Nyheter från en förlorad kontinent: att utforska de tidigaste tetrapodernas värld med synkrotrontomografi och maskininlärning

Tidsperiod: 2020-12-01 till 2024-11-30

Projektledare: Per Ahlberg, Per Erik Ahlberg

Finansiär: Vetenskapsrådet

Bidragstyp: Projektbidrag

Budget: 3 265 200 SEK

The origin of tetrapods is poorly understood. The oldest trackways are much older than the oldest body fossils, and while most Devonian tetrapod fossils are from Laurussia, the tetrapod lineage probably originated in South China. This confused pattern partly reflects poor biogeographical sampling of Devonian vertebrate faunas. Some areas are densely sampled, others almost blank; one of the latter is the continent of Siberia, where we know almost nothing about the non-marine vertebrate assemblages. The aim of this project is twofold: to illuminate the Late Devonian vertebrate biogeography, with emphasis on Siberia, and to develop machine learning software for segmenting (i.e. modelling the contents of) difficult tomographic data sets. The primary focus is a synchrotron tomography data set of a Late Devonian (Frasnian) bone bed from Ivanovka on the Uryup river, southern Siberia. The scans are of high quality but the bone-rock contrast is so low that conventional brightness-threshold segmentation fails. In collaboration with Thomas Schön at Uppsala University´s AI hub we will be developing new AI-based segmentation tools to solve this problem. The Ivanovka vertebrate assemblage (a fish fauna that may also contain tetrapods) will then be described and compared with other contemporary faunas, notably that from Sosnogorsk in European Russia, and biogeographical and faunal patterns will be analysed. We will also describe the enigmatic tetrapod-like fish Polyplocodus from Scotland.