Romania · 16 to 2 million years ago
When Romania was a sea
The Carpathian foothills, the Transylvanian Basin and the Danube plain were once under water. Scroll back 16 million years, then forward as the water shrinks. Find your commune at the end.
Scroll to go back in time.
Romania · 16 to 2 million years ago
Was your commune under the sea?
Type the name of a commune, town or city. You get the stages in which it lay inside the drawn water, and the nearest dated fossil site with what was found there.
How the water was drawn
The shape of the water is a reconstruction. Geologists draw it from rocks and fossils, and every published map is a sketch. It is drawn here on today's relief, without flooding the elevation model: the Carpathians were lower and the basins have moved since, so a flat sea level on today's map would be wrong.
Three kinds of line, with the error stated for each.
Dacian Basin, Black Sea shore of the Maeotian map: traced from the printed maps of Jipa and Olariu (2009) by colour, and placed on the cities and rivers printed in those figures (placement error 1–7 km). Error of the shore: 25 km for the Dacian Basin, 40 km for the Black Sea and for the edge of Lake Pannon, because the source maps are simplified.
Transylvanian basin, western plain, Moldavian foredeep, Dobrogea: no published figure could be placed on the map to commune scale. These are polygons drawn by hand after the basin sketches in Palcu et al. (2023), cut to today's low ground. Error 60 km. This is the weakest layer.
Ages are ranges from the published charts. The chronologies in Jipa and Olariu differ by up to about a million years for the Pontian and Dacian stages; the page gives round ranges.
Do the drawn seas agree with the fossils?
A test was written down before any outline existed (the file is linked below). The Paleobiology Database holds 946 Romanian collections, all dated. 439 had coordinates to the arc-minute or better, an age interval under four million years and a clear setting. Across the six stages that gives 60 checks of a water site and 1 of a land site (one site can count in more than one stage). The rule: at least 80% of the water sites inside the outline widened by its stated error, and at least 60% of land sites outside it.
| Stage | Water sites inside the plain outline | All collections, any age or setting, inside the plain outline | Water sites inside the widened outline | All collections inside the widened outline | Result on the rule as written |
|---|---|---|---|---|---|
| Badenian | 10 / 10 | 35% | 10 / 10 | 95% | passes the written rule |
| Sarmatian | 24 / 31 | 48% | 31 / 31 | 99% | passes the written rule |
| Maeotian and Pannonian | – | 9% | – | 47% | too few sites |
| Early Pontian | 2 / 5 | 6% | 3 / 5 | 25% | fails |
| Late Pontian to Dacian | 2 / 2 | 9% | 2 / 2 | 27% | too few sites |
| Romanian | 0 / 12 | 0% | 0 / 12 | 9% | fails |
The Badenian and Sarmatian outlines pass the rule as written, but the rule is weak: the widened outline holds 95% of all precise collections of any age or setting, so almost any site would pass. The plain outlines are the sharper check, and the column beside them is what chance would give: Badenian 10 of 10 water sites inside, against 35% of all collections; Sarmatian 24 of 31 (77%, under the 80% bar), against 48%. That is modest support, not a match. The Early Pontian outline fails (3 of 5). The two misses are sites the database dates to the Messinian and codes as marine, in Transylvania and the west; their shells (Mactra, Cardium, Pirenella) and, for one, the source paper on the Tortonian–Sarmatian boundary suggest that the database date may be wrong, which has not been checked. The Romanian stage fails (12 sites, none inside): the registered rule counts lake deposits as water, and the drawing has no lakes at that stage. The Maeotian and Late Pontian–Dacian stages have too few sites, and land sites are too few in every stage to test the second condition.
Known misses, found afterwards and not corrected: Praid and Cacica, both on or beside Badenian salt, lie outside the drawn Transylvanian and foredeep outlines, because the cut to low ground removed part of those basin floors.
The source maps were drawn partly from the same fossil literature that feeds the database, so agreement is expected and is not independent confirmation. Fossil sites are also found more often where rivers cut rocks, so they say where water was, not how much. A site's age is an interval, and a site counts for a stage only if its interval contains the stage's age.
The pre-registration, with its dated addenda and the list of every run
When did the Danube cut the Iron Gates?
The papers disagree, and the page does not pick one.
- About 6–5.3 million years ago: Clauzon and colleagues (2005) interpreted an erosional surface near the gorge's outlet as Messinian and proposed an early Danube. Jipa and colleagues (2011) challenged the identification and age of the delta deposits it rests on.
- Pliocene or Pleistocene: ter Borgh (2013) places the Danube's present course there, and finds a separate late Miocene link between the basins.
- About 2.0–1.8 million years ago: Andreescu (2009) proposes that the old Danube entered the Dacian Basin through the Iron Gates then.
Limits
- This is a reconstruction. Ages are ranges and shores are lines with a stated error of 25 to 60 km, so a commune near a shore may have been dry or wet.
- Lake Pannon is drawn only at 9 million years; it probably lasted longer in the west, but no map of it could be placed on today's map.
- The Early Pontian panel is partly hidden by the next panel of its source figure, so its outline is cut. The source also shows the basin's area widening southward from the Early Pontian to the Late Pontian (the inset of Fig 2.11) while it fills with sediment.
- The dates are approximate intervals represented by the maps, not the full boundaries of each stage; published regional chronologies differ, and the six stages are snapshots, not a continuous record.
- No depth is shown or claimed: no published source gives the depth of the water at a given place.
- Not every stage was a sea. Lake Pannon was a lake, the Dacian Basin turned fresh, and the Romanian stage was a plain.
- For the Badenian, the southern part of the water (the Dacian area) is the Sarmatian outline reused; no Badenian map of it was found that could be placed on today's map.
- The Black Sea after the Maeotian map is reused for the later stages and the two earlier ones east of 27.9°E; it is not redrawn for each.
- The Danube below the Iron Gates is drawn from the country border to Silistra and by straight links from there to the delta; it is a guide, not a survey.
- Fossil sites are the ones in the Paleobiology Database. Many Romanian sites are not in it, and a dot is a collection, not a species range.
Sources
- Jipa, D.C. & Olariu, C. (2009), Dacian Basin: depositional architecture and sedimentary history of a Paratethys sea, GeoEcoMar Special Publication 3 (Zenodo 252418): Figs 2.5, 2.7, 2.8, 2.11, and the stage chronology of Fig 1.5.
- Palcu, D.V. et al. (2023), The legacy of the Tethys Ocean, Earth-Science Reviews 246, 104594 (CC BY 4.0): Figs 12–13, and the Badenian, Sarmatian and Lake Pannon history.
- Fossils: Paleobiology Database, collections and occurrences with country code RO, downloaded 14 November 2026 (CC0).
- Relief: Copernicus GLO-30 DEM, averaged to 0.6 km. © DLR e.V. 2010-2014 and © Airbus Defence and Space GmbH 2014-2018 provided under COPERNICUS by the European Union and ESA; all rights reserved. Communes: UAT outlines 2025 (ANCPI), one point per commune. Borders: Natural Earth.
- Iron Gates: Clauzon et al. 2005, Basin Research 17; ter Borgh 2013, PhD thesis, Utrecht; Andreescu 2009, Oltenia: Studii și comunicări, Științele Naturii 25; Jipa et al. 2011, Geo-Eco-Marina 17.