Modern Coral Taxonomy Requires Biologically Relevant Evidence. Reply to Cowman et al. Comments on “Veron et al. Review of Coral Taxonomy, Evolution and Diversity. Diversity 2025, 17, 823”

Abstract

Herein we reply to the commentary of Cowman et al. on our Review of Coral Taxonomy, Evolution, and Diversity. We demonstrate that many of their central criticisms are mischaracterisations, contain factual errors, or extend beyond the evidentiary scope of available molecular datasets. Our Review did not dismiss the legitimacy or importance of molecular approaches, nor the value of integrative taxonomy. Rather, it emphasised the evidentiary thresholds required for formal species-level revision in morphologically variable and geographically widespread coral taxa. Genetic differentiation should not, without comprehensive sampling and contextualisation, be treated as sufficient grounds for immediate species-level restructuring. We reiterate that the concept of reticulate evolution in corals is supported by a growing body of molecular and other evidence. Furthermore, the “biological entities” discussed in our Review are not subjective impressions or non-reproducible opinions but are empirically documented, repeatedly recognisable, and diagnosable, characterised by coherent suites of morphological, ecological and geographic traits. These constitute structured, testable species hypotheses within an integrative framework. Our Reply addresses issues of field variability, type specimens, sampling design, molecular and morphological incongruence, and the taxonomic and conservation implications of premature rank assignment. Where multiple, independent lines of evidence converge, taxonomic revision is warranted; where sampling remains limited, geographically restricted, or analytically unstable, and where different lines of evidence conflict, nomenclatural stability and biological coherence are better served by restraint.

1. Introduction

We welcome the opportunity to respond to the commentary by Cowman et al. (2026) [1] (hereinafter referred to as “Cowman et al.”) on our recent Review article (Veron et al., 2025 [2]). Our Review sought to stimulate critical reflection on species-level taxonomy in corals, particularly in light of rapidly expanding molecular genetic datasets. However, many of the issues raised by Cowman et al. reflect misunderstandings, misinterpretations, or factual inaccuracies. Thus, we welcome this opportunity to clarify our position and address areas of disagreement (hereafter, our “Reply”). The main text of this Reply (principally Section 2 below) addresses the methodological and conceptual issues raised by Cowman et al., while the Supplementary Text and its Appendices provide a detailed point-by-point examination of their Commentary, enabling direct comparison of their critique and our response.

In their Acknowledgments, Cowman et al. imply significance to the breadth of authorship of their paper (33 authors from 31 institutions). However, we note that a majority of these authors are contributors to studies that we assessed critically in our Review, mainly on the grounds that the evidentiary standards were insufficient to support the species-level inferences drawn. We nevertheless acknowledge that many of these authors have made substantial contributions to coral genetics and systematics. Indeed, we accepted these findings where multiple lines of evidence converged and were non-conflicting.

Where apparent conflicts existed, however, we detailed them and questioned whether the evidence presented was sufficient to justify formal species-level revision. Several of the species-level cases raised by Cowman et al. illustrate our view (see their Section 7 and our corresponding responses below, Section 2.11 and in Supplementary Section S7 and its Appendices SA1–SA4). In those cases, lineage partitions were elevated to species status on the basis of limited sampling and/or a narrowly constrained interpretation of type specimens, without adequate assessment of representativeness within the known natural variability of taxa across habitats and ranges. In each case, contrary to the claim of Cowman et al., we did not dismiss the findings outright. Rather, we assessed the evidence, provided our considered view, and called for further work (see our Review Section 7 for details). Furthermore, and also contrary to Cowman et al.’s assertion, we have long recognised the essential role of molecular approaches in advancing coral systematics and explicitly acknowledged in our Review the substantial reshaping of higher-level phylogeny achieved through molecular research. Indeed, we stated explicitly at the beginning of our Review Introduction that:

“… molecular studies have provided much greater insight into the phylogeny of species, resulting in name changes at all levels. Corals are at the forefront of marine invertebrate taxonomy, with molecular studies providing insights into the relationships within and between species. As a result, there have been many changes to phylogenetic positions and a greater understanding of species relationships using population genomics.”

Where integrative datasets with adequate geographic and biological representation demonstrate consistent genetic, morphological, and ecological separation, we support formal taxonomic revision. Hence our principal disagreements are not about the legitimacy of molecular methods, but the strength of evidence required before lineage partitions are recognised as separate species. In morphologically variable and geographically widespread taxa, detection of genetic structure alone may be insufficient to establish diagnosable biological species unless sampling and type interpretation adequately capture known natural variability. Our position on species delimitation and taxonomic revision in relation to the principal papers under discussion in Cowman et al. is summarised in Section 2.11.2.

2. Central Methodological and Conceptual Issues

2.1. “Biological Entities” and Alleged Non-Reproducibility

Cowman et al. mischaracterise the “biological entities” discussed in our Review as subjective constructs, implying reliance on expert intuition rather than reproducible evidence. They suggest that this approach renders these entities effectively untestable, allowing the taxon concept to drift independently of the type specimen.

This is incorrect. The “biological entities” discussed in our Review are neither subjective impressions nor products of individual intuition, but structured species-level hypotheses grounded in empirical evidence. They are derived from studies by numerous previous and contemporary taxonomists and are based on repeated field observation, museum material, geographic comparison, ecological context, and reproductive and molecular evidence accumulated over many decades of study. The existing species hypotheses synthesise these studies and draw on an extensive cumulative comparative base of specimens, including types, as well as field observations across coral reef regions globally.

The following quote is drawn from Bridge et al. (2023: p. 6) [3], one of the principal papers in focus in Cowman et al.:

“Primary species hypotheses (PSHs) were identified as distinct molecular lineages … This allowed us to designate species names to particular lineages when the type material was available. However, it is important to note that this process does not necessarily capture the intraspecific variation required to delineate species boundaries in the field. Although the capacity to identify species in the field based on morphology is important for many research questions, it is beyond the scope of this study and will require numerous detailed studies focusing on a manageable number of species at a local scale.”

(our bolding)

Existing species delineations are indeed grounded in this broad and extensive comparative framework, from integrated assessments of the literature, types, and many other collected specimens, and repeated observations across depth gradients, habitat types, and numerous geographically dispersed sites throughout the distribution ranges of the taxa concerned. The above authors acknowledged the importance, indeed necessity, of such studies but nevertheless enacted numerous taxonomic actions in their absence, disrupting long-standing species concepts based on extensive and structured research in the process.

Building on the substantial body of prior taxonomic work by earlier taxonomists, along with many collaborations with regional specialists, our team has collected and examined approximately 30,000 specimens evaluated against thousands of other museum specimens and numerous in situ encounters across almost all major coral reef biogeographic ecoregions (Review Section 3.3.5). Recognition of these taxa requires training and experience, as does taxonomic practice in any morphologically complex group. However, that experience does not generate the patterns observed; rather, it enables consistent recognition of integrated character combinations that persist across regions, habitats, and independent observers and that remain diagnosable even in the presence of related taxa. These entities are therefore reproducible by independent observers, testable, and in many cases historically stable components of coral taxonomy, requiring a similar level of specialist training and comparative familiarity necessary in other complex biological groups. Specific allegations by Cowman et al. regarding “subjectivity” are addressed in Supplementary Sections S2b, S4b and S6a,e.

2.2. Integrative Taxonomy and Alleged Resistance to Genomics

Cowman et al. imply that our Review resists or dismisses molecular evidence, framing our position as oppositional to modern reproducible science. This misinterpretation does not reflect the content of our Review.

We explicitly support integrative taxonomy and the use of genomic-scale data in coral systematics—this was clear in our Review. Contrary to Cowman et al.’s implications, we have long recognised the essential role of molecular approaches in advancing coral phylogeny [4]. We also explicitly acknowledged the substantial restructuring of higher-level taxonomy achieved through such research. We did not, in our Review, and do not in CoralsOfTheWorld.org, focus on taxonomy at Family level or above. Our commentaries were consistently directed at particular genus and species level revisions where morphological and/or molecular inferences were in conflict and taxonomic and nomenclatural conclusions lacked consistent, robust support. We also highlighted recent integrative studies that have provided robust, well-supported insights where inferences matched, rather than exceeded, the scope of the relevant studies.

Our Review did not advocate substitution of morphology for molecular methods, nor did it privilege “biological entities” as an alternative to reproducible quantitative approaches. Rather, we emphasised that the design of effective molecular studies depends upon accurate biological contextualisation—including reliable field identification, recognition of closely related taxa, and adequate representation of geographic and ecological variability.

Molecular genetic data provide powerful and often indispensable insight into evolutionary relationships. However, translation of lineage-level genetic structure into formal species-level taxonomic revision requires careful consideration of sampling scope, analytical assumptions, and the broader biological context of the organisms concerned. Our commentary was directed at limits of inference and evidentiary sufficiency, not at molecular methodology itself. Case-specific clarifications are provided in our Supplementary Text, particularly Sections S4–S7.

2.3. Reproducibility and Alleged Double Standard

Cowman et al. allege that we apply a double standard in evaluating taxonomic evidence. Specifically, they claim that we reject robust genomic evidence on the grounds of limited sample sizes, while simultaneously proposing sweeping morphological synonymies based on visual assessment of single type specimens. This presumably referred to nominal species recently delineated or elevated from synonymy by some of their author group, the support for which we questioned. They mischaracterise this as privileging expert judgement over reproducible quantitative data.

This conflates two fundamentally different operations in taxonomy. A type specimen, by definition, cannot represent the full variability of a species; it is, in most cases, a single historical specimen that anchors a name. Its taxonomic significance lies not in representing variation, but in fixing nomenclatural application. When we, as indeed many other taxonomists, are considering a nominal species for synonymy, the type specimen of that nominal taxon is not assessed solely against the type specimen of another species. Rather, it is evaluated against the documented range of morphological and geographic variability of the putative parent species—variability that has been characterised through large comparative series of specimens and extensive field observations. The question is therefore whether the type of the nominal species falls within this established spectrum of variation (see Figure 25 in our Review for additional context).

By contrast, in several of the cases discussed by Cowman et al. (their Section 7; see Section 2.11.2 below and Supplementary Section S7), lineage-level molecular partitions derived from limited sampling are treated as sufficient to define new species boundaries, without first establishing the full extent of population-level variability or evaluating the representativeness of the type material. In these instances, the species boundary itself is being inferred from restricted data, rather than being evaluated against a comprehensively characterised biological framework.

These two evidentiary contexts are not comparable. The first evaluates name application within an existing, extensively characterised biological framework. The second proposes the creation or redefinition of species boundaries based on limited biologically relevant sampling, or in several cases only on comparisons of nominal type specimens. Critiquing the latter approach for insufficient breadth of evidence while applying the former within a long-established comparative context does not constitute a double standard; it reflects different evidentiary requirements for different taxonomic actions. For further clarification see Supplementary Section S4b; the specific species-level cases underpinning this distinction are addressed in detail in Supplementary Section S7 and Appendices SA1–SA4.

Cowman et al.’s claim for reproducible quantitative data is precisely what many museum collections provide, and to which our own collections have substantially contributed. In the context of coral genetics, however, different teams, each using modern molecular tools, are deriving different topologies (e.g., see Review Section 7.3.6). In some cases these differences are methodological; for example, the same nominal species may be recovered in different positions within alternative phylogenomic topologies, or associated with different species groups (e.g., refs. [3,5]), or may reflect differences in geographic scope, biological sampling framework, and taxonomic interpretation (e.g., refs. [6,7]). In others, the results may be due to real genetic structuring across geographic distance, where different lineages are recovered from populations which may nevertheless exhibit ongoing connectivity, albeit in some cases, occasional (see Section 2.6). This is entirely consistent with a reticulate evolutionary framework. The issue of reproducibility across different studies, however, is not only restricted to corals and can result in widely differing conclusions even with the same datasets (see Section 2.10 below for more general discussion).

2.4. ICZN, Alleged Abandonment of Typification, and Taxonomic Stability

Cowman et al. allege that we misinterpret or selectively apply the International Code of Zoological Nomenclature (ICZN) [8], invoking it to preserve familiar names while rejecting the principle of typification when it conflicts with our “biological entities”. They further argue that maintaining names that cannot be tightly linked to a type specimen creates “unanchored taxon concepts” that are effectively untestable. In certain cases, they propose that an ambiguous or incongruent historical type should be designated as a nomen dubium to “clear the nomenclatural landscape” and imply that this should take precedence over other considerations.

This also misrepresents our position. We do not reject typification, nor do we invoke the Code selectively. Our concern lies not with the principle that names must be anchored to types, but with how those types are interpreted in relation to accumulated biological knowledge. Where types are fragmentary, ambiguous, or potentially unrepresentative, interpretive disagreement is possible—and that disagreement can lead to synonymy, resurrection, or designation as nomen dubium (Supplementary Sections S4 and S5).

As introduced above, type specimens fix the application of names. They do not, by themselves, define the biological limits of species independent of comparative evidence. Over many decades, links between types and field-recognisable entities have been established through extensive comparative study of specimens across habitats and geographic regions. When the name in question is already securely associated with a well-characterised biological entity supported by extensive field and museum material, taxonomic actions should be undertaken with caution to maintain nomenclatural stability.

The ICZN Preamble makes clear that the Code exists to promote stability and universality. Consistent with this objective, the Code itself provides mechanisms—lectotype designation, neotype designation, conservation of usage —to prevent unnecessary disruption where strict literal interpretation, or indeed misinterpretation, of type material would sever established biological continuity.

We therefore do not defend “unanchored taxon concepts”. Rather, we argue that nomenclatural actions should not detach names from well-documented biological entities without simultaneously establishing a secure and biologically coherent replacement. Stability and typification are not opposing principles; both are fundamental objectives of the Code.

In our Review we discussed the specific example of the name Acropora microclados which has been associated with a long-established, well-documented, and well-illustrated field entity for many decades. The designation of this nominal name as a nomen dubium essentially severed the link between the name and the recognised biological entity to which it had long been applied, risking the loss of the accumulated biological knowledge associated with that taxon. Our intention was to highlight the problems resulting from such a severance, particularly in relation to the ICZN objective of maintaining nomenclatural stability, when no alternatives were offered simultaneously. For further discussion relevant to this section and our commentary above, see Supplementary Sections S5, S6 and S7g.

2.5. Sampling Scope, Type Specimen Context, and Genus and Species-Level Inference

Cowman et al. argue that genomic clustering and phylogenetic structure recovered from recent molecular datasets provide sufficient evidence to justify genus and species-level revision. They contend that failure to recognise distinct lineages risks “lumping” evolutionarily independent units, undermining ecological applications and conservation initiatives (their Sections 6 and 7).

The importance of biologically relevant sampling to species-level revision was a key issue in our Review, and it remains the most fundamental component of our response to Cowman et al. As such, it warrants particular focus here. We agree that failure to recognise genuinely distinct evolutionary lineages can have biological and conservation consequences. We also agree that genetic structure can provide powerful evidence of lineage differentiation. We fully support these endeavours, as was clearly stated in our Review. Our disagreement concerns the evidentiary threshold required to translate lineage-level differentiation into formal genus or species-level taxonomic change.

Biologically meaningful taxonomic inference depends critically on appropriate sampling design and representation of population-level variability [9,10,11,12], including contextualising of type specimens within such variability (see this Section below and Section 2.11, Supplementary Section S7, and Appendices SA1–SA4). These same considerations determine whether lineage-level genetic differentiation can be interpreted as evidence for formal genus- or species-level taxonomic revision.

Lineage structure within coral species can arise through a range of processes including geographic subdivision, habitat- and depth-associated differentiation, phenotypic plasticity, temporal reproductive structuring, hybridisation and introgression, incomplete lineage sorting, admixture, and local adaptation. Several of these processes are well documented in reef-building corals, and together they show that genetic, morphological, and ecological differentiation may arise within species, often in the absence of complete reproductive isolation (e.g., refs. [13,14,15,16,17,18,19]). Early population genetic studies further demonstrated that colonies exhibiting distinct morphologies may belong to the same genetic population, while conversely colonies with similar morphology may represent different genetic lineages [13,14].

Geographic and environmental gradients frequently generate population structure in widespread coral species. Habitat-associated differentiation—particularly with increasing depth, notably historically under-sampled environments such as mesophotic reefs—can produce distinct ecological lineages within populations [17,20,21]. Phenotypic plasticity, a pervasive feature of scleractinian corals [22], further complicates interpretation by allowing substantial morphological divergence with or without corresponding genetic isolation (e.g., refs. [23,24]).

Reproductive processes add an additional layer of complexity. Temporal structuring of spawning can reduce gene flow among sympatric lineages without eliminating it, while gamete compatibility, fertilisation success, larval viability, and hybrid fertility may vary across taxa and environments [25,26,27,28]. These barriers may themselves be environmentally influenced or context-dependent [29,30,31], allowing opportunities for gene flow to be re-established under changing conditions.

In addition, incomplete lineage sorting and admixture can generate discordant genetic signals among loci, while hybridisation and introgression may produce reticulate patterns of relatedness that are not adequately represented by strictly bifurcating phylogenies [15,16]. Functional differentiation, including variation in host–symbiont associations and physiological traits, may correlate with lineage structure and does not necessarily correspond to species-level divergence [32,33,34].

Taken together, these processes demonstrate that lineage structure within corals is expected under a wide range of biological scenarios and does not, in itself, constitute evidence of species boundaries. Rather, lineage partitions may reflect population structure, ecological differentiation, demographic history, or reticulate evolutionary processes. Only where such partitions are shown to be consistent across adequately sampled populations, concordant with morphological and ecological differentiation, and robust to alternative explanations, can they be considered strong evidence for species-level divergence. At present, for certain coral taxa, boundaries between reticulation, ongoing divergence and fully established species remain incompletely resolved. In these cases, premature conversion of lineage partitions into formal taxa risks generating unstable nomenclature that may require subsequent revision as biological understanding improves. We provided several examples in our Review (Sections 6 and 7).

Accordingly, interpretation of lineage partitions requires sampling designs that capture variation within and among populations across habitats, geographic regions, and sympatric assemblages [11,12]. Without such sampling, lineage partitions may reflect population structure, historical isolation, incomplete lineage sorting, or reticulate processes rather than species-level divergence.

Adequate sampling across populations and habitats is therefore not merely a technical detail of study design but a prerequisite for reliable species-level inference. In this context, Cowman et al. suggest that our interpretation “confuses population genetics with taxonomic revision.” This distinction is misplaced: the issue is not confusion, but evidentiary sufficiency and its biological interpretation. Species delimitation inherently depends on comprehensive sampling of variation within and among populations across habitats and geographic ranges, and population-level genetic structure provides one line of evidence bearing on species boundaries. However, its interpretation must be integrated with morphological variability, ecological differentiation and geographic context, including contact zones. Accordingly, population-genetic sampling and taxonomic revision are not separate analytical domains but complementary components of species-level inference.

Lineages can only be interpreted taxonomically if they are anchored to names, and names are anchored to types—therefore type context must be evaluated within the same biological framework as lineage data.

Interpretation of lineage partitions in taxonomic terms necessarily requires explicit linkage to type specimens, which anchor the application of names under the ICZN Code [8]. While type specimens remain essential for nomenclatural stability under the Code, this linkage cannot be achieved through comparison with type material in isolation but must instead be evaluated within the full biological context of the taxon, including its documented morphological variability, ecological range, and geographic distribution.

A holotype, lectotype, or neotype, being a single specimen, cannot represent the full variability of a species. Historical descriptions are similarly limited, as they typically document only the individual specimen used to anchor the name. These early descriptions did not include accounts of the living species or of ecological, population, or geographic variability. Consequently, neither the type specimen nor its original description can be taken to define species-level variability without broader contextualisation.

This contextualisation is particularly critical in corals, where type specimens may not exhibit diagnostic features observable in fully developed colonies in situ. Their morphology may also reflect specific environmental conditions or growth stages, and they may be small or only part of a fully developed colony. Without explicit characterisation of how such specimens relate to variation within populations at and beyond the type locality, their representativeness may remain uncertain.

Recent descriptions can have limitations for different reasons. These typically represent observations of a population in situ but are initially based on relatively few specimens, particularly where the newly described taxon is rare. Although the nominal species in these cases will be considered distinct from previously described species by their authors, its range of morphological variability, occurrence across habitats, and geographic distribution may not yet be characterised.

The most reliable descriptions therefore tend to be of taxa that have been long-known and well-studied across habitats and geographic space, both in situ and from skeletal material. For these, many taxonomists, working at different times and locations, have provided independent treatments, contributing to understanding and stability.

Accordingly, interpretation of type material must be integrated with sampling designs that capture variation within and among habitats, across local populations, sympatric assemblages and—where species-level revision is proposed—across the known geographic range of the target taxon. In the absence of such integration, there is a risk that type specimens—and the names they anchor—are interpreted within a restricted framework that does not reflect the broader biological reality of the species. Resulting search images for species in the field may also be constrained to a restricted interpretation derived primarily from the type specimen rather than the full documented variability of the species.

Cowman et al. in their Section 7, specifically cite examples from Bridge et al. (2023) [3] and Rassmussen et al. (2025) [7]; it is therefore towards these papers that we direct the following comments. Available evidence from the methods and taxonomic accounts included in these papers indicates that type specimens, associated topotypes and derived sequences have been interpreted within the limited frameworks discussed above. There was no explicit demonstration of how these relate to the documented variability of the species at the type locality and across its broader range (see also quoted text from the relevant papers in Supplementary Appendix SA1).

While this approach provides a practical means of linking genomic data to nominal taxa, it has been acknowledged by the authors themselves that it “does not necessarily capture the intraspecific variation required to delineate species boundaries in the field” (Bridge et al., 2023: p. 6) [3].

In other cases, historical descriptions of type specimens have been treated as if they characterise species-level variability rather than the individual specimens on which they are based, and variability described in the literature has not been fully incorporated into the interpretation (see Section 2.11 below and Supplementary Sections S1e, S7 and Appendix SA1). In certain cases, these interpretations led to the resurrection of junior synonyms based on type specimens and relatively minor distinctions in morphology of those types, without documented evidence of biologically relevant variability in either the parent species or the resurrected synonym (see Appendix SA1 for further discussion).

Given that taxonomic names are anchored to type specimens, such limitations in type contextualisation can directly affect the interpretation of lineage partitions and therefore the validity of subsequent genus- or species-level revisions.

Further, in considering the revisions discussed by Cowman et al. (their Section 7), the concept of replication requires careful distinction. Replication has two fundamentally different meanings: technical replication—the number of loci, markers, or samples required to stabilise an inferred topology or clustering pattern within a given analytical framework (e.g., ref. [35]); and biologically relevant replication—sampling across habitats, geographic regions, sympatric congeners, and intervening populations sufficient to characterise natural variability (e.g., ref. [36]).

While recent genomic datasets may achieve high levels of technical replication, this does not, in itself, ensure that biologically relevant variation has been adequately captured. A topology may therefore appear analytically stable yet remain biologically uninformative for species-level inference. In morphologically variable and geographically widespread taxa such as reef-building corals, biologically relevant replication is essential to determine whether lineage partitions correspond to diagnosable species or instead reflect population structure, ecological differentiation, or reticulate evolutionary processes.

The considerations regarding sampling and the interpretation of lineage partitions apply not only to species delimitation, but also to the interpretation of species distributions and to higher-level taxonomic revision. In each case, inference depends on adequate sampling across the geographic, ecological, and morphological range of the taxa concerned.

For species distributions, absence of particular lineages or morphotypes in restricted sampling cannot be taken as evidence of true absence without comprehensive geographic and ecological coverage. Distributional limits inferred from limited datasets may therefore reflect sampling gaps rather than underlying biological reality. This is not to suggest that some lineages will not reflect genuine evolutionary independence once studies of population and geographic structure, particularly across contact zones, have been undertaken; rather, that such investigations are required prior to species recognition and elevation.

Similarly, genus-level revisions require evaluation of character states and lineage relationships across the full range of variation within and among constituent species. Where such variation is incompletely sampled, apparent phylogenetic or morphological distinctions may not be stable under broader biological representation. As with species-level inference, robust higher-level revision depends on the integration of genomic, morphological, ecological, and geographic evidence within a sufficiently comprehensive sampling framework.

Characterising variability across habitats, geographic regions, and sympatric assemblages requires substantial investment of time, resources, and taxonomic expertise. To date, no coordinated body of molecular studies has evaluated any coral taxon across their full geographic, ecological, and morphological ranges to a degree sufficient for confident species-level revision on genomic evidence, with or without morphological support. Rather, existing studies have provided important but as yet incomplete insights into processes such as population structure, connectivity, depth-associated differentiation, and holobiont variation (e.g., refs. [6,37,38,39,40,41,42,43,44,45,46,47,48,49,50]), several of which include authors of Cowman et al. themselves. Taken together, these studies have identified significant lineage complexity within species, and across species boundaries, along with apparently conflicting results. We illustrated examples of this complexity in our Stylophora case study (Review Section 4.7) and discussed conflicts elsewhere (Review Sections 6 and 7).

In summary, lineage-level genetic differentiation, while highly informative, cannot be interpreted in taxonomic terms without adequate biological context. Robust genus- and species-level inference requires the integration of genomic data with sampling designs that capture variability across habitats, geographic regions and sympatric assemblages, together with explicit contextualisation of type material within that variability. Where these conditions are not met, lineage partitions may reflect population structure, ecological differentiation or reticulate evolutionary processes rather than species-level divergence. As stated in our Review, taxonomic inferences thus derived, while potentially useful, should be considered as provisional, providing guidance for further detailed work, rather than sufficient for significant taxonomic revision per se.

Our consistent call has been for biologically relevant study design, and restraint to avoid premature destabilising taxonomic actions. As noted above, our purpose was, and is, to ensure that taxonomic and nomenclatural decisions are robust. Biologically relevant sampling, and its implications for species inference, were a major theme of our Review and remains central to our discussions throughout this Reply.

2.6. Reticulate Evolution and Model Testing

Cowman et al. characterise our emphasis on reticulate evolution as an attempt to insulate morphological “biological entities” from genetic falsification, suggesting that reticulation is invoked as an untestable axiom whenever molecular results conflict with established taxonomy.

This characterisation reflects a fundamental misunderstanding of our position. We do not invoke reticulate evolution as a default explanation when there is molecular–morphological incongruence. Rather, we question molecular–morphological incongruence when the evidence is lacking for the inferences presented. We do, nevertheless, reiterate our view that reticulation is a key evolutionary process in reef-building corals, and other groups. Indeed, reticulate evolutionary processes are widely documented across the tree of life, including in plants such as Eucalyptus (e.g., refs. [51,52]), birds (e.g., ref. [53]), reptiles (e.g., refs. [54,55]), insects (e.g., refs. [56,57]), and in marine organisms such as the seaweed Halimeda [58], fish [59], and mussels [60] among many other groups.

Processes associated with reticulate evolution are frequently reported in molecular and reproductive studies of reef corals. Evidence consistent with reticulate evolutionary histories has been documented in several non-Acropora genera. For example, molecular analyses of Montipora have revealed patterns interpreted as probable reticulate evolutionary histories [61]. Similarly, studies of the Stylophora pistillata species complex have identified semi-permeable species boundaries and patterns consistent with introgressive hybridisation and reticulate evolution [62]. Population genomic analyses of Pocillopora have likewise demonstrated introgressive hybridisation among closely related lineages [63], while experimental fertilisation studies have shown that hybridisation among broadcast-spawning coral species can occur in various genera, including Montipora and Platygyra [64]. Similar examples have been documented in the Caribbean (Review Section 3.1.2).

Within Acropora, hybridisation and introgression have long been documented, including among the Caribbean Acropora palmata, A. cervicornis, and their derivative A. prolifera [65,66]. In the Indo-Pacific, fertilisation success among twelve species of Acropora on the Great Barrier Reef was observed for 16 interspecific pairings, and more than 80% of the successful crosses developed into apparently healthy hybrid larvae [64]. In the Acropora aspera group, differences in spawning time and reproductive compatibility have been shown to coexist with natural hybridisation and semi-permeable species boundaries [15].

Of particular relevance to this Reply, similar complexities have also emerged in the Acropora tenuis complex, where population genomic and phylogeographic studies have revealed substantial lineage structure and cryptic diversity across broad geographic regions [67,68], including genetically differentiated populations occurring in sympatry and across environmental gradients. Such studies indicate that lineage partitions within the A. tenuis complex may reflect population structure, ecological differentiation, and historical gene exchange rather than simple bifurcating divergence. Collectively, these studies demonstrate that processes consistent with reticulate evolution are widely documented in reef corals.

We do not attempt to insulate biological entities from scrutiny as implied by Cowman et al. Indeed, we expect there to be considerable intra-entity genetic structure, some of which may be raised to species level justifiably in future. On the other hand, the theory of reticulate evolution in corals was developed as a direct result of detailed study of living corals, because it was highly explanatory of the observations made in the field [4]. There is therefore an expectation that the species identified in Veron (1995: pp.279–285) [4] and in subsequent publications as most likely to exhibit reticulate patterns—or conversely as unlikely to do so—may display these predicted characteristics when examined genetically. For example, Veron (1995: p.285) [4] commented in the context of reticulate processes that Acropora hyacinthus and A. cytherea were part of a complex system where the two species were relatively easy to distinguish on the Great Barrier Reef but seemed to coalesce in the Pacific. That prediction proved prophetic and also has particular relevance to our Reply.

Among reef corals, the Acropora hyacinthus–cytherea complex provides one of the clearest and most extensively studied examples of reticulate processes, with multiple studies documenting extensive sympatry, cross-fertility, introgression, and lack of reciprocal monophyly among lineages. Early molecular work demonstrated that colonies identified as Acropora hyacinthus and A. cytherea are highly cross-fertile and genetically similar despite consistent morphological differences [69]. Subsequent population genetic analyses revealed extensive sympatry and cryptic diversity within this complex, together with evidence of introgression and a lack of reciprocal monophyly among lineages [6]. These studies showed that genetically differentiated lineages can occur within the same reefs and across large geographic regions, while still sharing substantial genetic variation and showing evidence of introgression, with ongoing gene flow likely in at least some cases. Such complexities urge the need for caution and strong support for taxonomic changes.

Later genomic studies have continued to document complex lineage structure within the Acropora hyacinthus group across the Indo-Pacific. Analyses of population genomic datasets have identified multiple genetic clusters occurring in sympatry and across environmental gradients, again without clear correspondence between lineage partitions and traditional species boundaries [40]. Of particular relevance to the case studies discussed in Section 2.11 is the inclusion, in this study of Acropora cytherea, of samples from Tanzania together with Pacific Island populations. Although the Tanzanian samples were genomically differentiated from the Pacific populations, they nevertheless fell within the same broader lineage as one of the recognised Pacific clusters (C2), suggesting regional population divergence within a wider lineage framework, rather than an isolated or unrelated lineage.

Regional studies have likewise revealed differentiated lineages associated with ecological and geographic variation [41], while experimental and reproductive studies have documented semi-compatibility among lineages within the complex [38]. Taken together, these studies indicate that the Acropora hyacinthus–cytherea complex exhibits extensive sympatry, cross-fertility, lineage structure, and introgression across broad geographic scales, characteristics consistent with a reticulate evolutionary system in which genetically differentiated lineages can persist despite ongoing gene exchange.

In systems of this kind, species-level revision requires sampling designs capable of distinguishing among lineage structure, population differentiation, and ongoing gene exchange across the geographic and ecological range of the complex. The extent to which such conditions are met in the case studies discussed below therefore warrants careful examination.

In summary, our position is not that reticulation overrides molecular evidence, but that reticulate processes have significant support. In this context, molecular inference must consider the full range of plausible evolutionary processes before species boundaries are formalised. For case-specific responses see Supplementary Sections S2, S6 and S7.

2.7. Interpretation of Morphological Characters and Phylogenetic Signal

Cowman et al. argue that integrative analyses frequently demonstrate congruence between morphological and molecular datasets, particularly where micromorphological or microstructural characters are used, and that such congruence provides a robust and reproducible basis for formal taxonomic action (see their discussion of integrative frameworks).

To be clear, where such congruence has been clearly demonstrated with adequate support, we have accepted the findings. Those cases were not detailed in our Review, which dealt principally with the biological, ecological, biogeographic, and evolutionary underpinning of coral taxonomy, and issues arising. We did, nevertheless, state (Review Section 7):

“Although this section deals with discrepancies, we note that we have accepted some of the recent shuffles of species across genera (see CoralsOfTheWorld.org (2026 in prep.) [236]) where these are warranted from both molecular and morphological viewpoints. We also note, however, that overt conflicts between morphology and molecular results require careful review.”

Similarly, we also accepted some recently described species, as detailed in the relevant species factsheets of CoralsOfTheWorld.org (2016, 2026 in prep.) [70,71]. Congruence among independent lines of evidence strengthens taxonomic inference when datasets are evaluated rigorously and interpreted appropriately.

Morphometric character matrices may include major architectural features that define colony organisation alongside minor, potentially labile traits without explicit consideration of their differing biological significance. Statistical analyses typically weight such characters equally unless otherwise specified. Equal numerical weighting, for various macro- and micro-morphological features does not necessarily reflect equal phylogenetic informativeness (see Supplementary Sections S6 and S7).

Major architectural characters—such as meandroid versus phaceloid colony integration and wall structure—reflect fundamental developmental organisation and likely capture underlying evolutionary relationships, even though aspects of their expression may remain responsive, to varying degrees, to local environmental conditions [72,73,74]. In contrast, certain micromorphological traits—for example those expressed in corallite architecture—may also carry biological information but are often more sensitive to positional, ontogenetic and environmental influences: see for example, Veron and Wallace (1984) [75], Wallace (1999): pp. 50–58 [76] and our Review Figure 48.

Even characters regarded as diagnostic rarely occur as a single fixed state. Rather, they tend towards a modal expression—that mode being the character used as the diagnostic feature—with variation occurring both within and among colonies across a species’ range (e.g., refs. [22,75,76,77].

In Acropora, for example, branching architecture, axial corallite extension, radial corallite form, and other skeletal features may differ within a single colony depending on position and growth history, and commonly vary among colonies occupying different habitats (e.g., refs. [75,76,78,79] and see Review Supplementary Figure S1). Such intracolonial variation can include the coexistence of multiple corallite forms and character modalities within a single colony, rather than discrete, non-overlapping character states. Although the degree of variability differs among species, inter-colony variation in macro- and micro-morphological characters is common in most corals [4,22,76].

Importantly, this pattern of modal expression with overlapping intra- and inter-colony variability means that individual specimens—including type material or fragments thereof—may capture only a subset of the total character space expressed by a species (as outlined in Section 2.5). Consequently, interpretation of diagnostic characters requires explicit consideration of this variability before characters are treated as fixed or discrete.

When a morphometric subset of such characters is treated equivalently—particularly where sampling is limited or ecological context is not considered—interpretation can be disproportionately influenced by traits that are developmentally plastic or environmentally responsive. As outlined below (Section 2.11) and discussed in Supplementary Section S1e in relation to the synonymy of Symphyllia with Lobophyllia, major integrative character systems such as ‘phaceloid’ or ‘meandroid’ colony architecture could themselves introduce biases if weighted more heavily. The difficulty is not that differential weighting would be unjustified, but that the appropriate extent of such weighting would inevitably be subjective. It would nevertheless be instructive to evaluate how morphological topologies change when alternative weighting schemes are applied. We emphasise that treating all characters as equivalent already represents a significant analytical assumption that is not necessarily biologically justified.

Under such conditions, apparent “synergy” between morphological and molecular datasets may reflect non-independent corroboration (e.g., refs. [80,81] and see Supplementary Section S1e), analytical constraints (see Section 2.10), or misinterpretations (see Section 2.11 and Supplementary Section S1e). Interpretation of morphometric data therefore requires explicit consideration of the phylogenetic significance of characters, their developmental context, and ecological variability, particularly in morphologically plastic coral taxa.

A further complicating factor is the relationship between some variable morphological characters and genotypes. In some cases, morphological characters have been shown to vary across genotypes and/or lineages [68,69], and in others, within genotypes and/or lineages (e.g., ref. [82]; and see refs. [14,22]). Thus, an understanding of these processes in the species concerned is a prerequisite for interpretation of lineage and species boundaries. Our concern is not with the use of quantitative morphology, but with the assumption that such analyses consistently resolve differences of biological significance among characters without the detailed corroboration of supporting studies. As with molecular data, morphological datasets require biologically and ecologically informed interpretation before they are translated into species-level taxonomic decisions (see Section 2.11 below and Supplementary Sections S6 and S7).

Bridge et al. (2023: p.5) [3] state:

“We use the unified species concept (de Queiroz 2007), which defines a species simply as an ‘independently evolving metapopulation lineage’, and therefore allows the delineation of distinct evolutionary lineages without requiring species to be distinguishable morphologically or to exhibit intrinsic reproductive isolation.”

(our bolding)

This interpretation is misleading. De Queiroz (2007) [83] was careful to distinguish the concept of species as ‘separately evolving metapopulation lineages’ from the operational criteria used to infer them. Those criteria include, among others, reproductive isolation, morphological diagnosability, ecological distinctness, monophyly, genetic differentiation, and geographic distribution. He did not suggest that recovery of lineages in a particular analytical framework is, on its own, necessarily sufficient for delimitation, particularly where biologically relevant sampling is incomplete. Rather, he emphasised that multiple lines of evidence provide stronger corroboration of lineage separation and that geographic variation is crucial in delimitation.

Although cryptic lineages are expected in cases of recent divergence or where lineages continue to exchange genes, species delineations that lack clear morphological distinctions should require high evidentiary standards. Such delineations are therefore not appropriate where sampling across relevant spatial scales—particularly geographic regions and contact zones—is incomplete.

2.8. Convergence and the Interpretation of Morphological Concordance

Cowman et al., echoing several recent molecular treatments, interpret incongruence between molecular results and traditional morphology as evidence that extensive morphological similarity among lineages must reflect convergence or homoplasy. In this view, discordance is taken to imply repeated independent evolution of similar forms. Convergence is a recognised evolutionary phenomenon, and phenotypic plasticity is common in corals. However, the explanatory weight placed upon convergence must be proportionate to the extent, integration, and biological significance of the similarities under discussion (Figure 1, and see Supplementary Sections S1f, S6 and S7).

Isolated trait similarity may be readily attributable to functional convergence under comparable environmental pressures. In contrast, many of the species-level entities of Bridge et al. (2023) [3], Rassmussen et al. (2025) [7], and Cowman et al. assessed here exhibit coordinated similarity across multiple skeletal systems—including colony-level architecture, corallite integration, septal patterning, and wall construction—involving characters that are developmentally and structurally interrelated (see Supplementary Section S7).

While independent evolution of individual characters is well documented across the tree of life, repeated parallel reconstruction of integrated skeletal frameworks in widely separated species would require coordinated re-emergence of multiple character complexes. Such examples, of near-identical phenotypic expression, are clearly different to superficial morphological resemblance (Figure 1). These require stronger evidentiary support—beyond the presence of molecular incongruence—before alternative explanations for morphological continuity can be set aside. Plausible alternatives—such as geographically widespread species exhibiting lineage-level variation—require explicit evaluation before being rejected. Indeed, there are numerous examples of widespread coral species with populations extending across the Indo-Pacific region. In our Review we gave Diploastrea heliopora as the archetypal example (see Reply Supplementary Figure S2).

Where the genetic signal is itself heterogeneous across loci (Section 2.5; Supplementary Section S7), alternative evolutionary processes provide feasible explanations for morphological continuity without requiring repeated independent assembly of nearly identical structural systems. Accordingly, we do not deny convergence. Rather, we question its routine attribution as a default interpretation of discordance, particularly where morphological similarity spans multiple character domains and broad geographic distributions.

2.9. Higher-Level Phylogeny Versus Species-Level Entities

Cowman et al. argued that recent molecular research is “fundamentally incongruent” with much of the taxonomy synthesised in Veron (2000) [84]. They disputed our statement that there have been “few fundamental challenges to the overarching biological units designated as species” (our Review Introduction), asserting that genomic-scale data reveal comparable instability at species level. They further claimed that the species-level taxonomy is “just as problematic” as the higher-level framework.

Firstly, as we point out in the Supplementary Text (Sections S1b–d), Veron (2000) [84] was published more than twenty-five years ago, and much has changed since then in all fields. As far as higher-level phylogenetic structure is concerned, molecular studies have substantially revised family-level and deeper relationships. These advances were specifically acknowledged in our Review (e.g., see our Review quote in Section 1 above). Higher-level phylogenetic rearrangements are not the focus of CoralsOfTheWorld.org (2016, 2026 in prep.) [70,71] and were not the subject of critiques in our Review.

If, however, Cowman et al.’s implication is that most field-recognisable species-level entities lack biological coherence or are broadly undermined by current genomic data, we disagree, and it was largely to species-level issues that our Review commentary was directed. Much of the genomic diversity now being revealed reflects fine-scale lineage structuring within these entities rather than wholesale reorganisation of the species concepts themselves. Such structuring is unsurprising, and was already foreshadowed in earlier work on coral biogeography, population structure, reproductive differentiation and speciation in corals (e.g., refs. [14,85,86,87]), before being discussed explicitly in relation to coral species boundaries by Veron (1995) [4].

This distinction has important practical implications. The relative robustness of these long-recognised species has provided a critical framework for molecular studies, allowing specimens to be assigned, at least provisionally, to broadly comparable biological field units. Interpreting the biological complexity revealed by genomic data therefore requires careful integration of molecular evidence with the natural variability of corals as observed in the field. Coral species recognition, both in the field and in the laboratory, requires substantial training and field experience, as is the case for other morphologically variable and ecologically diverse taxonomic groups. Without this broader biological perspective, type specimens may not be interpreted within the full range of variation of the taxa they represent, and study designs, topotype selections, and subsequent inferences may therefore be compromised (see Supplementary Section S7 for examples). Our Review drew attention to the importance of such context and emphasised that a focus on field experience is an essential component of all molecular studies pursuing taxonomic insights.

2.10. Interpreting Phylogenomic Structure Under Gene-Tree Conflict and Incomplete Biological Sampling

Recent phylogenomic studies have increasingly emphasised that genome-scale datasets frequently contain extensive gene-tree conflict and that species-tree inference must therefore be interpreted cautiously. Gene-tree discordance can arise from multiple biological and analytical processes, including incomplete lineage sorting, hybridisation and introgression, model misspecification, orthology error, and gene-tree estimation error. No single analytical framework can fully accommodate all of these sources simultaneously [88]. In large phylogenomic datasets, high bootstrap support can also coexist with low genome-wide concordance, such that strongly supported branches may nevertheless represent only a minority of gene histories [89]. Consequently, modern phylogenomic practice increasingly emphasises explicit evaluation of concordance factors, conflict metrics, and alternative topological signals rather than treating a single recovered topology as definitive [90].

Recent work has also demonstrated that analytical decisions can influence evolutionary interpretations even when identical datasets are analysed. In a collaborative study reported by Gould et al. (2025) [91], multiple research teams independently analysed the same evolutionary dataset using their preferred analytical approaches and reached differing conclusions regarding evolutionary relationships. These differences arose from variation in data filtering, model specification, and analytical workflows, illustrating that evolutionary inference may be sensitive not only to biological signal but also to analytical decisions. Such considerations are particularly relevant to molecular studies of reef corals, where choices regarding genomic markers, filtering thresholds, clustering approaches and species-delimitation frameworks must be aligned with the biological questions being addressed. Gould et al. (2025) [91] further demonstrated that conclusions may change markedly when analyses are conducted on restricted subsets of the available data: in their case study, analyses based on radically smaller subsets produced dramatically divergent effect sizes, reflecting both sampling error and biological heterogeneity within the full dataset. This finding highlights the critical importance of adequate geographic, ecological, and population sampling when interpreting genomic partitions in taxonomic contexts.

Phylogenetic structure may also be strongly influenced by incomplete sampling of the biological variability present within taxa. Restricted geographic, ecological or population sampling can produce topological patterns that appear well resolved but fail to capture the full spectrum of intraspecific variation, thereby creating artificial lineage separation or misleading phylogenetic structure when broader sampling is undertaken (see Section 2.5 above). This interaction between sampling scope and inferred character distributions has long been recognised in phylogenetic analysis [9]. It remains particularly relevant in organisms such as reef-building corals where morphological and genetic variability can be structured across habitats, depth gradients, and geographic regions.

These factors—analytical decisions, genome-level conflict, and incomplete sampling of biological variability—may also interact, such that limited sampling can exaggerate apparent lineage separation or obscure signals of gene flow that would become evident with broader geographic and ecological coverage. These methodological considerations are not merely hypothetical; they are directly relevant to several recent coral taxonomic revisions. In the following Section, we briefly outline the principal case studies referred to by Cowman et al. in which these issues arise. We address them in more detail in our Supplementary Text (Section S7).

2.11. Overview of Case Studies Discussed by Cowman et al.

Before addressing summaries of our key concerns with these case studies, we reiterate, as we did in our Review, that the studies under discussion offer insights into genomic variability of the groups they are examining. As emphasised throughout, our comments focus on specific issues with the studies, particularly relating to inferences beyond their evidentiary support. Cowman et al. address specific genus and species revisions in their Sections 1 (Symphyllia/Lobophyllia), and 7 the latter section focused on the studies of Bridge et al. (2023) [3] and Rassmussen et al. (2025) [7].

2.11.1. Symphyllia and Lobophyllia

We deal with the synonymy of Symphyllia with Lobophyllia in further detail in Supplementary Section S1e. In summary, this synonymy rests on several methodological limitations. In general terms, these include reliance on morphometric matrices that do not distinguish between minor structural traits and fundamental colony architecture (Figure 2 and Section 2.7 above), over reliance on characters of type specimens without full characterisation of their biological context (and see Supplementary Section S7 and Appendices SA1–SA4), limited genomic sampling (see Section 2.5 above), and phylogenetic trees that exhibit low support and incongruence at the internal nodes relevant to generic boundaries (see Section 2.10 above).

In this context, repeated recovery of a single clade demonstrates close evolutionary affinity within the constraints of the dataset but does not constitute decisive evidence against the long-recognised major architectural distinction between the phaceloid Lobophyllia and the massive meandroid Symphyllia. Rather, the available evidence is consistent with limited resolution under current sampling and marker regimes, and potential misinterpretation of clear morphological incongruence, rather than positive demonstration of generic non-separation.

In particular, the following points are of relevance (for further detail see Supplementary Section S1e):

• Developmental differences in colony form from juveniles to adults is common for species of both genera. However, adult colonies of Lobophyllia hataii (and all other Lobophyllia) are clearly differentiated from the massive meandroid Symphyllia valenciennesii, as indeed are all other Lobophyllia from Symphyllia.
• Juvenile and small colonies of Symphyllia valenciennesii and Lobophyllia hataii can share morphological similarity. Each develops along a growth trajectory leading to distinct mature morphologies, the former becoming distinctly massive and meandroid (like all other Symphyllia) and the latter flabello-meandroid to sub-phaceloid with deeply divided walls (like all other Lobophyllia).
• The syntypes of Symphyllia valenciennesii Milne Edwards and Haime, 1849 are mid-growth stage specimens (e.g., Figure 3 left).
• The holotype of Lobophyllia hataii Yabe, Sugiyama and Eguchi, 1936 is a juvenile colony (e.g., Figure 3 right).
• In their morphometric analysis Huang et al. (2016) [92] coded all Symphyllia valenciennesii specimens as phaceloid, although this species does not develop a phaceloid form at any stage of its growth. Rather, it may exhibit intercalicular grooves, which are typically minor but can appear relatively broad in mid-growth stages.
• Miscoding most plausibly reflects misinterpretation of intercalicular grooves as phaceloid wall separation, particularly since syntypes are not yet mature and exhibit a distinct intercalicular groove.
• Misclassification of this fundamental architectural character may have accounted for the anomalous placement of Symphyllia valenciennesii in the morphological topology (Figure 2).
• Confusions caused by similarity of these two species in their developmental stages may also have contributed to morphological and molecular discordance in the dataset.
• Huang et al. (2016) [92] did not recover strong internal resolution within the relevant Lobophyllia clade. Although the broader clade was recovered, most species-level relationships within subclade I—including taxa then assigned to Lobophyllia and Symphyllia—were unresolved. The implications are discussed further in Supplementary Section S1e.
• Further targeted molecular sampling across habitats and geographic ranges will be required to resolve these relationships robustly.
• Without substantial field experience, juveniles of many, if not most, Lobophyllia and Symphyllia species, and indeed many other corals, can be difficult to identify in early stages of their development. Accordingly, unless studies are explicitly designed to address ontogenetic variation, we recommend that molecular sampling in these genera should specifically target fully developed, mature colonies.

We discuss the consequences of over-dependence on morphometric data and topologies where biological context is not fully included in interpretations in Section 2.7 above. This case also illustrates the risks of reliance on type specimens in isolation from broader biological variability of their taxon (see our Review Section 5.1.1; and Section 2.5 above).

In summary, the synonymy of Symphyllia and Lobophyllia is not adequately supported by the evidence currently available. Furthermore, the continued recognition of these genera provides clear and practical advantages for identification of their constituent species.

2.11.2. Acropora tenuis and Acropora hyacinthus

The two studies, addressing the Acropora tenuis complex by Bridge et al. (2023) [3] and the Acropora hyacinthus complex by Rassmussen et al. (2025) [7], have provided molecular insights from the samples analysed. However, some related taxonomic actions were not sufficiently supported, requiring further study, as discussed in our Review Section 7.3.7, in Supplementary Section S7, and its Appendix SA1, and summarised here below.

The purpose of our Review was to present general cautions and provisos and to illustrate them briefly using selected examples from the literature. These were included only to demonstrate that the issues discussed arise in real analyses and are not hypothetical concerns. However, the criticisms and mischaracterisations presented in Cowman et al.’s response make it necessary for us to address these two studies more explicitly.

Both papers inferred that the recovered phylogenomic lineages represented substantially greater species diversity than recognised in earlier taxonomic treatments, and both translated those lineage partitions into formal taxonomic actions—including new species descriptions and removal of nominal taxa from synonymy. As outlined in our Review (principally Sections 7.3.5–7.3.7) and in greater detail in our Supplementary Section S7 and Appendices SA1–SA4, we consider that some assessments of the foundational taxonomic literature on which earlier species concepts were based did not correctly interpret the relevant texts. These interpretations appear to result from misunderstandings of species variability and over-reliance on type specimens without full consideration of their biological context.

Some level of future taxonomic refinement in these groups is likely and, where adequately supported, appropriate. Indeed, our own observations, together with the documented levels of hybridisation, morphological and genetic variability, and the widespread distribution of these species, indicate that such complexity is to be expected (Review Sections 3.1.2, 5.2 and 7.3.7; and see Veron, 1995: p. 285 [4]; 2000: p. 306 [84]). However, as discussed in our Review, the specific refinements proposed by Bridge et al. (2023) [3] and Rassmussen et al. (2025) [7] are, at present, insufficiently supported and premature.

Our commentaries in earlier sections and the supporting citations therein provide documented support for biological and methodological prerequisites relevant to species-level taxonomic revision. Although not exhaustive,

Table 1. Summary of potential limitations arising from methodological and interpretative inferences.

Methodological Step	Potential Interpretative Limitation	Relevance to the Studies Under Discussion
Type specimens are located and used as the starting point for taxonomic interpretation	Historical types may be juvenile, fragmentary, atypical, or otherwise unrepresentative of the broader species	Type condition and representativeness may affect how names are linked to modern samples
Original descriptions based on single or few specimens are treated as species diagnoses	Historical descriptions often capture only single specimens; recent descriptions may not capture species variation	Limited contextualisation of original descriptions may affect how names are linked to modern samples
Topotypes are selected primarily on the basis of resemblance to the type	Topotype selection depends on prior interpretation of the type and may not reflect broader local variability	Single morph-selected topotypes may disproportionately and inappropriately anchor names
Limited numbers of specimens are chosen for molecular analysis	Small sample sizes may not capture within-population, habitat or depth-related structure	Recovered lineages may reflect incomplete representation of biological variability
Ecological documentation of samples is limited	Habitat-linked or depth-linked structure may not be distinguished from species-level divergence	Environmental context may be under-characterised relative to known coral plasticity
Sampling is centred on the type locality, or inferred locality	The type locality alone may not capture the range of variation within a widespread species	Population-level context at and beyond the type locality may remain insufficiently characterised
Morphometric analyses are conducted on a small number of colonies or fragments	Small morphological datasets may not reflect intracolonial or population-level variation	Apparently diagnostic differences may fall within broader uncharacterised variability
Morphological matrices treat characters as equivalently weighted	Characters of very different developmental, architectural or ecological significance may be weighted alike	Minor variable characters may disproportionately influence interpretations
Genome-scale phylogenetic analyses recover distinct lineage partitions	Lineage discovery does not by itself establish species boundaries	Genetic partitions may represent population structure, ecological structure or reticulate history
Clustering and delimitation methods are applied to the same underlying dataset	Concordance among analytical methods may reflect technical consistency rather than biological replication	Multiple analyses do not substitute for broader biologically representative sampling
Morphology is examined after lineage recovery	Morphological assessment may function as interpretation of predefined lineages rather than an independent test	Topology-first workflows require caution when used for species validation—risks confirmation bias
Low gene concordance and/or admixture are present in the dataset	Discordant loci may indicate incomplete lineage sorting, introgression or other complex processes within species	Bifurcating topologies may not adequately represent the evolutionary history of the group
Lineages are interpreted as species without explicit testing of alternative evolutionary processes	Lineage partitions may reflect incomplete lineage sorting, introgression or population structure rather than species boundaries	Additional demographic and reticulation analyses required before taxonomic revision
Diagnostic morphological characters are identified for recovered lineages	Small differences may be environmentally responsive or fall within broader species variability	Putative diagnostic characters require testing against wider sampling
Historical synonyms are revisited in light of the recovered topology	Type comparison and geography may be used where direct molecular evidence is unavailable	Nomenclatural decisions may extend beyond the taxa directly sampled in the phylogeny
Some nominal taxa are resurrected from synonymy without direct sequencing or population level studies	Taxonomic actions may rely on morphology, geography, or type interpretation alone	Evidentiary basis is minimal and non-comparable to sequenced lineages; insufficient for species designation
Geographic separation is used as part of species interpretation	Geographic distance alone does not establish species boundaries in widespread corals	Intervening populations and broader regional sampling are required to assess gene flow
Taxonomic actions are formalised following lineage recovery and interpretation	Species-level nomenclatural changes may outpace biological characterisation of the taxa involved	Revisions premature where broader variability and alternative processes remain untested
Museum collections are potentially reinterpreted in light of the revised taxonomy	Unsupported or weakly supported revisions may propagate through reference collections	Subsequent identifications may inherit propagated uncertainty from the revisions

summarises those most directly relevant to the studies under discussion. It allows readers to assess how closely the available evidence supports the proposed species-level revisions and resurrection of synonyms.

When examining the summary below and Table 1, we stress that the implied requirements are not necessary for all studies, but they are necessary for studies which progress to species-level revisions. Specific evidence for the points made below can be found in Bridge et al. (2023) [3] and Rassmussen et al. (2025) [7], particularly their Methods and taxonomic accounts. Further supporting discussion can be found in the cited sections and in Supplementary Section S7 and Appendices SA1–SA4). In summary:

• In several cases the available evidence presented suggests a strong reliance on type specimens without adequate contextualisation:

  ○

  Certain lectotypes and holotypes have been interpreted with very limited documented consideration of their broader context and representativeness, including whether they represent whole or fragmentary colonies, their ontogenetic state, the morphological expression and associated habitat they may represent, and their overall position within the documented variability of the species (see Section 2.5);

  ○

  Historical descriptions have in some instances been treated, in their published interpretations, as if they characterise species-level variability, rather than the individual specimens to which they relate (see Section 2.5);

  ○

  This overall limited interpretation of types and original descriptions has led to variability described in the literature not being incorporated into the resulting interpretations (Supplementary Appendices SA1–SA4);

  ○

  Such interpretations also raise the possibility that similarly restricted search images may have been applied to locate and identify the species at type localities and elsewhere.

• No explicit characterisation of taxon variability at the type localities is presented that demonstrates how the type specimen or selected topotype fits within the species’ broader variability (see Section 2.5).
• No habitat-stratified sampling regime is described or evidenced in the sampling design presented that would allow assessment of potential habitat or depth related lineage structure (see Section 2.5).
• No explicit comparative framework is documented or evidenced in the analyses presented showing that similar species or potentially confusable morphologies were comprehensively sampled and evaluated at the type localities (see Section 2.5).
• Specifically in the case of the Acropora hyacinthus complex, despite published evidence consistent with reticulation, hybridisation and/or introgression involving Acropora cytherea (e.g., refs. [6,40,69]), that species was not sampled or analysed in the study (see Section 2.6).
• Sampling for molecular study is not demonstrated, on the basis of the information presented, to be sufficiently replicated or biologically representative for species-level revisions or distributional modifications (see Section 2.5).
• Recovered topologies were reported to have low gene concordance values (generally <10% in Bridge et al., 2023 [3] and <11% in Rassmussen et al., 2025 [7]) complicating species-level interpretation (see Section 2.10).
• Evidence of incomplete lineage sorting, admixture, hybridisation, and introgression is reported by one or both studies and may affect interpretation of the recovered topologies (see Section 2.10).
• Despite these analytical issues, neither study appears, based on the analyses presented, to have explicitly modelled such alternative evolutionary processes before proceeding to species delineation (Section 2.10).
• The lack of biologically representative sampling, together with these molecular uncertainties, does not, on the basis of the evidence presented, provide a sufficient foundation for conversion of lineages to species-level taxa.
• Morphometric and descriptive assessments were based on limited samples and remain strongly dependent on type specimens that are not fully characterised within the variability of the species and/or which were known by the authors, or documented by others, to be juveniles or otherwise atypical (Section 2.5 and Section 2.7).
• Some morphological characters were evaluated as diagnostic. For some of these characters, habitat-related variability is well established, yet this variability is not explicitly characterised in the relevant datasets presented (Section 2.7)
• According to the methods of the published studies, the morphologically diagnostic characters used to separate lineages were sought after lineages had been recovered, which introduces a potential risk of circular interpretation and confirmation bias in such topology-first workflows, particularly where the recovered topologies were not fully stable (e.g., refs. [80,81]).
• Some synonyms were taken out of synonymy to link to a recovered lineage based on comparison of a type specimen with a small number of specimens without corresponding analysis of variability in the proposed ‘diagnostic’ characters.
• Other synonyms were taken out of synonymy without molecular support on the basis of type comparison, morphology, and/or geography alone, and without demonstration of a distinct, consistently identifiable field entity (see Section 2.5, Section 2.6 and Section 2.7). This includes cases such as Acropora flabelliformis, which was reinstated by Rassmussen et al. (2025) [7] on the basis of a single type specimen, without field study or molecular support.

In summary, as detailed in our Review, the principal concern is not that genetic structure or distinct lineages were recovered—indeed such discoveries are valuable. Our concern is that formal species-level resolution was undertaken from datasets and sampling designs that did not adequately test the extent of within-species biological, morphological, geographic, and genomic variability, or potential processes affecting gene flow. These limitations affect both the interpretation of lineage partitions and the evidentiary basis for subsequent taxonomic and nomenclatural actions.

By contrast, many of the studies we have quoted herein (see in particular, Section 2.5), adopt a more cautious interpretive framework, explicitly recognising unresolved taxonomy, morphological overlap, environmental influence on characters, potential gene flow between lineages, and the possibility of additional diversity beyond the scope of individual datasets or the regional study frame. Our position, as clearly stated in our Review, is not to dismiss these studies, but to emphasise that further integrative work—incorporating comprehensive geographic, habitat and population-level sampling—is required before stable species-level revision of the Acropora tenuis and A. hyacinthus complexes can be achieved.

3. On Mischaracterisations and Errors of Fact by Cowman et al.

Many statements throughout Cowman et al.’s commentary frame our Review in a manner that does not accurately reflect its content or intent, or that constitute errors of fact. Since such framing influences readers’ interpretation of the issues under discussion, we address some of these characterisations here in summary form, focusing on statements made in their Abstract and Final Remarks as examples (Section 3.1, Section 3.2, Section 3.3 and Section 3.4), and offer a few others briefly in Section 3.5. Detailed point-by-point responses to specific statements are provided in our Supplementary Text.

3.1. Alleged Rejection of Molecular Approaches

Cowman et al. characterise our Review as resistant to, or dismissive of, reproducible molecular evidence. This is incorrect. Our Review explicitly recognised the transformative contribution of molecular data to coral phylogeny, particularly at family and deeper phylogenetic levels. We highlighted multiple recent studies that have provided robust integrative insights. We did not argue against genomic approaches; rather, we emphasised that species-level taxonomic revision requires evidentiary standards commensurate with the geographic and biological scope of the taxa concerned.

Disagreement with the interpretation or sufficiency of particular datasets should not be conflated with rejection of molecular methodology. Our critique concerned limits of inference and sampling design, not the legitimacy of genomics, or indeed other methods, as tools.

3.2. “Static” Criticisms and Alleged Lack of Engagement

In their Final Remarks, Cowman et al. suggest that our criticisms have “remained static while the science has transformed,” implying a lack of engagement with recent advances. This assertion overlooks and misrepresents both the historical record and the content of our Review. Molecular approaches were anticipated as central to coral systematics decades ago [4], and our Review engaged extensively with contemporary genetic, reproductive, and population-level studies.

Our position has evolved alongside the science. What has remained consistent, but not ‘static’, is the principle that species-rank decisions require comprehensive contextualisation. Continuity of methodological standards should not be mistaken for resistance to new data. Indeed, Veron (1995: p.32) [4] foresaw the issue and could well have been writing about the present situation:

“Molecular techniques present a powerful array of tests of morphological taxonomy and, in the long-term, may displace morphology as the primary basis for separating species. Hillis (1987) provides a general review of the subject, concluding that disagreements among morphological and molecular systematists over ‘species’ definitions usually represent a disagreement of concept without due reference to biological realities. This is likely to increasingly be the case with corals.”

Although now historical, Hillis’s (1987) [9] review foreshadowed issues that are particularly pertinent to this discussion by highlighting how sampling interacts with character polymorphism: synapomorphies, whether morphological or molecular, inferred from restricted datasets may not remain diagnostic when geographic and ecological sampling is widened, and small sample sizes are only defensible where informative characters are shown to be fixed within species—an assumption that reef-building corals are especially prone to violate.

3.3. Reticulate Evolution as an “Untestable Axiom”

Cowman et al. described reticulate evolution as having “calcified into a putative axiom”, suggesting that it functions as an untestable shield against molecular contradiction. This characterisation is also inaccurate. Reticulate evolution was presented more than three decades ago as an evolutionary framework to explain observed patterns and documented features of coral biology and variability, including hybridisation. In our Review, we explicitly stated that the extent and evolutionary significance of reticulation are likely to vary among taxa and that it must be evaluated empirically. Recognition of reticulation as a plausible evolutionary process does not assume its presence in every lineage, nor does it preclude the identification of predominantly bifurcating, or non-reticulating histories. These caveats notwithstanding, the evidence to date is consistent with its fundamental role in coral evolution and biogeography. Our position is that where gene flow is biologically plausible and has been documented in related taxa, explicit testing for introgression and demographic history strengthens species-level inference. Advocating model comparison is not equivalent to invoking an axiom.

3.4. Alleged Decoupling of Names from Specimens

Cowman et al. assert that our reliance on “biological entities” effectively decouples names from physical specimens, allowing species concepts to drift according to observer opinion. This interpretation misrepresents the nature of taxonomic practice. Biological entities, as discussed in our Review, are structured species hypotheses grounded in repeated field observation, museum collections, comparative morphology, and the published literature. They are not informal constructs insulated from empirical scrutiny. Nor, contrary to Cowman et al.’s assertion, do they exist independently of type material. Rather, type specimens are interpreted within a broad comparative framework that includes geographically and ecologically replicated material accumulated over decades. Broad-scale experience in situ, which includes cues and colony characters unavailable in museum specimens, such as full colony architecture, colour, crowding, antagonistic interactions, commensals, habitat association, co-occurrence with similar species, and population level variability, enhances interpretation and identification of the taxon and its links to museum specimens. It does not ‘decouple’ the taxon from its specimens, historical and modern confusions notwithstanding (e.g., Supplementary Section S7g Acropora microclados). Indeed, it is clear from our discussion on the consequences of over-reliance on type specimens that an understanding of the field entity is crucial to their interpretation (Section 2.11 above).

All molecular sampling presupposes prior identification of specimens. In practice, genetic datasets are anchored to existing taxonomic frameworks and reference material. Sequences are not generated in a taxonomic vacuum; they are attributed to names through specimen identification, which itself depends on comparison with type material and historically characterised populations. The question is therefore not whether names are linked to specimens, but how lineage-level partitions are evaluated relative to historically and biologically characterised species concepts.

3.5. Specific Examples of Mischaracterisation

Below we list several examples among many instances in which our Review has been mischaracterised.

Cowman et al.: “The recent review by Veron et al. (2025) posits that quantitative genomic evidence used to understand coral evolution should be secondary to species hypotheses derived from expert opinion based on field experience.”

We did not posit this (see Supplementary Section S0a).

Cowman et al.: “The authors argue that morphological “biological entities” should take precedence over molecular evidence when conflicts arise.”

We did not argue this (see Supplementary Section S0b).

Cowman et al.: “If the morphology of two ‘species’ overlaps (due to plasticity or convergence), or if the identity of the name-bearing type specimen is ambiguous and lacking in diagnostic morphological characters, then the species is rejected as a “variant” rather than being evaluated as a potentially distinct evolutionary lineage.”

This was not our position; rather we considered that the specific evidence presented in certain cases was insufficient for the inferences (see Supplementary Section S2e).

Cowman et al.: “Contrary to the assertions of Veron et al., most of the studies cited above explicitly incorporate morphological evidence in their analyses and identify synapomorphies that delineate families, genera and species morphologically where they exist.”

We did not assert or imply that studies did not incorporate morphological evidence. Indeed, we explicitly stated (Review Section 4.8b):

“A number of new species descriptions have been published since 2010 based primarily on molecular divergence among samples but also including macro- and micromorphological comparisons of specimens.”

However, in some of these cases, we questioned whether the combined evidence was sufficient for the inferences (see Supplementary Section S2f).

Cowman et al.: “The authors invoke the Code to preserve familiar names while rejecting the fundamental tenet of typification when the type material does not suit their ‘biological entities’.”

We did not do this (see Supplementary Section S5a).

Cowman et al.: “There are obvious practical consequences for maintaining an artificially stable taxonomy as proposed by Veron et al.”

We do not propose an “artificially stable taxonomy”; rather, before destabilising well established taxa, we argue for restraint and alternatives when there are other biological considerations that have not been taken into account (see Supplementary Section S5c).

Cowman et al.: [Veron et al. prioritise] “…convenience over precision and the concomitant risk of obscuring biodiversity and accelerating extinction”.

We do not do this (see Supplementary Section S5f).

Cowman et al.: [we] “discard typification because it is inconvenient” and [we argue that] “because the physical type specimen is ambiguous, the evolutionary lineage it represents must be invalid”.

Both statements are mischaracterisations (see Supplementary Section S6a).

3.6. Errors of Fact

In addition to the various misinterpretations and misrepresentations outlined above, Cowman et al. make other statements that are factually incorrect. We provide three examples here; fuller documentation is provided in Supplementary S1.

(a) 

Peer review of Corals in Space and Time

Cowman et al. state that our taxonomy is based on an evolutionary hypothesis established in Corals in Space and Time [3], “which was not peer reviewed and presents no primary data.”

This is incorrect. As acknowledged in the volume itself, the book underwent formal academic peer review prior to publication by University of New South Wales Press and subsequently by Cornell University Press. It also included specialist input from numerous acknowledged experts and named reviewers. It synthesised decades of primary research, including field and museum-based studies, and was reviewed in major scientific outlets (e.g., Grigg, 1995, Science [94]). While the book was a synthesis rather than a primary research article, to characterise it as unreviewed or unsupported by primary data is factually inaccurate.

(b) 

Misrepresentation of our position on the Unified Species Concept

Cowman et al. state that we “incorrectly asserted” that the Unified Species Concept (USC) permits delineation without requiring species to be morphologically distinguishable or reproductively supported.

This too is incorrect. We were not commenting on de Queiroz’s (2007 [83]) formulation of the USC itself. Rather, we quoted directly from Bridge et al. (2023) [3], who stated that their application of the USC allows delineation of distinct evolutionary lineages “without requiring species to be distinguishable morphologically or to exhibit intrinsic reproductive isolation.” Our critique addressed that specific interpretation and application, not the USC per se. Our statement was not a misunderstanding of the USC. It was a direct quote from Bridge et al. (2023) [3]. The misunderstanding lies elsewhere.

(c) 

Dismissal of Indian Ocean records of Acropora hyacinthus

Cowman et al. describe our response to the dismissal by Rassmussen et al. (2025) [7] of Indian Ocean records of Acropora hyacinthus as a “textbook example of Argumentum ad Populum,” suggesting that historical records reflect circular reliance on Veron’s taxonomy. This claim is demonstrably false. Independent records of Acropora hyacinthus from the Indian Ocean pre-date both Wallace (1999) [76] and Veron (2000) [84] and were made by multiple researchers across different regions. Their identifications were based on detailed skeletal comparison, field observation, and regional taxonomic assessment, not on circular citation of a single synthesis. Documentation is provided in Review Supplementary Table S2.

Furthermore, Cowman et al. apparently misrepresented the work of some of their own authors. Rassmussen et al. (2025) [7] reported only that none of their limited Indian Ocean molecular samples were assigned to the Acropora hyacinthus complex (and from that inferred that A. hyacinthus did not occur there). However, Cowman et al. further reformulated this as the Fijian lineage being “genetically distinct from any tabular Acropora species from the Indian Ocean,” (our bolding) a broader claim not directly tested by the sampling presented, nor a claim made by Rassmussen et al. (2025) [7]. Absence from a restricted dataset does not constitute comprehensive regional exclusion, particularly in a widely distributed and morphologically variable complex. And particularly when numerous workers, including a member of the Cowman et al. team, had previously reported A. hyacinthus at various localities (including Gulf of Aden and Maldives) (Pichon, Benzoni, Chaineau and Dutrieux, 2010) [95]. Given the wide geographic separation, the Fijian lineage may well be distinct from Indian Ocean populations, but such lineage differentiation does not necessarily denote distinct species (see Supplementary Section S7h for more detail).

3.7. Scope of Disagreement

Finally, the Abstract and Final Remarks of Cowman et al. give the impression of a fundamental opposition between our view and contemporary coral systematics. This is both incorrect and overstated. The principal areas of disagreement concern the following:

• the evidentiary thresholds appropriate for species-rank recognition;
• the proportionality between sampling scope and taxonomic revision;
• the interpretation of discordance among molecular datasets; and
• the application of nomenclatural rules in a manner consistent with stability.

These are methodological and interpretative questions within an integrative framework, not a rejection of that framework.

4. Concluding Remarks

Differences in perspective regarding what constitutes a species and the evidence required to substantiate these taxa lie at the core of the present discussion. As we noted in our Review, the “species problem” is not unique to corals; it is a longstanding issue across biology. In corals, the challenge is amplified by extensive phenotypic plasticity, broad geographic ranges, and documented reticulate evolutionary processes. These biological realities demand methodological proportionality in species-level taxonomic revision.

We agree that evidence must be quantitative and reproducible. Our disagreement concerns evidentiary thresholds, sampling scope, and interpretation. Robust molecular lineage discovery provides valuable insight into evolutionary structure. However, lineage discovery and species recognition are not equivalent processes. Genetic clusters represent evolutionary hypotheses whose taxonomic status must be evaluated within the broader biological context of the species concerned. Formal species recognition requires population-level geographic and ecological representation, contextual understanding of type specimens and morphological variability, and explicit evaluation of gene flow where biologically plausible. More broadly, recent work has shown that analytical decisions alone can produce substantial variation in scientific conclusions even when researchers analyse the same dataset ([91]—see Section 2.10 above). Such outcomes may also reflect incomplete sampling, methodological sensitivity, or the complex reticulate genetic histories characteristic of corals (as illustrated in our Review Stylophora case study, Section 4.7).

We emphasise five principles that underpin stable and biologically meaningful species-level taxonomy in corals:

4.1. Biological Context and Replication

Molecular study design and species-level inference must be grounded in the documented biological complexity of the taxon, including phenotypic, molecular, reproductive, and ecological or functional variability across habitats and regions, as well as known interactions with related taxa. Close correspondence in multiple traits—including those observed in type specimens—should not be attributed to convergence by default where those traits fall within the empirically established variation of the species across its range. Replication must encompass biologically relevant variation as well as technical sufficiency. Sampling design and topotypes should be informed by a comprehensive understanding of the taxon at the type locality, and beyond.

4.2. Type Specimens and Nomenclatural Anchoring

Type specimens serve as nomenclatural anchors, not exhaustive representations of natural variability. Their interpretation must occur within the context of accumulated comparative material. Where historical types are deficient or ambiguous, ICZN mechanisms exist to preserve stability through lectotype or neotype designation, or temporary retention of a name pending resolution, rather than precipitate nomenclatural disruption.

4.3. Lineages and Species

Detection of molecular lineages is an important first step. Translation of lineage partitions into formal species status, however, requires integration of genetic structure with morphological, ecological, and biogeographic data at appropriate sampling scales.

4.4. Reticulation and Model Testing

Gene-tree discordance, incomplete lineage sorting and introgression are documented features of coral evolution. Analytical frameworks should therefore explicitly evaluate alternative evolutionary models rather than presuppose strictly bifurcating histories. Testing for gene flow strengthens, rather than weakens, species hypotheses.

4.5. Stability and Conservation

Taxonomic decisions carry implications beyond systematics, influencing conservation policy, ecological interpretation and biodiversity assessment. We do not accept Cowman et al.’s assertion that our approach risks ‘silent extinction’ or renders lineages invisible to conservation. Conservation priority does not depend critically on taxonomic elevation. Genetic lineages can, and regularly do, inform conservation planning in many diverse taxonomic groups, irrespective of formal species rank. Management frameworks frequently operate at population, lineage, or regional scales where appropriate. Our position is simply that formal species designation should be based on comprehensive and geographically informed evidence, so that conservation actions are grounded in robust and durable taxonomy.

Furthermore, rather than ‘accelerating extinction’, silent or otherwise, we are highly cognisant of the risks to corals and the reefs they build (as explained in detail in refs. [96,97,98]), having consistently and strongly advocated for conservation. Indeed, early work by Veron and colleagues at the Solitary Islands and Lord Howe Island, east Australia [99,100,101] helped in the establishment of protected areas, including at the type locality of A. harriottae, other management measures (e.g., Crown-of-thorns starfish, Acanthaster, control), and provided the basis for future studies. Our work elsewhere also has led to significant conservation outcomes, via establishment of Marine Protected Areas, informed by our many studies across the Coral Triangle, western Indian Ocean and Arabia (e.g., refs. [102,103]). We advocate for conservation efforts towards species and lineages, particularly those with apparent physiological advantages in our rapidly changing oceans. This is not inconsistent with restraint over designation of new species where studies are not yet sufficiently comprehensive. Stability and universality, as articulated in the Preamble of the ICZN, remain central objectives. Prudence in species-level revision safeguards long-term reliability of biological knowledge.

5. A Final Word

Our aim in our Review and in CoralsOfTheWorld.org (2016, 2026 in prep.) [70,71] has been to provide a structured, evidence-based framework for recognising and documenting coral diversity—integrating morphology, field observation, and molecular insights. We have been aided in this endeavour by numerous dedicated researchers and underwater photographers from many nations, for which we are grateful. We expect that many of the authors of Cowman et al. made good use of this resource—indeed, some have contributed to it. As predicted by Veron (1995) [4] more than three decades ago, we anticipate that continued advances in genomic methodologies, combined with comprehensive biologically relevant sampling, will further refine species hypotheses. Restraint in the face of incomplete evidence does not impede scientific progress; it ensures that revision is durable.

Point by point responses to specific claims by Cowman et al. are provided our Supplementary Text and its Appendices.