Background & Summary

Trait-based approaches are one way to correlate different characteristics, or traits, of organisms to environmental changes with the ultimate goal to predict community and whole-ecosystem responses to such changes. Traits of reef corals have been explored to identify their life-history strategies1 and extinction risk2,3. Several palaeontological studies have examined trait-based extinction using the fossil record4,5,6. Changes in ancient reef ecosystems can provide valuable insights into drivers of reef proliferation and decline. Understanding why fossil reefs collapsed and how they recovered may help to inform conservation activities in modern coral reefs in light of anthropogenic climate change. Many reefs of the Phanerozoic eon (the last 540 million years) can be considered analogues of modern tropical coral reefs with regards to reef architecture and environmental controls7. Phanerozoic reef-building is not restricted to scleractinian corals but also a multitude of other sessile hypercalcifying animals (animals with a large skeletal to biomass ratio) such as extinct coral clades, calcifying sponges and rudist bivalves.

In the first release of this database, we focus on scleractinian corals, which have a rich fossil record from the Middle Triassic (~245 Ma) and became the dominant reef builders in the Late Triassic (~225 million years ago)8. Scleractinian corals have an extensive fossil record due to their calcified skeletons. Accordingly, morphological traits are well preserved and several of those traits have been shown to be linked to species extinction risk: examples are corallite integration, corallite diameter, growth rate (linked to morphology or measured directly from growth bands), and colony longevity (linked to size)3,9,10,11,12. Spatial and environmental traits are also important predictors of extinction risk: habitat breadth and maximum water depth are two key parameters linked to climate change vulnerability in corals9.

The traits of extant reef corals have been compiled in the openly accessible Coral Traits Database (CTD)13,14, which in addition to morphological traits, also contains data on physiology, biology and reproduction. However, there is no single, exhaustive resource for fossil reef building organisms including extinct scleractinian corals and the traits of extinct corals are currently scattered in online repositories and research publications. Some trait compilations that cover only one type of trait (i.e., corallite integration) and/or span one period (i.e., Triassic) have previously been published15 but are not openly available. The online resource Corallosphere (www.corallosphere.org) contains a description of genus-specific traits for both extinct and extant scleractinian corals, while the Paleobiology Database (PBDB; www.paleobiodb.org) contains information for a limited number of traits such as inferred symbiotic status or preferred environment. The database on Neogene Marine Biota of Tropical America (NMITA) mostly contains trait information of Pleistocene to Holocene scleractinian corals from the Caribbean region16,17. Therefore, there is a need for an updated, comprehensive and openly accessible compilation of fossil coral traits.

Here, we present the Ancient Reef Traits Database (ARTD) which fills this gap. ARTD is a unique, specimen-based compilation of reef builders’ traits. At the initial stage, the database contains 15 traits of scleractinian corals and covers the time period from the mid-Triassic until the mid-Holocene. The value of ARTD lies not just in its data coverage but in its interlinkage with other databases. ARTD is compatible with CTD and is also designed in a way that provides easy integration with the biggest resource of fossil occurrence data, namely the PBDB.

Methods

ARTD has a similar structure as the CTD and is designed to contain specimen-level traits for identified species and genera (Fig. 1). The basic unit of entry is that of the trait of a single specimen, e.g. a coral specimen, that is also accompanied by contextual characteristics such as the geological stage(s), the present-day geographic region and the present-day coordinates of the locality in decimal degrees, in which the specimen was found. Reconstructed palaeo-coordinates were computed from the present-day coordinates using the rotation file supplied by C. Scotese in the ‘PALEOMAP PaleoAtlas for GPlates’ package18. These metadata are crucial for analyses of traits in a spatial, temporal and environmental context. For example, the corallite diameter or colony size of a species may vary due to genealogical trends or environmental factors19. A specimen can be linked to a number of observed traits as per the source of the data. In the case where there are multiple observations for a specimen or group of specimens (usually reported as such in the primary source), text-based information is entered and separated by “-“, e.g. trabecular-substyliform for columella structure. We also employ a hierarchical taxonomic structure where any inheritable trait (i.e. symbiotic status or corallite integration) of a taxon, is automatically applied to all the lower taxonomic levels and specimens of that particular taxon. For example, if a genus’ corallite integration is qualified as cerioid all its containing species and specimens are also assigned to a cerioid trait, unless exceptions are known. In the case of growth forms, heritability was assigned only to genera where one growth form has been assigned, e.g. Isastraea (massive) as growth forms tend to vary among some groups due to factors such as environment20.

Fig. 1
figure 1

Overview of the ARTD back-end structure. The observation table is highlighted in red. The remaining tables contain supplementary information that complement the observation records.

Full size image

The taxonomic and contextual information are stored separately and linked to the “observation” table where each entry is provided with a unique identification number (id; Fig. 1). This observation table contains the respective specimen-level traits which contains information of the trait(s) being entered, the value (numeric or character depending on the trait being entered), the value type (whether raw, aggregated, or based on the opinion of experts), and the method used to obtain the measurements (e.g. observations made in the field or laboratory). This structure is used so that multiple specimens from one specific site and one taxon can be linked several times to the same traits. For example, published articles provide several measurements of corallite widths21. In the case when a range of values is provided for a measured trait, the mean value is entered, except for the traits “corallite width minimum” and “corallite width maximum” where the minimum and maximum values are entered, respectively. Traits are divided using the categories that are comparable to the CTD: morphological, reproductive, or physiological. However, it is clear that only morphological traits are directly accessible in fossils, whereas the latter categories need to be inferred. Currently the only inferred trait in our database is that of symbiotic status (trait_name: “Zooxanthellate”) which relies on morphological criteria of extinct coral species as defined by Kiessling and Kocsis22. The original source of the entered data is also included in the database (called the primary source), and if available, the reference number for this primary source in the PBDB which can then be used to integrate both databases. The taxonomic information is automatically validated against the PBDB before each database release to ensure that the most up-to-date information is available.

Data Records

Here, we provide a release of 15 traits of 505 genera and 1129 species, comprising a dataset of 17,841 trait values obtained from localities around the world (Fig. 2). This first release has broad taxonomic and temporal coverage and comprises more than 70% of all coral genera in the PBDB in each geological stage since the mid-Triassic (Figs. 2, 3a). The availability of traits per specimen or taxon also varies (Fig. 3b) depending for example on the preservation of the specimen being reported in the primary source, with some traits such as corallite diameter or corallite integration more readily available than others such as colony size (Fig. 3c).

Fig. 2
figure 2

Map of localities, categorised by geological period, from which trait data of scleractinian corals were compiled in ARTD.

Full size image
Fig. 3
figure 3

Data coverage in ARTD: (a) Taxonomic coverage of ART-only data and ART data combined with CTD compared to taxonomic occurrences in the PBDB; (b) Number of trait values in ARTD for coral genera in the PBDB; and (c) Data completeness of traits as a percentage of specimens in ARTD.

Full size image

A static release of the database is available directly from the ART platform (https://art.nat.fau.de) as well as Zenodo23. The data release is in the form of a compressed folder containing two files:

  1. 1.

    data.csv: A csv-formatted file containing the contextual information and measurement of specimens.

  2. 2.

    references.csv: A csv-formatted file containing the bibliographic information of data sources24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114.

The details and descriptions of the available trait variables are available in Table 1. Up-to-date data are directly available from the database. However, as data entry and validation (see Technical Validation) is ongoing, users are recommended to use the data made available through the static releases to maximise reproducibility of analyses and results. Both the static releases23 and direct downloads are accompanied by the primary sources24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114 which should be credited.

Table 1 Overview of traits available in ART v1.0, including descriptions and standard categories used.
Full size table

All data in the ARTD and included in this release are linked with published (e.g. peer-reviewed papers, taxonomic monographs, books) references24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114. The final dataset consists of one row per trait for each specimen entered. Each specimen is given a unique identification number (observation_id) which can be associated with various traits. The geographic data is available for each specimen, such as the country or region in which it was found (location_name), present-day coordinates (longitude, latitude), reconstructed palaeo-coordinates (paleolng, paleolat), and taxonomic information (identified_name: as entered in the database, accepted_name: based on updated taxonomic information, genus_name: genus of the specimen as per the updated taxonomic information, species_name: the species name as per the updated taxonomic information). The time period identified for the specimen (early_interval: the first interval in which the specimen was found, late_interval: the last interval in which it was found, min: the minimum identified age, max: the maximum identified age) is also provided. For each specimen, the available traits are entered. Each trait (trait_name) is assigned to a category (trait_class), and a trait entry for a specimen contains the trait value and unit (standard_unit) if applicable. Additional information (value_type) about any measurements such as whether the entered measurement is raw or an aggregated value (mean, minimum, maximum) based on expert opinion, model-derived, the unit for the measurements (standard_unit), and the methodology used to obtain the data (methodology_name) is also provided if available. Each entry contains the identification number of the data source (reference_no), whose bibliographic details are provided in the reference.csv table, which can then be used for citations purposes.

Technical Validation

The database is curated by the managerial board who undertake the tasks below. Curation and quality control of the data include:

Updating taxonomic information

All taxonomic data are cross validated taxonomic names against the PBDB to ensure that reliable taxonomic data is available.

Dealing with duplicates

All duplicated records identified based on the combination of data source, taxonomic information, location, time period, and measurement values are flagged and then removed.

Applying inherited traits

Any newly entered traits that are identified as inherited traits are applied to all species of the genus that contains that particular trait.

Standardising text-based fields

Text-based fields such as corallite integration or growth form are standardised (Table 1) to facilitate analysis on these traits.

Contributor approval

Anyone wishing to contribute to the database should become a formal contributor and any observations entered by them will be associated with their user account.

The database is hosted on the server of the Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) and will be maintained on the long-term by WK, MCK and other staff members of the FAU.

Usage Notes

ARTD offers many new opportunities for incorporating trait-based approaches in addressing macroevolutionary and macroecological questions using the fossil record of reef-building organisms. The correlation of certain traits with environmental conditions and their vulnerability or resilience to environmental change can be widely used to understand the evolution of these organisms and the reefs they build over time, their extinction risk in light of global warming and recovery after such an event. A trait-based framework allows the analysis of variation and evolution of traits within and across reef-building organisms over time and their responses to environmental change in the past. For example, the data from ARTD combined with the PBDB show that the diversity trajectories for colonial and solitary scleractinian corals were different (Spearman’s ρ = 0.241, p = 0.129 of first differences), with colonial corals showing an increase in diversity during the late Jurassic and Cretaceous but declining again in the Late Cretaceous (Fig. 4). On the other hand, the diversity of solitary corals remained relatively low compared to colonial corals.

Fig. 4
figure 4

Range-through diversity of solitary and colonial corals since the Triassic using data from ARTD and the PBDB. Recent corals were not included in the range-through analysis.

Full size image

Trait-based frameworks are also widely used in modern ecology to study the climate impacts of organisms and their extinction risk1,14,115,116,117,118. Focusing on such frameworks would therefore allow the integration of palaeontological and neontological data using similar concepts and methods to address urgent questions on biodiversity and extinction at multiple scales3,119,120. Such an integrated approach might contribute to the conservation of modern coral reefs, which are the most threatened ecosystems from climate change121.