Assessing amino acid racemization variability in coral intra-crystalline protein for geochronological applications

Over 500 Free Amino Acid (FAA) and corresponding Total Hydrolysed Amino Acid (THAA) analyses were completed from eight independently-dated, multi-century coral cores of massive Porites sp. colonies. This dataset allows us to re-evaluate the application of amino acid racemization (AAR) for dating late Holocene coral material, 20 years after Goodfriend et al. (GCA56 (1992), 3847) first showed AAR had promise for developing chronologies in coral cores. This re-assessment incorporates recent method improvements, including measurement by RP-HPLC, new quality control approaches (e.g. sampling and sub-sampling protocols, statistically-based data screening criteria), and cleaning steps to isolate the intra-crystalline skeletal protein. We show that the removal of the extra-crystalline contaminants and matrix protein is the most critical step for reproducible results and recommend a protocol of bleaching samples in NaOCl for 48 h to maximise removal of open system proteins while minimising the induced racemization. We demonstrate that AAR follows closed system behaviour in the intra-crystalline fraction of the coral skeletal proteins. Our study is the first to assess the natural variability in intra-crystalline AAR between colonies, and we use coral cores taken from the Great Barrier Reef, Australia, and Jarvis Island in the equatorial Pacific to explore variability associated with different environmental conditions and thermal histories. Chronologies were developed from THAA Asx D/L, Ala D/L, Glx D/L and FAA Asx D/L for each core and least squares Monte Carlo modelling applied in order to quantify uncertainty of AAR age determinations and assess the level of dating resolution possible over the last 5 centuries. AAR within colonies follow consistent stratigraphic aging. However, there are systematic differences in rates between the colonies, which would preclude direct comparison from one colony to another for accurate age estimation. When AAR age models are developed from a combined dataset to include this natural inter-colony variability THAA Asx D/L, Glx D/L and Ala D/L give a 2σ age uncertainty of ±19, ±38 and ±29 year, for the 20th C respectively; in comparison 2σ age uncertainties from a single colony are ±12, ±12 and ±14 year. This is the first demonstration of FAA D/L for dating coral and following strict protocols 2σ precisions of ±24 years can be achieved across different colonies in samples from the last 150 years, and can be ±10 years within a core from a single colony. Despite these relatively large error estimates, AAR would be a valuable tool in situations where a large number of samples need to be screened rapidly and cheaply (e.g. identifying material from mixed populations in beach or uplift deposits), prior to and complementing the more time-consuming geochronological tools of U/Th or seasonal isotopic timeseries.