HIGH PRECISION 40Ar/39Ar DATING OF PLIOCENE TURKANA TUFFS

The Omo-Turkana Basin in Kenya and Ethiopia is renowned for fossil-rich, Plio-Pleistocene sedimentary sequences that include important hominin specimens. Interbedded volcanic tuffs provide critical time constraints, mostly from ⁴⁰Ar/³⁹Ar dating of K-feldspar crystals from pumice clasts. Unfortunately, most previous age results have limited precision (~20-60 ka), inhibiting refinement of fossil ages and the testing of paleoclimate proxy models.

Recent high precision ⁴⁰Ar/³⁹Ar analyses of pumice feldspars from Turkana tuffs reveals an array of ages from individual pumice clasts, a feature not apparent from prior work. Here, tuff eruption times were determined using a Bayesian estimation approach, which assumes that the youngest ages best reflect the time of eruption. This statistical method yields high-precision ages with uncertainties typically ±2-3 ka, although it is unclear whether the approach provides the most accurate measure of tuff eruption.

In this study, we analysed pumice feldspar crystals from several well-known Pliocene Turkana tuffs (including Naibar, Topernawi, Moiti, Lokochot, Tulu Bor, Toroto, Ninikaa, Burgi tuffs) to improve age resolution, evaluate neutron flux gradient effects, and assess the robustness of Bayesian age estimates. Our analyses reveal small fluence gradients across single sample packets (~0.05 – 0.2%); for perspective, these gradients are similar to uncertainties for individual analyses. Nonetheless, it is likely that the Bayesian method favours ‘younger’ feldspar grains that receive lower neutron doses, leading to an underestimation of eruption ages. We suggest that more accurate ages are provided by the youngest subset of results that constitute >20% of data-points and conform to a normal distribution This method gives marginally older eruption age estimates (1.6 – 6.9 ka), although all results are within uncertainty of previous estimates. The new data permit resolution of tuff ages that could not be distinguished previously. For example, we calculate an age of 3.9976 ± 0.0008 Ma (2s) for the Topernawi Tuff, distinct from a value of 3.9818 ± 0.0015 Ma (2s) for the Moiti Tuff.