Coral geochemical records: An overview of their use as climate proxies and of available databases

The following was contributed by Dr. Allison Lawman, June, 2023:

Overview

The tropical oceans are powerful drivers of temperature and hydroclimate patterns worldwide. Paleoclimate records from shallow-water, reef-building corals (Figure 1) extend the instrumental record and provide decades to centuries of high-resolution (annual to sub-annual) temperature and hydroclimate information from the tropical to subtropical oceans (~30°S-30°N). Coral geochemical proxies are widely used to reconstruct past changes in tropical climate variability and investigate its response to natural and anthropogenic forcings. Prominent modes of tropical climate variability include the El Niño-Southern Oscillation (ENSO), tropical Pacific decadal variability, the Indian Ocean Dipole, and tropical variability in the Atlantic.

As a coral grows, tiny coral polyps secrete a skeleton made of aragonite, a crystal form of calcium carbonate (CaCO₃). Massive corals can grow several meters in height and live for hundreds of years, providing continuous high-resolution records of environmental conditions. Corals have annual density banding that is visible when thin ~5-7 mm slices of coral skeleton are x-ray imaged perpendicular to the growth axis (Figure 2). The annual density banding along with radiometric dating techniques (Cheng et al., 2013; Shen et al., 2012) provide well-dated chronologies with age uncertainties typically on the order of a few years or less. Fast growth rates of ~0.3-2 cm/year (Thompson, 2021) and precise micro-milling techniques permit annual to sub-annual sampling along the coral’s growth axis.

Changes in the chemical composition of a coral’s skeleton reflect oceanic conditions at the time of growth. Strontium-to-calcium ratios (Sr/Ca) and the oxygen isotopic composition (δ¹⁸O) of coral skeletal aragonite are well-established climate proxies (Fairbanks et al., 1997; Lough, 2010). Sea surface temperature (SST) exerts the dominant climate control on coral Sr/Ca (Beck et al., 1992), whereas coral δ¹⁸O is jointly influenced by SST and the oxygen isotopic composition of seawater (δ¹⁸O_sw) (Weber and Woodhead, 1972). δ¹⁸O_sw is impacted by similar processes as sea surface salinity (e.g., rainfall, evaporation, advection of different water masses, and freshwater runoff; LeGrande and Schmidt, 2006). Hydroclimate signals from coral-inferred δ¹⁸O_sw values are calculated using paired coral Sr/Ca and δ¹⁸O measurements to remove the SST contribution from coral δ¹⁸O.

Together, modern (living when collected) and fossil (dead when collected; Figure 1) corals provide climate information from the tropical oceans, a region largely underrepresented among other high-resolution paleoclimate records from tree rings and ice cores.

Key Strengths

• Corals provide climate information from underrepresented tropical regions.

• Annual to sub-annual resolution.

• Well-dated chronologies.

• Paired Sr/Ca and δ¹⁸O measurements jointly provide temperature and hydroclimate information.

Key Limitations

• Most fossil coral records are short (<100 years) and do not extend to the 20th or 21st centuries.

• Inter-coral colony replication is limited.

• Sr/Ca-SST calibration methods and equations are not standardized and often site-specific.

• Most published fossil coral records do not extend to time intervals prior to the Holocene. That said, recent advancements in coral drilling provide the opportunity to recover older samples.

Types of Uncertainties in Coral Proxy Records

• Analytical uncertainty

• Chronological uncertainty (layer-counting errors and radiometric dating errors)

• Calibration uncertainty

• “Geological” uncertainty due to uncertainties about the oceanographic setting (for fossil corals) or sampling

Publicly Available Coral Data Compilations

Many coral proxy records are available from public repositories such as the NOAA National Centers for Environmental Information World Data Service for Paleoclimatology (WDS Paleo) (https://www.ncei.noaa.gov/access/paleo-search/?dataTypeId=4) and the World Data Center PANGAEA data publisher (https://www.pangaea.de/). These repositories enable users to search for coral records by investigator, location, or time interval.

The PAGES CoralHydro2k database (Walter et al., 2023) is a comprehensive, community-sourced compilation of peer-reviewed coral records from the Common Era (CE). The database is available for download from the WDS Paleo (https://www.ncei.noaa.gov/metadata/geoportal/rest/metadata/item/noaa-coral-35453/html). Version 1 of the database includes 54 paired Sr/Ca-δ¹⁸O records, 79 unpaired δ¹⁸O records, and 29 unpaired Sr/Ca records that span the global tropics (Figure 3). This dataset is available in the Linked Paleo Data (LiPD) standardized format and includes serializations in Python, MATLAB, and R. Example scripts that demonstrate how to query the database are provided on the CoralHydro2k GitHub repository (https://github.com/CoralHydro2k/).

The CoralHydro2k database builds upon the work of PAGES Ocean2k (Tierney et al., 2015) and PAGES Iso2k (Konecky et al., 2020). Many coral records will appear in multiple PAGES2k databases. A notable difference is that the CoralHydro2k database includes short coral records (minimum 10 years in length) that did not meet the length criteria for other PAGES2k databases. All records in the CoralHydro2k database are quality control checked, machine-readable, and include standardized metadata fields.

Fossil coral records from time periods prior to the Common Era are rarer but nonetheless provide important insight about Holocene tropical climate variability. Numerous studies (e.g., Cobb et al., 2013, Emile-Geay et al., 2015, Grothe et al., 2019, Lawman et al., 2022) use Holocene coral records to understand the response of ENSO to orbital and anthropogenic forcing during the Holocene. Emile-Geay et al. (2015) compiled ~2000 years of published coral and mollusk δ¹⁸O data from ENSO-sensitive locations across the western, central, and eastern Pacific. The compiled records are available as a MATLAB file from GitHub (https://github.com/CommonClimate/EmileGeay_NatGeo2015/tree/master/data/obs). All records have annual or finer resolution.

Grothe et al. (2019) present an updated modern and fossil coral δ¹⁸O data compilation from the Line Islands in the central equatorial Pacific. This synthesis includes over 1700 years of coral data from multiple studies. Collectively, this data compilation covers intervals from ~7,000 years ago through 2016 CE. The average length of an individual fossil coral record is approximately 30 years and provides discrete “snapshots” of climate variability through time. The modern and fossil composite MATLAB data files from Grothe et al. (2019) are available from the WDS Paleo (https://www.ncei.noaa.gov/access/paleo-search/study/22415).

Additional Details

Coral Age Models

The layer-counted chronologies for modern corals are assigned ages based on the date of collection. Radiometric dates provide chronological tie points for fossil corals. For corals sampled at approximately monthly resolution, the sub-seasonal chronology is typically determined by assigning annual local maxima and/or local minima in the coral geochemical data a calendar month based on the climatology at the study site.

Proxy Calibrations

Coral proxy calibrations may use local in-situ measurements or gridded climate data products. At present, there is no standardized Sr/Ca-SST calibration equation. Slope values for the linear Sr/Ca‐SST transformation typically fall within the −0.06 ± 0.01 (±1σ) mmol/mol/°C range for the Indo‐Pacific (Corrège, 2006). The temperature dependence for δ¹⁸O at individual coral sites may range from −0.10 to −0.34‰/°C (Evans et al., 2000), whereas studies that synthesize the results from multiple locations report values of −0.20 (Evans et al., 2000) and −0.22 (Lough, 2004), which are close to the inorganic slope of −0.22‰/°C (Epstein et al., 1953).

Non-climate Influences

Corals are complex living organisms so there may be non-climatic factors or “vital effects” that influence the measured Sr/Ca and δ¹⁸O values. To minimize non-climatic influences, it is crucial to carefully sample a coral along its maximum growth axis (DeLong et al., 2013). Primary coral aragonite material should also be screened for diagenesis before sampling.

Strategies for Comparing Coral Proxy Data with Model Output

Coral climate reconstructions provide important out-of-sample tests of climate model simulations. Typically, paleoclimate proxy data is transformed into a climate variable (e.g., SST) using empirically determined calibration equations.

Alternatively, forward modeling via a proxy system model (PSM) transforms observations or climate model output into “pseudoproxies” that estimate the proxy signal (Evans et al., 2013; Dee et al., 2015). This is a common approach for both model-data comparisons and data assimilation. A PSM mathematically models how different processes impact a climate signal that emerges from the proxy data. Simple coral PSMs use observations or climate model output to forward model coral δ¹⁸O as a linear combination of temperature and salinity or δ¹⁸O_sw (if available) (Thompson et al., 2011).

Other more complex coral PSMs (Lawman et al., 2020) include other processes and assumptions inherent to coral paleoclimatology. The simulated δ¹⁸O generated using a PSM is more directly comparable to coral δ¹⁸O observations. Coral-model data comparisons have a wide array of climate applications. For example, Lawman et al., (2022) perform a model-data comparison to investigate the response of ENSO to orbital forcing during the Holocene (Figure 4).

Due to model biases, coral data users should be cautious about direct point‐to‐point comparisons between coral observations and climate model output from a single grid point. Care should be taken to select a region in the model that best matches the climate conditions observed at the proxy site.

Summary

Coral records provide invaluable information about tropical temperature and hydroclimate variability on interannual, decadal, and multi-decadal timescales. New insights about the full range of tropical climate variability and its response to external forcing will help improve the representation of the tropics in climate models and improve projections for the future.