Life on the moving map

Every fossil has two addresses. One is where it was dug up: a quarry in Montana, a sea cliff in Dorset, a dry riverbed in Kenya. The other is where the animal actually lived, and for anything more than a few million years old that was somewhere else. Continents creep a few centimetres a year, and over hundreds of millions of years that adds up to thousands of kilometres. The fossils found in Britain tell the story: its Silurian sea creatures lived around 23 degrees south of the equator, its coal-swamp animals of 300 to 320 million years ago lived on it (the median collection lies within a degree of it), and its early dinosaurs and ichthyosaurs at about 38 degrees north, the latitude of the Mediterranean today.

The Paleobiology Database, a huge shared catalogue built over decades by hundreds of palaeontologists, records both addresses. For each collection of fossils it gives the present-day place and, using the PALEOMAP plate model of Christopher Scotese, the position on the ancient Earth. I took the records of 24 animal groups that people ask about, from trilobites and sea scorpions through the dinosaurs to whales, elephants and our own family, and put them on a map of the moving continents.

01 · The mapEarth through time

Press play to run from 540 million years ago to today, drag the slider or the timeline, or jump to a key moment. Each frame shows the fossils dated within 5 million years of the moment shown (by the midpoint of each collection's age range); at the five mass extinctions it shows the 10 million years before. Click a dot to see what was found there, when, and where that patch of ancient ground is today. The layers panel switches every dot between its ancient position and today's, turns groups on and off, and changes the background.

02 · PangaeaWhen nearly all the land was one

For about the first half of this story the continents were scattered: Gondwana, a giant made of today's South America, Africa, Antarctica, Australia and India, lay across the south, with smaller continents such as Laurentia (North America) and Baltica (northern Europe) around the tropics. From about 335 million years ago they began to collide, and Pangaea came together in stages through the Carboniferous and Permian. To put a number on it I merged the PALEOMAP blocks at each age into joined land masses (closing gaps narrower than about a third of a degree) and measured the largest land mass as a share of all the land.

By 270 million years ago 88% of the land was one block, and it stayed above 85% for 90 million years, peaking at 90.2% around 180 million years ago, in the Early Jurassic. Then it came apart: 72% at 170 million years ago, 50% at 150 as the central Atlantic opened, and 23% by 100 million years ago, with the South Atlantic open between South America and Africa. Today the biggest block, Africa joined to Eurasia, holds 54%.

One caution about the outlines. They are today's coastlines and continental blocks carried back in time by the plate model, not ancient shorelines. Sea level rose and fell by hundreds of metres, and at times shallow seas flooded much of the continents (a seaway split North America in two for much of the Cretaceous), so some dots that sit on 'land' here were in fact on a sea floor. Many of the oldest fossils, trilobites and ammonites above all, are sea creatures for exactly that reason.

03 · Who ruled whenThe record, group by group

Counting fossils is not counting animals. A group with hard shells that lived in shallow seas, like ammonites, fossilises far more readily than one that lived in forests. So the chart below asks a simpler question for each group: when is it most diverse in the record? It counts the distinct genera found in each 5 million years and scales each group to its own peak, so a pale lane is a quiet time for that group, not necessarily a rare animal.

By that measure trilobites were the dominant group of the early seas: they make up 57% of all genera from the Paleozoic era in these groups, peaking between 495 and 490 million years ago. In the Mesozoic the largest share is not the dinosaurs but the insects (42%), and in the Cenozoic the mammals (39%). Insects are absurdly diverse (12,832 genera in this data), and their record is lumpy: a quarter of all insect finds (25%) come from just 50 of 5,580 collections, rare deposits where delicate bodies were preserved. Ammonites, with 2,790 genera, are the other great presence of the Mesozoic seas, until the very end of the Cretaceous.

04 · DinosaursFrom one continent to many

The data hold 20,231 occurrences of dinosaurs other than birds, in 1,588 genera: 693 genera of theropods (the meat-eaters and their kin), 380 of sauropods and their relatives (the long-necked plant-eaters) and 471 of ornithischians (the horned, armoured and duck-billed plant-eaters), plus 1,596 finds identified only as 'dinosaur'. The oldest records are from the Triassic, and the busiest 5 million years in the record is 80 to 75 million years ago, with 316 genera, 75% of their finds from North America, where rock of that age is unusually well exposed and well searched.

Dinosaurs appeared on Pangaea and spent their first 50 million years or so on one connected land mass, then watched it break apart. That shows in where their genera turn up. Of the 203 genera from the Late Triassic to the Early Jurassic, 11.3% (about 1 in 9) are found on more than one of today's continents. Of the 1,180 Cretaceous genera only 5.1% are (1 in 20): the separating continents grew their own faunas. Every tyrannosaur and horned dinosaur (ceratopsid) in the data is from North America, Asia or Europe, while 88% of abelisaur finds are from the southern continents and India.

Caution is needed. More genera are named from the Cretaceous, palaeontologists in different countries sometimes give similar animals different names, and a genus 'on two continents' may be two closely related animals. But the drop from about 1 in 9 to 1 in 20 is the direction you would expect as an ocean opens between populations. Dinosaurs also lived much nearer the poles than the word 'reptile' suggests: 245 dinosaur collections lay more than 60 degrees from the equator, and 53 more than 75 degrees, most of them in Alaska (23) and in Victoria, Australia (17), places that had weeks or months of winter darkness.

05 · LatitudeTropics, poles and drifting ground

Because the palaeo-positions are known, we can ask how far from the equator each group lived. The chart shows the middle 80% of each group's collections by palaeo-latitude, every 10 million years. Trilobite collections lay a median of 20 degrees from the equator; turtles 41; insects 43. Hominids sit at 8 degrees: 88% of their record is from Africa and Asia, most of it in the tropics. Much of this is the shape of the record rather than of life: most of today's fossil hunting is in the northern mid-latitudes, so recent groups look northern too.

06 · Five extinctionsWhat was lost, in these groups

Five times in the last half-billion years a large part of life vanished in a geological moment. A crude way to see them in this data: take the genera known from at least two collections that were around in the 10 million years before each extinction, and ask how many are never found again. At an ordinary 10 million year step the median is 37% (genera come and go, and many are rarely found), with 1 step in 10 above 63%. The end-Permian, the 'Great Dying', stands out at 74%. At the end of the Cretaceous every genus of non-bird dinosaur, plesiosaur and mosasaur in the data is last seen before the line, and 97% of ammonites, while insects (1.5%) sail through.

The end-Triassic extinction barely registers here (33%), even though it is one of the five: much of what it killed (reef animals, many marine reptiles and the crocodile-line archosaurs that rivalled early dinosaurs) is outside these groups or thinly recorded. The groups chosen and the patchy record shape every number on this page.

07 · The gapsA map of where palaeontologists dig

The fossil record is not a random sample of ancient life. It depends on which rocks of each age survive at the surface, and on where people have looked. North America and Europe hold 32% of the world's land but 66% of the finds here; the United States alone has 28%. Africa has 8.5% and South America 6.5%. The ancient continents look busiest where the modern ones have the most universities, museums and roads cut through rock. Dots are records, not abundance: an empty ocean or continent on the map usually means no rock, or no one looking, rather than no life.

A dot on this map is a place where rock of the right age reached the surface and someone went to look.

08 · How I did thisMethods and caveats

Fossils. I downloaded every occurrence in the Paleobiology Database for 23 named groups through its data service (480,977 rows), asking for palaeo-coordinates from the PALEOMAP model at each collection's midpoint age. The groups overlap (the dinosaurs include the birds, the mammals include the whales), so every occurrence is kept once, in its most specific group (47,354 repeats removed). Three names reach further than meant and were corrected: 'Proboscidea' also matches an insect name (5,679 insect records dropped from the elephants, which the insect group already holds); the database places frogs, salamanders and caecilians inside Temnospondyli, so I split these 4,301 records off as their own group; and 1,138 records of Mesozoic and living birds that came back under 'theropods, not birds' were moved to the birds. 5 non-bird dinosaur records dated after the end-Cretaceous extinction were left out.

Ages and positions. Each occurrence takes the midpoint of its collection's age range. Collections dated only to within more than 25 million years (3,155 occurrences) are left out of the analysis. PALEOMAP cannot place some collections; for the 37,781 occurrences younger than 1 million years I used the present position (plates move a few tens of kilometres at most in that time), and 14,889 older occurrences without a palaeo-position are left out. That leaves 409,895 occurrences in 98,147 collections from 211 present-day countries. On the map the fossils of each frame are binned by group into 1 degree squares of palaeo-latitude and longitude. The positions are for each collection's own age, which can differ from the frame's age by up to 5 million years, so dots can sit a little off the outline of that frame. Present-day places come from Natural Earth countries and provinces.

Diversity. Because ages are the midpoints of ranges that often span a whole geological stage, the 5 million year bins fill unevenly, and one, 225 to 230 million years ago, holds no midpoints at all; it shows as a gap in the charts. Genera are counted as distinct genus names found in each 5 million years, with no correction for how much rock or effort each interval has. For the extinctions, a genus is 'last seen' before a boundary if no collection younger than the boundary has it; one-off finds are left out. These are simple counts, chosen to be transparent rather than to give the best estimate of true diversity, which needs methods that correct for sampling.

What this is not. It is not a census of ancient life. The record is dominated by animals with hard parts, by marine sediments, and by the places and ages where rock is exposed and has been studied. The coastlines are present-day outlines moved by a plate model, which is itself a reconstruction with uncertainty that grows with age. The time scale follows the International Commission on Stratigraphy, rounded.