New mass-spectrometric 234U-230Th ages for the Key Largo Formation confirm earlier, less precise alpha-counting measurements that place the unit within marine oxygen isotope stage 5. Ages derived from the petrographically freshest samples (>99% aragonite) of this study range from 125 to 138 ka. Despite careful sample selection and preparation, the initial 234U/238U activity ratios [denoted (234U/238U)T] range from 1.163 to 1.176 and so are all significantly higher than modern-day or inferred last-interglacial seawater. This result indicates open-system behavior of U-series nuclides, as has been observed in many other Pleistocene coral samples, and casts doubt on the accuracy of our age determinations. We have used our samples to investigate the processes and magnitudes of age perturbations.
Repeats of the same sample dissolution (analytical reproducibility) are within measured uncertainties, which are generally less than ±1 k.y. In contrast, different parts of the same coral, similarly screened for alteration, can show age differences of as much as ∼3 k.y. Evidence from this study and others suggests that alteration generally increases 234U-230Th ages. We were able to determine, by using field relationships, that the differences in 234U-230Th ages of samples lying on either side of an erosion feature are at least 5 k.y. in error; the stratigraphically younger corals above the erosion surface have 234U-230Th ages that are at least 5 k.y. too old. Notably, these apparently more perturbed samples have lower (234U/238U)T ratios than the underlying corals. This observation emphasizes that, although (234U/238U)T is a sensitive indicator of open-system behavior, the (234U/238U)T ratio does not provide a simple index of the magnitude of age disturbance.
Diagenetic additions of both 234U and 230Th to corals seem necessary to explain both elevated ages and (234U/238U)T ratios. We argue that these additions result from redistribution of nuclides during meteoric alteration of the reef in the first ∼40 k.y. after formation. Variations in the amount of 230Th added relative to 234U can be used to model the spread of ages at similar (234U/238U) ratios, although variable amounts of recent U loss may also have been important. The youngest age of this study (125 ka), just below an erosion surface, is compatible with the ages of similar features in the southern Bahamas. We suggest that 125 ka is the maximum age of a short-lived sea-level lowstand within the last-interglacial (5e) highstand.