The trade of all stony corals (Cnidaria, Hexacorallia, Scleractinia) is regulated under Appendix II of the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES). Specimens regulated include, but are not limited to, live samples and dead ornamental pieces. All traded corals should ideally be identified to species, allowing for the establishment of export quotas of threatened species, traceability and to enhance conservation efforts (Cohen et al. 2013). Recently, CITES proposed revisions to definitions pertaining to “Trade in stony corals”, acknowledging the difficulties in species-level identification in stony corals due to nomenclatural inconsistencies and identification challenges (CITES 2023). Specifically, consultation revolved around clarification of terminology, but, ultimately, all live or dead corals are to be identified at least to genus because of difficulties associated with species identification (CITES 2023; see also CITES (2013)). Indeed, visual identification of species can be confounded by morphological plasticity (Foster 1979, Todd 2008) and highly speciose groups like the genus Acropora are particularly challenging (CITES 2000, Bridge et al. 2023). Field identification of corals is problematic for regulation enforcement personnel who are not expertly trained and the difficulties of ex situ identification may be exacerbated by the issues associated with traded coral pieces that might either be cultured or fragmented from another colony (CITES 2000, Cohen et al. 2013).

Acropora is one of the most diverse and geographically widespread reef-building corals (Wallace 1999, Wallace et al. 2012), which only adds to the difficulty of identifying species. While recent studies have strongly recommended a modern integrated taxonomic approach that combines morphological and genomic data to delimit and identify Acropora species (Cowman et al. 2020; see also Quek and Huang (2021)), the extraordinary diversity of the genus remains unchallenged (Bridge et al. 2023). Within the coral trade, Acropora ranks amongst the most popular taxon of choice amongst aquarists, fuelled by their fast growth and diversity of forms and colours (Knop and Moorhead 2012). In fact, the popularity of Acropora warranted its own entry in a recent technical report by the Food and Agriculture Organisation (FAO) of the United Nations, with some 19 million pieces traded between 2007 and 2016 (Pavitt et al. 2021).

The molecular revolution has transformed our understanding of coral systematics and resulted in extensive taxonomic revisions (Kitahara et al. 2016). Recent developments of new phylogenomic approaches capitalising on high-throughput sequencing (HTS) have generated an unprecedented volume of data and further accelerated phylogenetic research on corals (Quek and Huang 2021). Nevertheless, due to the diversity within Acropora, it remains challenging to revise the clade, so numerous incongruences between its current morphological taxonomy and the phylogeny remain (Cowman et al. 2020, Bridge et al. 2023, Quek et al. 2023). It is also not feasible to monitor the identification accuracy of aquarium trade corals using HTS approaches on a regular basis due to time, cost and manpower limitations. Instead, traders utilise common names that might be region- or even country-specific, often relying inconsistently on growth form, colour and other morphological characteristics visible only in the aquarium setting, with no bearing on their valid scientific names. Therefore, until taxonomically consistent and visually identifiable characters are applied on traded Acropora corals, it is likely that misidentification of Acropora species will remain pervasive in the aquarium industry.

The importance of species identification cannot be overstated. Accurate species identification is important for conservation efforts. For example, overharvesting of rare corals can deplete local populations, leading to extirpations and downstream ecological consequences (reviewed in Dee et al. (2014)). On the regulatory front, CITES only requires all scleractinians to be minimally identified to genus, but the United States of America (USA), according to its Endangered Species Act, requires Acropora to be identified to species level (see Tlusty et al. (2023)). To enhance species identification in the aquarium trade, a recently proposed molecular framework for Acropora identification leverages two commonly used genes in Acropora phylogenetic studies — mitochondrial putative control region (mtCR) and nuclear PaxC intron (henceforth referred as PaxC) (Colin et al. 2022). Based on their workflow, they were able to assign species names to 44% and provisionally identify up to 80% of the 127 Acropora colonies tested.

Both mtCR and PaxC sequences are used widely in Acropora genetic studies, including investigation of species relatedness (Van Oppen et al. 2001, Van Oppen et al. 2003, Rosser 2015, Fukami et al. 2019). GenBank hosts > 1000 sequences of mtCR and > 400 sequences of PaxC from Acropora species (as of November 2023). Both markers span non-coding regions, therefore exhibiting a higher degree of sequence variability than coding genes. However, their suitability for species identification may be limited. Critically, the accuracy of deposited sequences is contingent on proper identification and curation of samples. Furthermore, the reliability and usability of either or both markers in Acropora remain contentious — while mtCR is able to delineate between genetically divergent clades (Suzuki et al. 2008, Fukami et al. 2021), the slow rate of evolution in mitochondrial DNA amongst scleractinians (Shearer et al. 2002) precludes fine-scale delimitation of all Acropora species (see also Van Oppen et al. (2001)). The PaxC intron alone may also not be ideal for species delimitation and identification. For example, non-monophyly of many Acropora species, based on PaxC sequences, was previously observed (Van Oppen et al. 2001, Rosser et al. 2017) Indeed, there is reason to suspect that species identification, based solely on these two markers, might be limited for Acropora.

In this study, we obtained 127 Acropora specimens from commercial aquarium vendors in Singapore in an attempt to identify them to species, following recent recommendations by Colin et al. (2022). Singapore has a bustling coral trade and corals are imported predominantly from Indonesia or Australia (data available at https://trade.cites.org/), making it ideal as a test-bed to investigate Acropora diversity in the aquarium trade. We sought to critically evaluate the applicability of DNA barcoding techniques on species identification from a diversity of Acropora species spanning different morphologies. Furthermore, to ascertain if the use of more molecular markers — apart from mtCR and PaxC — would improve species identification rates, we evaluated the potential for complete mitochondrial genomes to help delimit Acropora species. Our study is critical for determining if DNA barcoding can be broadly applied to recognise the diversity of Acropora corals in the aquarium trade, as well as understanding where improvements are needed for the enforcement of trade regulations.

Sampling strategy for traditional PCR and sequencing

DNA extraction, amplification and sequencing

Species assignment using barcoding markers

Evaluation of mitochondrial genomes for species delimitation

Sequencing and species assignment, based on mtCR and PaxC

Sequences for mtCR were recovered from 125 samples and PaxC from 70 specimens. All sequences generated were deposited at Zenodo (https://doi.org/10.5281/zenodo.12538519). Based on mtCR, 40 samples had a preliminary species assignment after excluding samples with no unambiguous species epithet (i.e. not assigned as sp. or had qualifiers “cf.” or “aff.”), of which only 37 samples had a percentage identity of over 99%. The mean percentage identity for sequences with an assigned species was 99.76% (± SD 0.53%). PaxC performed better, with 65 specimens preliminarily assigned a species identification after following the same exclusions in mtCR. However, the percentage identity inferred from BLASTn to the reference database sequences for those with an assigned identity ranged between 95.70% and 100%, with a mean of 98.43% (± SD 1.01%) (Fig. 1A). Amongst the 57 samples with only mtCR and no PaxC sequences, only 24 could be preliminarily assigned a species identity after accounting for two samples that matched to reference sequences without a clear species epithet.

Figure 1.  

Frequency of sequences of each molecular marker according to (A) highest percentage identity of the queried sequence to the GenBank database and (B) number of matched species with 100% identity between the queried sequence and the GenBank database.

Between the two loci, mtCR had more species sharing 100% percentage identity matches compared to PaxC, making it less likely to have a definitive match and subsequent species identity, based on a perfect identity match alone (Fig. 1B). Finally, only a single sample had a species assignment (i.e. A. abrotanoides), based on concurrence in identity for both PaxC and mtCR (Suppl. material 1), out of 68 samples with both loci recovered.

Mitochondrial genomes in for species delimitation

Complete mitochondrial genomes were neither able to accurately identify nor delimit species despite their length. After the removal of nine samples with a single record for the BLASTn analysis, only 35 out of 110 samples (31.82%) were correspondingly identified as the supposed species (Suppl. material 2). After excluding the Caribbean taxa (A. cervicornis, A. palmata and A. prolifera), only two species — A. florida and A. robusta — had an accurate hit for two out of the three samples in each. Out of a nominal 25 Acropora species used for phylogeny reconstruction, bPTP discerned only eight species, of which only three species had correspondingly one named species in the clade. Apart from the Caribbean clade (A. palmata, A. prolifera and A. cervicornis), all remaining species had PP > 0.9, based on bPTP. This generally corresponded with high bootstrap support (> 50) throughout the phylogeny (Fig. 2).

Figure 2.  

Maximum Likelihood phylogeny (RAxML-NG) of complete mitochondrial genomes of Acropora, with Montipora as an outgroup. Clades are collapsed, based on bPTP results for species delimitation and tips represent the sequenced-matched species identity with the number of samples denoted in parentheses. Numbers adjacent to nodes indicate bootstrap/posterior probability values from RAxML-NG and bPTP, respectively.

In this study, we sequenced barcodes for both mtCR and PaxC from 127 Acropora samples, for which we recovered mtCR sequences for 125 samples and 70 PaxC sequences. However, we were unable to provide a nominal species identity for all the samples other than A. abrotanoides with concurring results between both genes. These results suggest that both PaxC and mtCR are not useful for Acropora species identification and neither is the entire mitochondrial genome.

Databases are only as useful as the volume and accuracy of sequences and identities present within them. Coral identification is particularly difficult due to a lack of distinguishing characters in the skeleton, compounded by morphological plasticity (Todd 2008). Recent phylogenomic studies in Acropora have confirmed the complexity of the clade and the lack of consistency in determining species groups (sensu Wallace (1999)), based on morphological features. Indeed, the current evolutionary hypothesis in Acropora reveals trajectories inconsistent with traditional species groups delimited by morphology (Cowman et al. 2020, Bridge et al. 2023; see also Quek et al. (2023)). Most likely, current morphological characters used may not be sufficient to delimit species within Acropora (but see Ramírez-Portilla et al. (2022)). Furthermore, it is highly likely that many of the entries in GenBank might contain misidentifications considering the complexity of the clade, which reduces the efficacy of barcodes for species identification (Groenenberg et al. 2011). This is confirmed by our results, whereby only one sample was unambiguously assigned a species identification (i.e. A. abrotanoides), based on both mtCR and PaxC. Studies over the years have come up short when using mtCR and PaxC for species delimitation in Acropora, although they may be helpful in supporting lineages that are known to be phylogenomically distinct (Rosser et al. 2017, Ramírez-Portilla et al. 2022). We advise that neither of these two loci should be utilised for species identification without independent verification with alternative approaches, such as morphological and nuclear phylogenomic analyses.

Genome skimming is an easy and cheap method used to recover entire mitochondrial genomes in scleractinians (Quek and Huang 2021). However, in Acropora, we observe that variation across the entire mitochondrial genome is also highly limited, as evidenced by the short branch lengths and the inability of bit-score sorting to accurately classify species (Fig. 1, Fig. 2, Suppl. material 2). Comparing between phylogeny reconstruction and BLASTn, it appears a phylogeny reconstruction might be more useful for some species such as A. robusta. In this instance, considering the evolutionary distance between the monophyletic A. robusta clade comprising two samples (Fig. 1), the third A. robusta identified as A. florida (Fig. 2, Suppl. material 2) could possibly be a misidentification. Nevertheless, both are generally not sufficiently informative for accurate and consistent species identification. Mitochondrial evolution in scleractinians is notoriously slow and phylogenetically uninformative (Shearer et al. 2002, Hellberg 2006, Huang et al. 2008, Pratlong et al. 2017). Taken together, PaxC, mtCR and even complete mitochondrial genomes are unreliable for species identification of Acropora corals from the aquarium trade.

Fortunately, despite having hundreds of nominal species, not all Acropora species are common in the aquarium trade. Over the course of this study, our personal communications with aquarium vendors found that certain morphospecies tend to be traded more typically (e.g. A. millepora, A. microclados, A. tenuis and A. spathulata) and corals are also obtained from a limited set of localities (see also https://trade.cites.org/). Critically, users in the coral trade usually use names that are not rooted in scientific nomenclature, such as “Oregon blue tort” (A. tortuosa), “strawberry shortcake” (A. microclados), “Bali green slimer” (A. yongei) or “Tricolor valida” (A. valida), amongst others (Suppl. material 1). Moving forward, it would be worthwhile building a curated inventory of common names and species names that are verified by phylogenomic data and experts on coral taxonomy. This is especially useful in cultured corals, as only the parent colony needs to be identified to species. Subsequently, all stock has to be declared appropriately, with any deviations from the list to be reported with origin, photographs and finally, an identity assigned by an expert, ideally supported or at least periodically verified by phylogenomic analysis. As taxonomic revisions arise (e.g. Bridge et al. (2023)), this list can be updated accordingly. If an unknown Acropora colony is harvested from the wild and is of interest to regulators, nuclear phylogenomics would be required to accurately identify it to species (but see Ramírez-Portilla et al. (2022)). Hybrid capture analysis may be the most feasible amongst available phylogenomic approaches given currently available resources (Cowman et al. 2020, Quek et al. 2023). However, this method relies on a rich set of genomic data for many species and will likely only become viable in the coming years with new datasets being produced (e.g. Cowman et al. (2020), Bridge et al. (2023)). In the meantime, in silico methods can augment available data beyond those recovered from hybrid capture alone (see Quek et al. (2023)).

Identification of traded flora and fauna should ideally be performed at the species level, as the data can provide precise information for understanding trends and tracking the global trade of threatened species. In particular, robust and reliable species-level identification empowers CITES parties to implement targeted conservation policies and measures to improve the enforcement of wildlife trade regulations. Unfortunately, for scleractinian corals, species-level identification accounts for a small proportion of identification and, even so, it is often discarded as unreliable and unusable for analysis due to the high margin of error (Green and Shirley 1999). In this study, we found many Acropora colonies harvested from the wild for sale in the aquarium trade without proper taxonomic identification. The IUCN Red List (https://www.iucnredlist.org/) for Acropora lists many species of traded Acropora corals that run the gamut of categories, in ascending order of priority: least concern (e.g. A. valida, A. yongei), near threatened (A. digitifera, A. tenuis), vulnerable (A. aspera, A. jacquelineae) to even endangered (e.g. A. suharsonoi). Therefore, the identification of traded corals only to genus for permitting and reporting purposes may limit the conservation outcomes of trade regulations, especially in alleviating the threats of exploitation and extinction.

To reduce overharvesting of corals from wild populations, particularly amongst species at greater risk of extinction, we recommend that, as far as possible, corals for the aquarium trade should be sourced from aquaculture and coral farms instead. Furthermore, with the development of more accurate and rapid molecular methods to generate and analyse coral phylogenomic data in the future, as is currently occurring for other marine taxa (e.g. But et al. (2020), Naaum et al. (2021)), more effective regulatory and trade enforcement policies should be established once reliable species identities can be obtained for corals. Meanwhile, however, we emphasise that the two most popular Acropora barcoding markers — mtCR and PaxC — are inadequate for accurate species identification and more suitable loci need to be developed and tested.

There is a pressing need to move towards a working model for the identification and traceability of traded scleractinian to boost conservation outcomes. The current CITES requirement with respect to corals allows declaration of species where possible, although genus identification is accepted. This is largely due to practical limitations and trade data can be adulterated where there is disregard for accurate species identification or lack of expertise in taxonomic identification. Furthermore, DNA barcodes are extremely limited in utility insofar as Acropora is concerned and should, therefore, not be the only means for species identification (but see Hoeksema and Arrigoni (2020)). To this end, we propose the following steps to improve traceability of the coral trade: (1) establish an inventory of the commonly traded corals with information on IUCN Red List status, source country, common name and scientific name, of which the specific species identified is to be verified by coral taxonomists and/or using nuclear phylogenomic analyses (at least until suitable barcodes for Acropora species identification are available), with the list updated in tandem with taxonomic revisions and conservation status updates; (2) all imports and exports to be declared against the list and any deviations from the list shall be appropriately recorded; (3) in the event of a deviation, images and a tissue sample shall be taken and retained by enforcement bodies for future verification; (4) periodic audits to be conducted, with stipulated identities verified by coral taxonomic expertise; (5) uncertain samples should be sent for nuclear phylogenomic analysis to ascertain their identities more accurately. Importantly, we recognise that some of these suggestions may not be feasibly adopted given the limited resources. These proposals are available as considerations for future conservation efforts and can be implemented in parts. Finally, the import and export of threatened species should be avoided; instead, coral farms can be a good avenue to supply the demand for all taxa, whether common or rare, thereby reducing the burden on wild populations.

We would like to thank Lionel Ng, Steffi Tan and the staff of the aquarium shops for sharing their knowledge and insights.

This research is supported by the Temasek Foundation under its Singapore Millennium Foundation Research Grant Programme (A-0003203-01-00).