Selection and stability of quantitative stratigraphic age models: Plio-Pleistocene glaciomarine sediments in the ANDRILL 1B drillcore, McMurdo Ice Shelf

Interpretation of glacimarine sedimentary records from Antarctic shelf drillholes has been hampered by the ambiguous age of strata where erosional unconformities and coarse diamictite deposits truncate or omit the magnetostratigraphic and biostratigraphic units used for correlation. However, new quantitative biostratigraphic techniques enable the correlation of sparse, incomplete, and often reworked Plio-Pleistocene records of Ross Sea fossil diatom flora with the more extensively documented but potentially diachronous offshore history of species' first and last appearance datums (FADs and LADs). The approach uses comprehensive regional databases of fossil records and computer-automated search algorithms to find the multidimensional line of correlation (LOC) that best fits local observations, and to identify regionally isochronous biostratigraphic markers. Different model configurations can be used to produce LOCs that represent alternative working hypotheses regarding reworking and other sources of misfit in the biostratigraphic record, and that together provide an envelope of uncertainty for age interpretation.An integrated, quantitative chronostratigraphic model for the ANDRILL-1B drillcore was developed iteratively: an initial age model was constructed solely from preliminary on-ice observations of fossil diatom highest and lowest occurrences (HOs and LOs) and their correlation with a database of other local event records from 24 DVDP, CIROS, and IODP drillcore sections. The model was subsequently updated as off-ice work yielded additional biostratigraphic marker events and revised event horizons, Ar/Ar ages for volcanic material, better-constrained magnetostratigraphic interpretation, and refinements to computational/analytical methodology. The current quantitative biostratigraphic age model for the AND-1B hole integrates the local ranges of 29 diatom taxa, five dated volcanic samples, and independently constrained ages of five paleomagnetic reversals. During analysis, three model families were compared, each of which make different assumptions about the scale and influence of reworking and caving on observed biostratigraphic ranges, varying from assumption of no reworking/caving to widespread reworking/caving. Of these, the Hybrid Range Model, which allows for intermediate levels of reworking expected in a glaciomarine depositional environment, was found to give the most robust and reliable single LOC, even with immature input biostratigraphic data. However, alternative models used together provide an envelope of uncertainty around this preferred LOC. Results corroborate almost all of the on-ice geomagnetic polarity reversal age interpretations, but identify a previously unrecognized disconformity (~ 800 kyr hiatus) near 440 mbsf.In spite of substantial augmentation and progressive refinement of the AND-1B input dataset, successive versions of the output LOC retain the same fundamental structure and shape with only relatively minor, fine-scale differences. The age model's remarkable stability indicates that quantitative biostratigraphic analysis is capable of constructing robust and reliable regional correlation schemes and local section chronologies, even in cases of discontinuous stratigraphy and with relatively immature or incomplete datasets.