Coral indicators of past sea-level change: A global repository of U-series dated benchmarks

Fossil corals provide valuable data for reconstructing past sea levels, as they are often well preserved in the fossil record and can be dated with U-series methods. Here we present a global and internally consistent database of UTh dated fossil coral sea-level indicators, including full consideration of all (known) associated uncertainties (both vertical and chronological). We include carefully determined taxon-specific depth distributions, rather than blanket depth uncertainty terms as used in most previous work. This is based on a synthesis of extensive modern ecological information on depth ranges. These ranges are found to be spatially variable (between ocean basins, between regions, and on sub-regional scales) because depth itself is not limiting – instead, depth distributions arise from complex physical, chemical, and biological interactions with coral-reef growth, distribution, and composition. One of the main causes for recognition of the greater depth-variability of coral taxa has been the routine inclusion of deep-diving and ROV surveys in coral ecological studies over the past few decades, which has broken through the “shallow-water” bias of early surveys by adding frequent observations on deeper occurrences (although more are needed). It is also clear from our assessment that coral habitat-depth distributions must be determined on the species level to reduce uncertainties in reconstructions of past sea levels, and that application to sea-level studies then requires these studies also to identify fossil corals to the species level. Samples identified only to the genus level give rise to wide uncertainties in habitat depth and, hence, sea level. Our database contains extensive metadata to assist evaluations of dating quality, as well as geomorphic and stratigraphic metadata. We demonstrate with examples how such metadata can help to evaluate sea-level reconstructions, for example by identifying outlier points. One example discusses the Last Interglacial (LIG), where we use the available data with their uncertainties to assess probabilistically the time at which local sea levels exceed that of the present, which yields a mean age of 124.6 ka with 95% probability bounds at 118.5 and 129.5 ka. We conclude with identification of key outstanding issues relating to: (i) current incomplete understanding of tectonic setting (including the current lack of independent verification of uplift/subsidence rates and reliance of somewhat unsatisfactory, and cir