20.–25. Sept. 2026
Münster
Europe/Berlin Zeitzone

New insights into miospores from the Late Devonian of Armenia: diversity, palaeobiogeography and machine learning applications

24.09.2026, 11:00
15m
Auditorium F5 (Fürstenberghaus)

Auditorium F5

Fürstenberghaus

Talk CIMP Palaeozoic palynology CIMP Palaeozoic palynology

Sprecher

Vitalina Lokteva (Institute of Geological Sciences of the National Academy of Sciences of the Republic of Armenia, Yerevan, Republic of Armenia)

Beschreibung

Among the few sections in Armenia exposing the Frasnian–Famennian critical interval, the Ertych is the only one known to yield abundant and diverse palynomorph assemblages. Previous studies documented a well-preserved miospore assemblage assigned to the lower torquata–gracilis Biozone, indicating a Late Frasnian age for the lower to middle part of the section. However, its diversity and palaeobiogeographic significance remain poorly constrained. Therefore, the present study focuses on the re-examination of the palynological record of the section and documents 49 miospore taxa assigned to 25 genera, increasing the known diversity compared to the 37 taxa belonging to 22 genera reported previously. The revised assemblage confirms the high diversity of Late Frasnian terrestrial vegetation preserved in the South Armenian Block. Moreover, this study assesses the palaeobiogeographic affinity of the miospore assemblage recovered from the Ertych section using hierarchical cluster analysis (UPGMA) and non-metric multidimensional scaling (NMDS).

The results reveal strong affinities with miospore assemblages from northern Gondwana and southern Laurentia, with particularly close similarities to those from Türkiye, Saudi Arabia, and North Africa, suggesting broadly comparable palaeoclimatic conditions across the northern Gondwanan margin. In addition, we explored, for the first time, the application of machine learning methods for automated miospore identification, applied as a case study to three biostratigraphically important miospore species. A two-step artificial intelligence pipeline performing automated detection and taxonomic classification of Samarisporites triangulatus, Teichertospora torquata, and Geminospora lemurata achieved approximately 98% detection precision and 88% classification accuracy, highlighting the potential of deep learning approaches for accelerating palynological research.

Autoren

Vitalina Lokteva (Institute of Geological Sciences of the National Academy of Sciences of the Republic of Armenia, Yerevan, Republic of Armenia) Tamara Hambardzumyan (Institute of Geological Sciences of the National Academy of Sciences of the Republic of Armenia, Yerevan, Republic of Armenia) Pierre Breuer (Laboratoire de Paléobotanique, Paléopalynologie et Micropaléontologie, Université de Liège,Allée) Philippe Steemans (EDDY Lab/Palaeopalynology, Department of Geology, University of Liège, Allée du 6 Août, B18 Sart Tilman, B4000 Liège, Belgium) Martin Tetard (ESNZ, Lower Hutt, New Zealand) Taniel Danelian (Univ. Lille, CNRS, UMR 8198, Evo-Eco-Paleo, 59000 Lille, France) Vahram Serobyan (Institute of Geological Sciences of the National Academy of Sciences of the Republic of Armenia, Yerevan, Republic of Armenia)

Präsentationsmaterialien

Es gibt derzeit keine Materialien.