The rapid development of sequencing technologies and the ever-decreasing cost has led to a discrepancy between the generation of primary sequencing data (sequence reads) and their assembly, annotation, and curation (genomes, genes, etc.): we are producing more data than we can “consume” (Richards, 2015; Voolstra et al., 2017a). This inconsistency is highlighted by the now routinely required provisioning of primary sequencing data to a public database (NCBI nr, EMBL ENA, and DDBJ) prior to publication vs. the provisioning of assembled and annotated sequencing data (the type of data that most people work with), which currently is not a strict requirement (Liew et al., 2016; Voolstra et al., 2017a). Indeed, accessibility to assembled sequencing data is generally provided on a voluntary basis, and more often than not, relies on secondary databases, such as reefgenomics.org (Liew et al., 2016) or symportal.org (Hume et al., 2019) in the marine/coral reef domain. These secondary outlets often lack funding (or the availability of funding schemes that support such endeavors), rendering their continued upkeep financially challenging, e.g., CnidBase that is now no longer accessible (Ryan and Finnerty, 2003) or GeoSymbio which is no longer updated (Franklin et al., 2012). Even when processed sequencing data are available, another problem is version control, i.e., access to and documentation of previous transcriptome or genome versions (assemblies), which in some instances are critical to reproduce results. Public databases often put restrictions in place for the upload of genome/transcriptome assemblies or gene sets, resulting in different versions used for analysis, relative to those that are published with the respective study. This disparity is further complicated by the circumstance that sequencing databases often produce their “own” version of an uploaded genome based on a standardized analytical framework. In the case of the Aiptasia (Exaiptasia diaphana) genome (Baumgarten et al., 2015), for instance, a comparison of the submitted GenBank version (PRJNA261862¹) to the RefSeq version (PRJNA386175²) using a gene mapping file³ reveals different lengths and numbers of protein-coding genes. The same can be observed for the genome of the coral Stylophora pistillata (Voolstra et al., 2017b) with the author-published version featuring 25,769 genes⁴, the corresponding submitted GenBank version harboring 24,140 of these genes⁵, and the associated RefSeq version featuring 33,239 genes⁶ with no corresponding gene mapping file to cross reference the different genes and identifiers.

Large-scale sequencing projects often prioritize the generation of genomic and transcriptomic data over comprehensive formal descriptions of samples and their environmental/ecological setting (i.e., metadata). This is true even for species with high intraspecific variation in heritable functional traits, such as scleractinian corals, for which ecological and environmental context matters greatly (Ziegler et al., 2014; Sawall et al., 2015; Röthig et al., 2017; Thomas et al., 2018; Bongaerts et al., 2020; Kavousi et al., 2020). Underlying this problem is that most molecular databases focus largely on sequencing data deposition and do not provide a comprehensive framework for the deposition of associated metadata (Riginos et al., 2020). The association between sequencing data and contextual, environmental (meta)data makes interpretation of these data more meaningful and allows the alignment of molecular patterns with phenotypes (Hume et al., 2020; Voolstra et al., 2020; Grottoli et al., 2021). The recently established Genomic Observatories Metadatabase (GEOME) aims to expedite and improve deposition and retrieval of molecular data and metadata for biodiversity research (Deck et al., 2017; Riginos et al., 2020). Here, we address a specific key issue relevant to this aim: the importance of accurate taxonomic descriptions of sequenced coral holobiont specimens and the deposition of specimen vouchers to provide a formal taxonomic framework for sequencing data, coupled with the ability to update existing descriptions. The absence of a proper taxonomic treatment associated with sequenced specimens makes cross-referencing and meta-analyses challenging and, in the worst case, can confound analyses due to taxonomic misclassification of sequence data (Bonito et al., 2021). Simply put, while everyone agrees on the value of properly curated specimens and associated sequencing data, what is missing is a guide or reference that details what should be provided when sequencing a genome.

Here we were motivated to provide such consensus guidelines as we embark on a new initiative to substantially improve the number and quality of genomes available from scleractinian corals and their associated Symbiodiniaceae microalgae (Supplementary Table 1). The ‘‘Coral symbiosis sensitivity to environmental change hub’’ is embedded in a phylogenetically broader effort to survey genomes across a wide variety of marine organisms and their microbial symbionts (octocorals, sponges, clams, nudibranchs, etc.) entitled the ‘‘Aquatic Symbiosis Genomics Project’’, which is jointly funded by the Wellcome Sanger Institute and the Gordon and Betty Moore Foundation⁷. We aim to provide consensus guidelines on the “minimal taxonomic information” that should be provided to maximize the utility of the generated sequence data. We further expand these guidelines to also include coral-associated prokaryotic genomes due to recent efforts in describing and collating the culturable fraction of the prokaryotic community of the coral holobiont (Sweet et al., 2021). We advocate for the provision of taxonomic information for the most important (i.e., best understood, most commonly researched) coral holobiont entities: the coral animal host, the Symbiodiniaceae microalgae, and the associated prokaryotes (bacteria and archaea). Although the focus of the guidelines is aimed toward shallow-water stony corals (Scleractinia), they are broadly applicable to all coral taxa, and we incorporate specific considerations for temperate, cold-/deep-water corals as well as octocorals (Octocorallia), black corals (Antipatharia), and other hexacorals (Hexacorallia) where applicable.

Standardized morphological and molecular taxonomic practices are not equally available for all coral holobiont entities, nor equally well tried-and-tested. For instance, coral skeletal-based taxonomy has a long history (Veron, 2000), but is not without discrepancies if compared against molecular-based analyses (Fukami et al., 2004; Kitahara et al., 2016; Terraneo et al., 2019a; Cowman et al., 2020). But therein lies the conundrum: while molecular analyses commonly achieve superior taxonomic resolution, they rely on initial expert review and annotation to prevent error-propagation through incorrect phylogenetic annotations of sequence database entries (Tripp et al., 2011). It is important to acknowledge that taxonomic identification is challenging because morphological characteristics that differentiate species in one genus may not be applicable to other genera, and the same is true for molecular markers (Veron, 2000; Shearer et al., 2002; Stolarski et al., 2021). In the case of many coral lineages, species-level molecular markers are simply not (yet) available (Quattrini et al., 2018; Cowman et al., 2020; Erickson et al., 2021), partially due to ongoing taxonomic revisions, but also due to corals exhibiting low levels of congeneric divergence for commonly employed (mitochondrial) gene markers, effectively hampering species-level resolutions (Shearer et al., 2002; Supplementary Table 2). Both circumstances support the necessity of skeletal voucher specimens as a reference to validate, synchronize, or update ascribed taxonomic annotations and allow later re-evaluation in case of taxonomic revisions. Importantly, specimens should be identified with reference to the original type specimens and descriptions, and not the most recent or most easily accessible revision, unless these provide a formal re-description (or illustration) of type material (or neotype specimen where applicable). Nevertheless, for most sequenced coral genomes to date, such information is not or not easily accessible (Supplementary Table 3). With most museums placing emphasis on digitizing collections, it should become easier to access photographs of type specimens and original descriptions—a major step forward from even a decade ago. Museum curators and collection managers can also facilitate this process by providing access to specimens (including digitized versions) in their collections—a valuable service to the broader scientific community.

By comparison, formal descriptions of Symbiodiniaceae are rather recent, with the vast majority established or formalized after molecular data began to be integrated (LaJeunesse et al., 2012; Wham et al., 2017; Lee et al., 2020). The updated taxonomy provided an overdue revision of this group of microalgal symbionts, acknowledging their substantial genetic divergence and discouraging the use of informal clade designations as auxiliary constructs (LaJeunesse et al., 2018). The majority of sequenced genomes are currently available from the genus Symbiodinium, with many genera not yet having genome assembliesavailable (Supplementary Table 4). Rather, Symbiodiniaceae associations are commonly described through means of marker gene elucidation using a range of different methodologies (Sampayo et al., 2009; LaJeunesse et al., 2018; Hume et al., 2019; Grottoli et al., 2021). Common markers that are sometimes used in conjunction include ITS, ITS2, psbA^ncr, SSU, LSU, and cp23S, which are utilized along with morphological data and host associations (Sampayo et al., 2009; LaJeunesse et al., 2018; Hume et al., 2019).

For coral-associated prokaryotes, much work remains to be done (Supplementary Table 5), but the recent assembly and genome-level description of bacteria associated with Porites lutea Milne Edwards and Haime, 1851 (Robbins et al., 2019) and the cataloging of cultured bacterial coral isolates (Sweet et al., 2021) provide a groundwork to build upon. Given that coral genomics is a nascent field, any guidelines put forward here must be considered provisional, and indeed current limitations should be a motivation rather than a barrier to begin to work on formulating the types of information that are most important to provide alongside sequencing data. While it is evident that multiple challenges are associated with taxonomy at all levels of the coral holobiont, we begin with a set of guidelines focusing on what should be provided when generating reference genomic data for the coral animal host, Symbiodiniaceae microalgae, and those prokaryotes that are either cultured or for which a full-length 16S rRNA gene reference sequence or a well-assembled (meta)genome is available (Supplementary Material). Our recommendations are not prescribed for metabarcoding, gene expression, or metagenomic/-transcriptomic surveys per se, as they may become overburdening for these latter types of studies. Although providing metadata descriptors for these data types in as comprehensive a manner as possible is desirable, they typically do not represent “reference datasets” because multiple studies are typically available for these types of sequencing data for any given species (e.g., 16S metabarcoding datasets exist for the same species from multiple locations). We further advocate establishing a well-curated set of specimen vouchers associated with primary reference sequencing data, which then allows alignment of samples against that reference. This should minimize misannotation and curtail error propagation caused by annotating tertiary sequencing data against secondary sequencing data.

The sequencing era has the potential to unlock the complexity of the coral holobiont by means of highly resolved genomic interrogation of its member species (i.e., coral animal host, Symbiodiniaceae microalgae, associated prokaryotes, etc.). While initially the focus was on sequencing “one genome at a time”, e.g., the Stylophora pistillata (Esper, 1792) holobiont genomics studies (Bayer et al., 2013; Aranda et al., 2016; Neave et al., 2017a, b; Voolstra et al., 2017b), there is now a suite of efforts to target the coordinated sequencing of all (or the most common) holobiont member species (Robbins et al., 2019). One of these efforts is the “Aquatic Symbiosis Genomics Project”. To maximize the utility of the generated data, a common commitment to formulate and adhere to consensus guidelines within a defined taxonomic framework is required. Here, we lay out the guidelines that the “Coral symbiosis sensitivity to environmental change hub” will follow to facilitate meta-analyses, cross-comparisons, and back-tracking of samples, with the intent that other initiatives can join and adopt this approach. Our first step was to decide on the coral species that would be part of this project (Supplementary Table 1). To do this, we collated the current state of play of coral genomes and assessed which key species were missing or suffered from incomplete and/or fragmented genome assemblies. We then compared where our selected corals were initially described from, that is the country of origin of the type specimen (type locality). Samples are currently in the process of being collected by in-country scientists and experts who are in charge of sampling and archiving specific types of metadata for each specimen. Without such data, type specimens and previous data collections cannot be ground-truthed or revised (Blom, 2021; Buckner et al., 2021), which ultimately limits the usefulness of -omics data for current and future analyses.

One additional barrier is the lack of a central repository that integrates (i) several or all types of data (genomic, taxonomic, physiological, chemico-physical, etc.) from (ii) multiple coral holobiont entities (cross-kingdom) with (iii) the inclusion of version control and access to “derived” data products. Recent initiatives aim to provide centrally available open-access databases that integrate primary data and some associated metadata (Box 1). While the broad centralized integration of data is meaningful, our point is not to suggest a single database to hold all data, as this is likely to affect implementation, focus, and usability. Rather, the coordination of efforts into a few collective and linked databases is desirable to avoid duplication of efforts. The power of a consensus framework was recently outlined for coral bleaching experimentation, which detailed response variables to increase comparability and hasten scientific insight (Grottoli et al., 2021). Given the pervasive lack of long-term funding for data centralization (including logistics, sorting, and collection), the alternative bottom-up, community-driven model is a more realistic goal to attain, and will be particularly valuable if it manages to incentivize data and meta-data deposition. Arguably, the burden to follow through with comprehensive data deposition lies with the individual researcher and is typically done ‘‘after the fact’’ (after publication). However, free-of-charge repositories, such as zenodo.org or figshare.org, provide digital object identifiers (DOIs) and by that a mechanism of citing and acknowledging well-curated data, ultimately incentivizing such efforts. For the ‘‘Aquatic Symbiosis Genomics Project’’, all sequence data will be openly accessible. All raw and assembled sequence data will be deposited in the European Nucleotide Archive (ENA) database which is part of the International Nucleotide Sequence Database Collaboration that also entails the DNA DataBank of Japan (DDBJ) and the GenBank at NCBI, which exchange data on a daily basis. Further, our intention is to rapidly publish all submitted genome assemblies alongside their associated meta-data as Wellcome Open Research Data Notes, which can be cited⁸. It is now up to us (the scientific community) to further foster these endeavors through proper acknowledgement and citation of non-traditional publication outlets. We hope that the consensus guidelines detailed here provide a path to broaden our understanding of coral holobionts, to accelerate discovery, and to facilitate novel solutions to mitigate coral degradation, which becomes ever more pertinent as we witness the continuous loss of reef ecosystems globally.

CV, KQ, SD, JP, RP, MA, ACB, CC, MZ, and MS conceived the idea. CV, KQ, SD, JP, RP, CB-L, TB, CC, DC, JF, TR, DS, MZ, and MS wrote the manuscript. CC conceived and created Figure 1. All authors collected the data, provided input, and contributed to the writing of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.