Abstract :
[en] The Late Devonian was affected by major, irreversible changes including two of the most severe
biodiversity crises in Earth’s history, the so-called Kellwasser and Hangenberg Events
(respectively near the Frasnian – Famennian (F-F) and the Devonian – Carboniferous (D-C)
boundaries). Currently, hypotheses for the Late Devonian extinctions include sea-level
fluctuations and regression, climate cooling, ocean anoxia, massive volcanism and/or bolide
impact. Unfortunately, testing these hypotheses is impaired by a lack of sufficient temporal
resolution in paleobiological, tectonic and proxy climate records. Recent advances in
astronomical calibration have improved the accuracy of the Frasnian time scale and part of the
Famennian. However, the time duration of the Famennian Stage remains poorly constrained.
During the Late Devonian, an epieric sea in North-America mid-continent occupied the Illinois
Basin where a complete Late Frasnian – Early Carboniferous succession of deep-shelf deposits
was archived. A record of this sequence is captured in three overlapping cores (H-30, Sullivan
Slough and H-32). The H-30 core section spans the F-F boundary; the Sullivan Slough section
spans almost all of the Famennian and the H-32 section sampled spans the D-C boundary. To
have the best chance of capturing Milankovitch cycles, 2200 rock samples were collected at
minimum 5-cm-interval across the entire sequence. Magnetic susceptibility (MS) was measured
on each sample and the preservation of climatic information into the signal was verified through
geochemical analyses. To estimate the duration of the Famennian stage, we applied multiple
spectral techniques and tuned the MS signal using the highly stable 405 k.y. cycle for Sullivan
Slough and the obliquity cycle (34.4 k.y.; Waltham, 2015) for the H-30 and H-32 cores. Based
on the correlation between the cores and the tuning, we constructed a Famennian astronomical
time scale, which indicates a duration of 13.3 m.y. An uncertainty of ± 0.5 m.y. was calculated to
assess the errors raising from the stratigraphic position of the F-F and D-C boundaries, and the
405 k.y. cycle counting error. Our estimate duration is very close to the GTS-2012 duration
(Becker, 2012) interpolated from the high-resolution (U/Pb) radiometric ages available for the
uppermost Devonian.