago), along with anything above it that then has no room. The hover card says which kind of date you're looking at, and where TimeTree gives a confidence interval, the card and the written answer give it too (“range 6–7”); it isn't drawn. A date moved to fit the others is marked as adjusted. More than two lines from one point means the tree doesn't resolve how they're related; the answer says so. The root is an assumption, not a measurement: 4.25 billion years. Dates deeper than about 600 million years are uncertain: estimates for the last universal common ancestor run from about 3.8 to 4.5 billion years (Moody et al. 2024 put it near 4.2), and for the first animals from about 550 million years (the fossils) to 800 (molecular clocks).
Line weight. How many living species science has named in a lineage, on a log scale: a family of ten species is a hairline, the insects (about a million) a broad stroke. They're counted in the Open Tree taxonomy, leaving out what it flags as extinct, names that aren't proper species (Aus sp. 3), species the Paleobiology Database lists as fossils, and, among four-legged vertebrates, names known only from GBIF, which there are mostly fossils and duplicates. The counts land within about a fifth of the standard checklists for most groups: about 5,800 mammals (the Mammal Diversity Database lists about 6,600), 10,800 birds (the IOC list, about 11,000), 40 of the 41 cats. They run high where the taxonomy still carries old synonyms (the fungi, daisies, pines) and low where recently described species are missing (amphibians, nematodes). A family the count finds none in keeps Open Tree's own count, which may include fossils, and is drawn lighter. Line weight shows what has been named, not how much of life a group is: the bacteria make up most of life's cells but have only about 22,000 named species.
How many are alive. Counts of living individuals, and their biomass (Bar-On, Phillips & Milo 2018), are published for only a few dozen groups: people and livestock, birds, ants, nematodes, krill, the cyanobacteria and a few more. They're on those groups' hover cards, with their units: cells for single-celled life, where a cell is an individual; shoots for land plants; cell-equivalents for fungi, which have no countable individual; individuals for animals. They aren't drawn. No source covers every family, sharing a group's total among its families by species count gives impossible numbers (over a million molluscs for every square metre of the Earth), and cells, shoots and animals can't be added up.
Space. A group's height follows how many families taxonomists recognise in it (compressed, so the insects don't swamp everything), not how diverse it is. Bacteria and Archaea hold most of life's genetic and metabolic diversity, and most of its cells, but have few named families and species, so they get a thin strip at the top and thin lines: the chart shows what has been described. Their hover cards give how many cells there are.
The unknown. Undescribed species are estimated from each group's published total as a share: for every species named, how many more are thought to exist (against the same source's own count of the named). About 86% of eukaryotes are thought to be unnamed (Mora et al. 2011), around 80% of insects (Stork 2018), and almost all of the fungi and prokaryotes. Each family takes the share of the nearest group around it that has an estimate, so a group with none of its own, like the diatoms and other protists, is taken to be as unexplored as the eukaryotes as a whole, not fully named. Estimates for prokaryotes run from a few million species to a trillion, which is why there are two settings: Conservative uses the low end of each range, Liberal the high end, and it is never lower. A group's total is the sum of its families, so it can differ from the published figure for the group. Treat these as orders of magnitude: Mora's extrapolation in particular has been questioned.
Missing links. The pencil lines are lineages we infer but haven't seen. Those reaching today stand for undescribed species, as many as each group's estimate allows. Those ending earlier stand for extinct relatives, more along older lineages, since nearly every species that ever lived has died out. How many veins there are follows those estimates; where each one runs is only illustration.
Silhouettes. The groups you select are drawn from PhyloPic, a library of free silhouettes of organisms, tinted to their lineage. A group without one of its own borrows its nearest relative’s, and the credits above the chart name whose it is.
Going deeper. The page's own data stops at families. Zoom until a family is tall on screen and it loads what's inside it (subfamilies, genera, then species) from the Open Tree of Life, a few levels at a time. Searching for a species loads the way down to it. Inside a family, a branch point takes Open Tree's date for it where there is one: the median age from the published dated trees it was built from. Where there isn't, it's estimated by a pure-birth (Yule) model from species counts, between the nearest dated point above it and today. Those counts include fossil species, so the estimates run old. The hover card says which kind of date you're looking at. A family's described species and unknowns are shared down by Open Tree's species counts, so a genus's figures are approximate; a species is one. A species with a census of its own, such as humans, chickens and cattle, gives it on its card. Extinct groups are left out using Open Tree's flags and the Paleobiology Database, and a branch point whose other lines were all fossils gives way to the living one. Species known only from GBIF or IRMNG are left out too, since that's where unflagged fossils tend to hide (it may drop an obscure living one). The few families Open Tree splits apart load from its taxonomy instead, with their members but not their branching order.
Sources and download
The sources: Open Tree of Life (Hinchliff et al. 2015, PNAS; synthesis and taxonomy as above), TimeTree (Kumar et al. 2022, Mol Biol Evol), and the Paleobiology Database (paleobiodb.org). Each figure's own source is in its hover card and in the table below.
Fossil data from the Paleobiology Database are used under CC BY 4.0; the tree also draws on Open Tree of Life, TimeTree and NCBI Taxonomy (common names). Each CSV download begins with the same credits. Silhouettes are credited one by one beside the selection.
Every node in the built data: its parent, name, Open Tree id, age and how it was dated, living or extinct with its fossil range, described living species and the undescribed estimates (as counts, each with its source), and the published population and biomass figures with their units, where a group has one. The Download button by the selection gives the same for what you've selected, with its common ancestor.
Every published figure used
| Group | Measure | Value | Range | Confidence | Source |
|---|