About this lab
Five topics, one thread: land and sea change for more than one reason, and the biggest is the slow movement of lithospheric plates across Earth’s surface.
Falling and rising sea level can make a land bridge appear and vanish while the two landmasses stay where they are. Plate motion is a different thing: continents drift with their plates, sea floor is born at ridges and lost at trenches, and earthquakes and volcanoes cluster on plate boundaries.
The continents in the globe topics are real: they are run back from today’s positions to 250 million years ago with the 2019 global plate model of Müller and colleagues. The sea-floor colours are today’s measured ages; crust younger than the chosen time did not exist yet, so it is erased and the dark “sea of that time, gone today” shows through.
Five conclusions
Five conclusions
- Sea level and land bridges: the seven survey points are +20, −20, −50, −80, −50, −20 and +20 m. At −60 m the −80 m trough still separates the shores; at −80 m or lower all seven points are exposed and the bridge is complete. When the water rises the bridge disappears, and the two landmasses never moved.
- Evidence for continental drift: a matching outline is only a clue. Run the continents back and the areas with Glossopteris, Mesosaurus, Lystrosaurus and glacial traces on the southern continents draw together. Each has its own age, rock layers and alternative explanations to test, and no single kind of evidence proves the continents were once joined.
- Spreading rate: new crust grows evenly outward from the ridge, older with distance from the axis. 60 km in 3 million years gives a half rate of 20 km/Ma = 2 cm/yr, and the two sides separate at 4 cm/yr. Sea-floor ages in the Atlantic give rates of the same order as those from the PB2002 plate Euler poles.
- Three kinds of boundary: at a spreading boundary new crust forms in the middle; an oceanic slab dives under a continent to make a trench, a volcanic arc and a deep earthquake zone; where two continents collide the crust thickens and rises; at a transform boundary the sides slide past each other and are only offset, with nothing created or lost.
- Hazard belts: from 1976 to 2025 the great majority of magnitude 6+ earthquakes lie near a plate boundary, and earthquakes deeper than 70 km go with subduction or collision zones, their epicentres shifted inland from the boundary by the dip of the slab. Earthquakes also occur inside plates; the lab keeps those events instead of forcing them onto a boundary. This is a statistical pattern and predicts no particular day or magnitude.
Model and sources
Model and sources
- Reconstruction: each 0.5° cell belongs to a static plate polygon of the 2019 Müller model and is turned to its place at the chosen time by that plate’s finite rotation relative to Africa (one set every 5 million years, interpolated on the sphere between them). Land shapes come from the Natural Earth 10 m land mask and ground colours from today’s NASA imagery, so the shapes are today’s coasts, not the coasts of that time; ancient shorelines, shallow seas and mountains are not drawn. Holding Africa fixed is a choice of reference, not a claim that Africa does not move.
- Sea-floor age: the 2020 global grid of Seton and colleagues (GTS2012 time scale), averaged onto a 0.5° grid. Crust older than the chosen time turns with its plate; sea floor that has already been subducted cannot be recovered and is drawn dark.
- Present-day velocity: the PB2002 Euler poles of each plate relative to the Pacific plate; relative velocity of two plates = the difference of their angular-velocity vectors crossed with the position vector. The second source of spreading rates is the sea-floor age profile: sample along a parallel on both sides of the ridge and fit, by least squares, the points older than 3 and no older than 20 million years. That gives a mean rate over roughly the last 20 million years, which need not equal today’s instantaneous velocity.
- Evidence circles: each centre is the approximate position of a fossil or glacial-deposit area; the 6° radius only means “found in this region”, not a find-spot boundary, and does not correspond to the age of a particular layer. Mesosaurus is about 275 million years old, beyond the model’s 250 million, so the 250 Ma positions are only an illustration. The shore marks Recife and Lagos are illustrative, not control points from any study.
- Strait and cross-sections: the seven elevations come from the report’s original simplified strait, the 5 km spacing is a teaching value, and the strait view is shrunk 40 times horizontally, equivalent to 40 times vertical exaggeration. The boundary cross-sections are schematic: relief of the sea floor and rift is exaggerated about 5 times, crust thickness about 2.9 times (7 km drawn as 20 km), the sea depth in the ridge section is compressed to about 0.4 and the lithosphere and asthenosphere are not exaggerated; in the subduction and collision sections the sea depth and the land above sea level are exaggerated about 5 times, as is the land relief in the transform section, and volcano heights are only illustrative, the 45° dip, lithosphere thicknesses of 70 and 100 km and continental crust of 35 km are teaching values; thickening in a collision follows conservation of area and uplift follows Airy isostasy (densities 2800 and 3300 kg/m³). It is not a measurement of the Himalaya and gives no collision age. Each section shows only a short stretch of asthenosphere below the lithosphere (36 km deep for the ridge, 115 km for subduction, 68 km for collision, 62 km for the transform); the mantle below is not drawn. The surface image of the transform section is taken from NASA Blue Marble over California, rotated, and is only an illustration, not the real location of any fault. The ridge stripes also show two short normal-polarity events (Jaramillo 0.99–1.07 Ma, Olduvai 1.78–1.95 Ma, rounded values, GTS2012, Ogg 2012), for display only. The asthenosphere is solid mantle that can flow slowly, not a pool of magma.
- Earthquake catalogue: USGS ComCat, 1976-01-01 to 2025-12-31, magnitude 6 or more, 7030 events; each event is tagged with the type of and distance to the nearest PB2002 boundary (measured to segment midpoints). Events deeper than 70 km are compared only with subduction and collision zones; this is a model assumption (deep earthquakes need a subducting slab), tested by requiring the epicentre to lie within 200 km plus the focal depth of the boundary, which about 95% of deep events satisfy.
The Earth imagery is from NASA.
- Müller, R. D. et al. · 2019 · Global plate model with lithospheric deformation (Tectonics 38) ↗ CC BY-SA 4.0; plate rotations and static plate polygons; the rotation table here is derived data, shared under the same CC BY-SA 4.0; downloaded and checked 2026-11-15
- Seton, M. et al. · 2020 · Global present-day oceanic crustal age grid (G-cubed 21) ↗ EarthByte open data, no licence terms given in the directory; 6-arc-minute grid; downloaded and checked 2026-11-15
- Bird, P. · 2003 · PB2002 plate boundaries and Euler poles (G-cubed 4) ↗ Author’s public data (the Ahlenius conversion is released under ODC-By); boundary types and present-day velocity
- U.S. Geological Survey · ComCat earthquake catalogue API ↗ US government work, public domain; magnitude 6+ earthquakes; downloaded 2026-11-15
- USGS · This Dynamic Earth · early evidence for continental drift ↗ Glossopteris and glacial deposits
- Kelso, N. V. and Patterson, T. · Natural Earth · 10 m land ↗ Public domain; land mask; downloaded 2026-11-15
- NASA Earth Observatory · Blue Marble: Next Generation ↗ Daytime surface imagery