What is a Species, Really? How Genomics is Solving Biology’s Oldest Debate

The Epistemological Crisis of the Species Rank

The species is the fundamental currency of biology. It is the unit of conservation, the node of phylogenetic analysis, and the primary subject of evolutionary theory. Yet, despite centuries of study, the definition of what constitutes a species remains one of the most contentious debates in the life sciences. From the morphological distinctiveness championed by Linnaeus to the reproductive isolation emphasized by the Biological Species Concept, taxonomists have long sought a universal yardstick for biodiversity. In the twenty-first century, this quest has moved from the museum drawer to the high-performance computing cluster. The advent of genomic data has precipitated a paradigm shift, transforming species delimitation from a qualitative art into a quantitative, process-oriented science.

A landmark review published in 2025 in the Annual Review of Ecology, Evolution, and Systematics by Sonal Singhal, Adam D. Leaché, Matthew K. Fujita, Carlos Daniel Cadena, and Felipe Zapata, titled "A Genomic Perspective on Species Delimitation," captures the state of this transformation.¹ This report provides an exhaustive analysis of the themes, methodologies, and biological insights presented in that review and the broader literature it references. It explores how the genomic perspective has redefined the boundaries of life, utilizing sophisticated models like the multispecies coalescent (MSC) to disentangle the complex history of divergence, gene flow, and lineage sorting that characterizes the speciation process.

The transition to genomic species delimitation is not merely a technological upgrade; it is a conceptual revolution. It acknowledges that speciation is not an instantaneous event but an extended process—a continuum where populations gradually accumulate genetic differences, ecological distinctiveness, and reproductive barriers.² This continuum creates a "grey zone" of speciation, where lineages may be genetically distinct but not yet reproductively isolated, or morphologically cryptic yet separated by millions of years of evolution. The genomic perspective seeks to navigate this grey zone by providing objective, statistical frameworks for testing species hypotheses, fundamentally altering our understanding of groups ranging from the cryptic night lizards of the North American deserts to the gelatinous predators of the deep ocean and the recombining microbes that sustain the biosphere.

Theoretical Foundations: The Multispecies Coalescent

To understand the modern approach to species delimitation, one must first engage with the theoretical engine that drives it: the Multispecies Coalescent (MSC) model. In the era of single-gene phylogenetics (often based on mitochondrial DNA), researchers frequently assumed that the history of a gene was identical to the history of the species. We now understand that this assumption is frequently violated due to the stochastic nature of inheritance.

The Mechanics of Lineage Sorting

The history of a genome is a collection of thousands of independent gene trees, each telling a slightly different story about the relationships among species. This discordance arises primarily from a phenomenon known as Incomplete Lineage Sorting (ILS) or deep coalescence.³ When an ancestral population splits into two descendant lineages, the genetic variation present in the ancestor does not instantly segregate into the daughters. Polymorphisms—different alleles of a gene—can be randomly sorted such that a gene copy in Species A is more closely related to a gene copy in Species C than to one in its sister Species B.

The MSC models this process explicitly. It treats the species tree not as a single branching line, but as a "container" or a set of demographic boundaries within which gene lineages evolve and coalesce (merge backwards in time).⁵ The model relies on two primary parameters: theta, which represents the effective population size (scaled by the mutation rate), and tau, which represents the species divergence time (also scaled by mutation rate).⁶

By integrating over the histories of thousands of loci, the MSC allows researchers to estimate the probability that a given set of gene trees could arise within a hypothesized species tree. This probabilistic framework transforms species delimitation from a pattern-recognition exercise into a model-testing procedure. A researcher can ask: "Is the genomic data more likely under a model where these three populations are distinct species, or under a model where they are a single panmictic species?".⁷

The Censored Coalescent

The MSC is sometimes referred to as the "Censored Coalescent" because the coalescent process (the merging of lineages) is constrained or "censored" by the speciation events.⁵ Lineages cannot coalesce between species after the divergence time tau (assuming no gene flow), but they must coalesce eventually in the ancestral populations. This mathematical structure allows the model to distinguish between the stochastic noise of population genetics and the signal of species divergence.

However, the basic MSC assumes strict isolation following divergence—a simplification that nature often violates. Extensions to the model, such as the Multispecies Coalescent with Introgression (MSci) or the Multispecies Network Coalescent (MSNC), have been developed to accommodate gene flow, hybridization, and reticulate evolution, which are increasingly recognized as common features of the speciation continuum.⁵

The Methodological Workflow: Discovery and Validation

The operational implementation of genomic species delimitation typically follows a two-step workflow: Discovery (or hypothesis generation) and Validation (or hypothesis testing). This distinction, emphasized by Singhal et al. (2025) and Carstens et al. (2013), provides a structured approach to handling the complexity of genomic data.¹

Step 1: Discovery – Generating Hypotheses

The discovery phase is exploratory. Researchers begin with a set of sampled individuals and genomic data (typically Single Nucleotide Polymorphisms, or SNPs) without necessarily assuming a priori how many species are present. The goal is to identify "putative species-level units"—clusters of genetically similar individuals that may represent distinct evolutionary lineages.

Clustering Algorithms

Standard population genetic tools are often repurposed for this stage. Algorithms like Structure or Admixture assign individuals to K clusters based on genotype frequencies, minimizing deviations from Hardy-Weinberg equilibrium. Principal Component Analysis (PCA) and Discriminant Analysis of Principal Components (DAPC) provide visual representations of genetic space, revealing discontinuities that suggest barriers to gene flow.¹

Phylogenetic Networks and Distances

Distance-based methods, such as the analysis of raw genetic distances or the construction of unrooted phylogenetic networks (e.g., using SplitsTree), can also reveal grouping patterns. These methods are computationally efficient and can handle the large datasets generated by methods like ddRADseq or whole-genome resequencing. However, they are descriptive rather than inferential; they show that clusters exist, but they do not prove that these clusters are species.⁷

Step 2: Validation – Rigorous Testing

Once putative units are identified, they are subjected to validation using model-based approaches. This is where the heavy lifting of the MSC occurs.

Bayesian Phylogenetics and Phylogeography (BPP)

Perhaps the most prominent tool in this domain is BPP (Bayesian Phylogenetics and Phylogeography). BPP implements a reversible-jump Markov Chain Monte Carlo (rjMCMC) algorithm to explore the space of possible species delimitation models. It can collapse nodes in a guide tree, effectively merging populations into a single species, or split them apart, calculating the posterior probability of each configuration.⁸ BPP is powerful because it simultaneously estimates the species tree topology and the delimitation model, accounting for the uncertainty in both.¹⁰

SNAPP and Bayes Factors

Another widely used tool is SNAPP (SNP and AFLP Package for Phylogenetic analysis), which calculates the likelihood of the species tree directly from biallelic markers (SNPs) without first estimating individual gene trees. This bypasses the computational burden and potential error of gene tree estimation for short sequences.¹⁰ When combined with Bayes Factor Delimitation (BFD*), SNAPP allows researchers to compare the marginal likelihoods of competing delimitation models (e.g., "Species A and B are separate" vs. "Species A and B are one species") and select the one with the highest support.¹²

The Validation Crisis: Structure vs. Species

Despite the sophistication of these methods, a significant controversy persists. Critics, such as Sukumaran and Knowles (2017), have argued that the MSC detects genetic structure, not necessarily species.¹³ If two populations are isolated by distance and have ceased exchanging genes, the MSC will correctly identify them as distinct lineages. However, if they have not yet evolved ecological distinctiveness or reproductive incompatibility, taxonomists might consider them merely structured populations of the same species. This tendency of MSC-based methods to "over-split" biodiversity—elevating every isolated population to species rank—is a central challenge discussed in the Singhal et al. review.¹⁴

To mitigate this, the review advocates for integrative approaches. Metrics like the Genealogical Divergence Index (gdi) have been proposed to quantify the magnitude of divergence, not just its statistical significance. The gdi scales the divergence time by the population size, providing a continuous index (0 to 1) of speciation progress. A high gdi (e.g., > 0.7) suggests strong reproductive isolation, while a low gdi suggests that lineages, while structured, are likely conspecific.¹⁶

Table 1: Comparative Overview of Genomic Delimitation Methods

Method	Algorithm / Basis	Input Data	Strengths	Limitations
BPP	Bayesian MSC (rjMCMC)	Multilocus Sequence / SNPs	Estimates theta, tau, and delimitation simultaneously. Handles gene tree uncertainty.	Computationally intensive. Can over-split population structure as species. Requires a guide tree (usually).
SNAPP	Exact MSC Likelihood	Biallelic SNPs	Bypasses gene tree estimation. Good for short markers (RADseq).	Restricted to biallelic data. Computationally expensive with many taxa.
BFD*	Bayes Factor / Path Sampling	SNPs (via SNAPP)	Provides rigorous statistical model selection/ranking.	computationally demanding. Requires pre-defined models (cannot "discover" species).
gdi	Genealogical Divergence Index	MSC Parameters (tau, theta)	continuous measure of divergence; reduces over-splitting.	A heuristic index, not a strict statistical test. interpreting thresholds can be subjective.
SODA	Quartet Frequencies	Gene Trees / Quartets	Extremely fast; scalable to large datasets.	Heuristic; less rigorous than full Bayesian MSC.
Structure	Bayesian Clustering	Genotypes (SNPs)	Excellent for visualizing admixture and identifying hybrids.	Does not model phylogeny or the speciation process. A "discovery" tool, not validation.

Case Study I: Cryptic Diversity in Vertebrates

The theoretical frameworks of genomic delimitation find their most compelling applications in the natural world. Singhal et al. (2025) highlight vertebrates as a key testing ground, revealing that morphological evolution and genomic divergence are often decoupled.

The Night Lizards of Xantusia

The genus Xantusia, a group of small night lizards inhabiting the rock crevices and yuccas of the North American deserts, serves as a paradigmatic example of "cryptic" diversity. For decades, herpetologists classified these lizards into a few widespread species based on scale counts, body size, and coloration. Phenotypically, a Xantusia from Baja California might look nearly identical to one from the Mojave Desert.

However, genomic analyses have shattered this view. As detailed in related research by Singhal, Leaché, and colleagues, Xantusia represents an ancient radiation where morphological evolution has essentially stalled.¹⁸ The application of MSC-based delimitation methods revealed that these lizards contain multiple, deeply divergent evolutionary lineages that have been separated for millions of years.²⁰

• Deep Divergence: Sequence divergence in mitochondrial and nuclear markers between these cryptic species often exceeds 20%—levels that typically separate distinct genera in mammals.²⁰

• The Allopatry Predictor: The study suggests that speciation in Xantusia is a predictable consequence of extended allopatry (geographic isolation). These lizards have extremely low dispersal capabilities and highly fragmented habitats. Over millions of years, isolated populations drifted apart genomically, but because they occupied similar ecological niches (rock crevices), there was no selective pressure to diverge morphologically.²¹

• Validation Success: This case illustrates the triumph of the genomic perspective. Traditional taxonomy failed because the signal was not in the phenotype. The MSC successfully validated these lineages by detecting the clear genomic signature of long-term isolation, correcting a significant underestimation of biodiversity.²²

The Paradox of Avian Speciation

In contrast to the cryptic lizards, birds often exhibit dramatic phenotypic variation in plumage and song, traits that are critical for sexual selection and reproductive isolation. Yet, the genomic perspective has revealed a different kind of paradox in avian speciation: the "permeable species boundary."

Research by Carlos Daniel Cadena and others, reviewed in Singhal et al. (2025), challenges the strict application of the Biological Species Concept (BSC) in the genomic era.²³

• Gene Flow: Genomic data frequently reveals that recognized bird species—taxa that are morphologically distinct and do not interbreed freely in the wild—often show signatures of historical or even ongoing gene flow. The MSC, which assumes strict isolation, can struggle with these cases.

• Plumage vs. Genome: In radiations like the Setophaga warblers or the diverse Scytalopus tapaculos, phenotypic traits can evolve rapidly under sexual selection while the rest of the genome remains relatively undifferentiated. Conversely, some bird lineages show deep genomic splits with little phenotypic change.²⁴

• The "Speciation Genes" Hypothesis: The review highlights that genomic divergence in birds is often heterogeneous. Differentiation may be restricted to small "islands of divergence" containing genes responsible for reproductive barriers (e.g., plumage color, song), while the rest of the genome is homogenized by gene flow.²⁵

• Implication for Delimitation: This suggests that a strictly genomic definition of species (e.g., based on genome-wide Fst) might lump together good biological species. Singhal et al. argue that for birds, genomic validation must be integrated with phenotypic data. Delimitation should not rely solely on the cessation of gene flow but on the persistence of distinct evolutionary trajectories despite it.²³

Case Study II: Complexity in the Open Ocean

Moving from the desert to the deep sea, the review examines the Siphonophora (Cnidaria), a group of colonial hydrozoans that includes the famous Portuguese man o' war (Physalia physalis) and the bioluminescent deep-sea species of the genus Nanomia. These organisms challenge the very concept of an "individual," and by extension, the concept of a species.

The Colonial Problem

Siphonophores are colonies composed of specialized bodies called zooids (e.g., feeding gastrozooids, swimming nectophores). Because these colonies are fragile and often disintegrate during net sampling, morphological taxonomy has been historically difficult. Genomic barcoding and transcriptomics have become essential for understanding their diversity.

Work by Felipe Zapata and colleagues has utilized genomic data to resolve the deep phylogeny of the group, particularly within the clade Codonophora.²⁸ This phylogenetic framework has been crucial for species delimitation, revealing extensive cryptic diversity that morphology missed.

Genomic Gigantism and Reduction

A striking finding in siphonophore genomics is the plasticity of the genome itself. In the genus Nanomia, researchers have discovered significant variation in genome size.

• Genome Dynamics: Nanomia septata and Nanomia cara possess relatively large genomes (1.5–1.7 GB), while the related Nanomia bijuga and an undescribed species show a secondary reduction to approximately 0.7 GB.²⁹

• Structural Rearrangement: The genome of Nanomia septata is highly rearranged compared to other hydrozoans, suggesting a turbulent evolutionary history. This variation in genome architecture (size, synteny) provides an additional layer of data for delimitation but also complicates alignment-based methods.²⁹

The Physalia Revision

For centuries, the Portuguese man o' war was treated as a single, globally distributed species (Physalia physalis). However, the combination of genomic data and citizen-science morphological scoring has led to the identification of at least four distinct lineages, likely representing separate species. Some of these correspond to taxa described in the 18th and 19th centuries that were later synonymized.³⁰ This case exemplifies how genomics can "resurrect" valid species that were lump-summed by conservative morphological taxonomy.

Case Study III: The Microbial Frontier

Bacteria represent the most significant challenge to the species concept. The models developed for sexually reproducing eukaryotes (like the MSC) often break down when applied to prokaryotes due to their asexual reproduction and the pervasive influence of Horizontal Gene Transfer (HGT).

The Recombination Challenge

In eukaryotes, recombination is inextricably linked to reproduction (meiosis). In bacteria, recombination (HGT) is decoupled from reproduction and can occur between distantly related lineages. This violates the assumption of a bifurcating species tree, creating a phylogenetic network rather than a clean hierarchy.³¹

• Homologous vs. Non-Homologous: Bacteria undergo homologous recombination (swapping allele versions) which can blur species boundaries, and non-homologous recombination (acquiring new accessory genes) which can create ecological differentiation without genome-wide divergence.³²

The Stable Ecotype Model

To address this, microbiologists often employ the Stable Ecotype Model as a theoretical basis for delimitation. This model posits that bacterial species are ecological units (ecotypes). Periodic selection sweeps (where a beneficial mutation sweeps through the population) purge genetic diversity within an ecotype, maintaining its cohesion.

• Genomic Species vs. Ecotypes: A standard "genomic species" in microbiology is often defined by an Average Nucleotide Identity (ANI) threshold of ~95-96%. However, Singhal et al. note that this genomic threshold does not always map to ecological units.

• Example: In Synechococcus bacteria from Yellowstone hot springs, distinct ecotypes (adapted to different temperatures or depths) can exist within a single genomic cluster (divergence < 0.5%). These ecotypes are maintained by selection on a few specific loci, despite the homogenization of the rest of the genome by recombination.³³ Conversely, in Ensifer, genomic clusters align well with ecological differences.³⁴

• The Delimitation Dilemma: The review highlights that for bacteria, delimitation is a search for "cohesive units." The challenge is determining whether that cohesion is driven by restricted gene flow (a "species" in the eukaryotic sense) or by ecological selection (an "ecotype"), and whether these units are stable over evolutionary time.³⁵

The Future of Delimitation: Machine Learning and Integration

As the volume of genomic data grows, the computational limits of likelihood-based methods like BPP are being tested. The Singhal et al. (2025) review points toward Artificial Intelligence (AI) and Machine Learning (ML) as the next frontier in species delimitation.

Supervised Machine Learning

Supervised methods, such as Convolutional Neural Networks (CNNs), are being trained to recognize the "visual" patterns of speciation in alignment matrices. By converting genetic data into image-like tensors, CNNs can classify evolutionary scenarios (e.g., "Species A split from B with gene flow") with high accuracy, often outperforming ABC (Approximate Bayesian Computation) in complex scenarios.³⁶ These methods are particularly promising for distinguishing between population structure and speciation in the presence of gene flow.

Unsupervised Machine Learning

Perhaps even more revolutionary are unsupervised methods, such as Variational Autoencoders (VAEs) and Self-Organizing Maps (SOMs). These algorithms do not require pre-labeled training data (which can introduce bias). Instead, they learn the inherent structure of the data, projecting high-dimensional genomic information into lower-dimensional space.

• Objectivity: Unsupervised ML offers a way to bypass the circularity of defining species based on prior taxonomy. Studies on arachnids and salamanders have shown that these methods can successfully cluster samples according to species-level divergences while ignoring intraspecific population noise.³⁷

• Integration: SuperSOMs allow for the simultaneous analysis of distinct data layers—genetics, geography, climate, and phenotype—offering a truly integrative path to delimitation that aligns with the Unified Species Concept.³⁷

Challenges and Synthesis

Despite the immense power of the genomic perspective, Singhal et al. (2025) conclude with a sobering analysis of the remaining challenges. The field faces two primary hurdles: Genome Heterogeneity and Divergence Mismatch.

Genome Heterogeneity

The genome is not a uniform record of history. It is a mosaic shaped by conflicting forces. Introgression, selection, and recombination mean that different parts of the genome have different genealogies. Relying on a "genome-wide average" can be misleading. High gene flow at neutral loci can mask speciation (as in birds), while deep divergence at neutral loci can mask the lack of ecological differentiation (as in Xantusia). The solution lies in methods that can parse this heterogeneity, identifying the specific genomic regions that underpin the speciation process.¹

Divergence Mismatch

The rate of genomic evolution does not always match the rate of speciation. This "mismatch" is the core tension of modern systematics.

• Rapid Speciation: In adaptive radiations, species can form rapidly through ecological selection without accumulating significant genomic divergence (the "grey zone" problem).

• Cryptic Stasis: In ancient non-adaptive lineages, genomic divergence can accumulate to high levels without the evolution of reproductive barriers or morphological change.

Conclusion: Toward a Process-Based Taxonomy

The "Genomic Perspective" on species delimitation is ultimately a move toward a process-based taxonomy. It encourages biologists to define species not just by what they look like (morphology) or how much their DNA differs (distance), but by how they evolve (lineage independence).

The tools of 2025—BPP, SNAPP, gdi, and emerging AI models—provide the statistical rigor to test these evolutionary hypotheses. They allow us to see the "invisible" biodiversity of the world, from the cryptic lineages of desert lizards to the reticulate networks of ocean microbes. However, these tools must be wielded with biological intuition. A genomic cluster is a hypothesis, not a conclusion. True species delimitation requires integrating the genomic signal with the ecological and phenotypic reality of the organism. As Singhal and colleagues demonstrate, the genome provides the map, but the history of life is the territory, and exploring it requires considering every layer of biological complexity.

The species boundary, once a static line drawn by taxonomists, is now understood as a dynamic, semi-permeable membrane. The genomic perspective has not dissolved the species concept; it has refined it, revealing a natural world that is more complex, more fluid, and more diverse than Linnaeus could have ever imagined.

Table 2: Biological Case Studies in Genomic Delimitation (Singhal et al. 2025 & Related)

Taxon Group	Key Organism(s)	Primary Evolutionary Characteristic	Key Insight from Genomic Analysis
Reptiles	Xantusia (Night Lizards)	Cryptic Diversity: Morphology is static despite millions of years of separation.	Speciation is driven by allopatry. Genomic splits are deep (>20% divergence) despite morphological similarity. The MSC validates "invisible" species that traditional taxonomy misses.
Birds	Setophaga, Scytalopus	Gene Flow & Phenotype: Phenotype evolves faster/slower than genome; gene flow is common.	"Good" biological species often exchange genes. Genomic divergence does not always predict reproductive isolation. Delimitation must integrate phenotypic barriers (plumage, song).
Siphonophores	Physalia (Man o' War), Nanomia	Coloniality & Deep Time: Complex body plans; difficult morphological ID.	Cryptic diversity is high in the open ocean. Genome size is highly plastic (Nanomia). Genomic data resolves deep nodes in the phylogeny (Codonophora).
Bacteria	Synechococcus, Ensifer	Recombination (HGT): Asexual reproduction + lateral gene transfer violates tree models.	The "Stable Ecotype Model" is preferred. Genomic species (ANI clusters) may contain multiple distinct ecotypes maintained by selection on specific loci.

Detailed Analysis of Key Insights

The Structure vs. Species Dilemma

A recurring insight in the modern literature is the behavior of the Multispecies Coalescent (MSC) when faced with population structure. As highlighted in the snippet ¹³, the MSC is mathematically designed to detect interruptions in gene flow. However, in nature, gene flow can be interrupted by geography (isolation by distance) without resulting in speciation (the evolution of reproductive incompatibility).

• The Consequence: When BPP or similar tools are applied to populations with high structure (e.g., organisms with low dispersal like Xantusia), they tend to delimit every population as a species.

• The Fix: Singhal et al. suggest using gene flow as a proxy for validation.¹ If two "delimited" units show high historical gene flow, they may just be populations. Alternatively, metrics like the Genealogical Divergence Index (gdi) ¹⁶ attempt to normalize the divergence by the population size, providing a continuous measure of "speciation progress" rather than a binary yes/no.

The Role of Natural History Collections

The ability to perform these analyses relies heavily on biodiversity resources. The snippet regarding Xantusia mentions the indispensable role of natural history museums.⁴⁰ Genomic delimitation often requires broad geographic sampling to distinguish isolation by distance from sharp species boundaries. Museums provide the historical and spatial breadth that modern sequencing projects require, serving as the physical library for the genomic age.

The "Grey Zone" of Speciation

Speciation is a continuum. At the beginning (populations), lineages exchange genes freely. At the end (species), they do not. In between lies the "grey zone"—a period of time where lineages are diverging but not yet fully isolated.⁴¹

• Genomic Signature: In the grey zone, different parts of the genome tell different stories. Some genes (under selection) will show sharp divergence (Fst = 1), while neutral markers may still flow freely (Fst ~ 0).²⁵

• Delimitation Implication: Delimitation methods that average this signal (like standard MSC) may fail in the grey zone. They might lump distinct ecological species because of neutral gene flow, or split connected populations because of drift. The "Genomic Perspective" advocated by Singhal et al. is to embrace this complexity, using tools that can model introgression (like the MSci model) to map where a lineage sits on this continuum.⁶

Future Outlook: The Integrative 2.0

The paper concludes with a call for solutions to the challenges of genome heterogeneity and divergence mismatch.¹ The path forward involves Integrative Taxonomy 2.0:

1. Process-Oriented: Moving from "pattern-based" (is there a gap?) to "process-based" (what caused the gap?).

2. Pluralistic: Accepting that "species" might mean something slightly different in bacteria (ecotype) than in birds (reproductive community).

3. Technologically Enhanced: Utilizing AI to detect complex non-linear patterns of divergence that linear statistical models miss.

In 2025, we no longer ask "Do these genomes look different?" We ask, "Do these genomes behave like independent evolutionary lineages?" The answer, as Singhal and colleagues show, requires a rigorous, genomic perspective.

Works cited

1. A Genomic Perspective on Species Delimitation - Annual Reviews, accessed January 11, 2026, https://www.annualreviews.org/content/journals/10.1146/annurev-ecolsys-102723-055311

2. Species boundaries to the limit: validating species delimitation methods is critical to avoid taxonomic inflation in the genomic era | bioRxiv, accessed January 11, 2026, https://www.biorxiv.org/content/10.1101/2022.12.28.522089v1.full-text

3. The multi-species coalescent model and species tree inference - University of Washington, accessed January 11, 2026, https://faculty.washington.edu/leache/wordpress/wp-content/uploads/2020/03/2020RannalaSpeciestree.pdf

4. Multispecies coalescent and its applications to infer species phylogenies and cross-species gene flow - Oxford Academic, accessed January 11, 2026, https://academic.oup.com/nsr/article/8/12/nwab127/6321855

5. Multispecies coalescent process - Wikipedia, accessed January 11, 2026, https://en.wikipedia.org/wiki/Multispecies_coalescent_process

6. Multispecies coalescent and its applications to infer species phylogenies and cross-species gene flow - PMC - NIH, accessed January 11, 2026, https://pmc.ncbi.nlm.nih.gov/articles/PMC8692950/

7. A Genomic Perspective on Species Delimitation - Annual Reviews, accessed January 11, 2026, https://www.annualreviews.org/doi/pdf/10.1146/annurev-ecolsys-102723-055311

8. bio.tools · Bioinformatics Tools and Services Discovery Portal, accessed January 11, 2026, https://bio.tools/bpp

9. bpp/bpp: Bayesian analysis of genomic sequence data under the multispecies coalescent model - GitHub, accessed January 11, 2026, https://github.com/bpp/bpp

10. Species Delimitation using Genome-Wide SNP Data | Systematic Biology | Oxford Academic, accessed January 11, 2026, https://academic.oup.com/sysbio/article/63/4/534/2848292

11. snapp - BEAST 2, accessed January 11, 2026, https://www.beast2.org/snapp/

12. Species Trees and Species Delimitation with SNAPP: A Tutorial and Worked Example Contents - Evolution and Genomics, accessed January 11, 2026, http://evomics.org/wp-content/uploads/2018/01/BFD-tutorial.pdf

13. Multispecies coalescent delimits structure, not species - PNAS, accessed January 11, 2026, https://www.pnas.org/doi/10.1073/pnas.1607921114

14. Species Delimitation Using Genomic Data: Options and Limitations - PubMed, accessed January 11, 2026, https://pubmed.ncbi.nlm.nih.gov/40026292/

15. Hierarchical Heuristic Species Delimitation Under the Multispecies Coalescent Model with Migration - PubMed, accessed January 11, 2026, https://pubmed.ncbi.nlm.nih.gov/39180155/

16. Species Delimitation with Gene Flow | Systematic Biology - Oxford Academic, accessed January 11, 2026, https://academic.oup.com/sysbio/article/66/5/799/2726792

17. Species boundaries in the messy middle—A genome‐scale validation of species delimitation in a recently diverged lineage of coastal fog desert lichen fungi - PubMed Central, accessed January 11, 2026, https://pmc.ncbi.nlm.nih.gov/articles/PMC8717302/

18. Genomic divergence, demographic histories, and male territorial response reveal asymmetric reproductive barriers in allopatric eastern versus western Nashville warbler subspecies (Leiothlypis ruficapilla) | Evolution | Oxford Academic, accessed January 11, 2026, https://academic.oup.com/evolut/advance-article/doi/10.1093/evolut/qpaf221/8297141

19. Testing Species Boundaries in an Ancient Species Complex with Deep Phylogeographic History: Genus Xantusia (Squamata: Xantusiidae) | The American Naturalist: Vol 164, No 3, accessed January 11, 2026, https://www.journals.uchicago.edu/doi/abs/10.1086/381404

20. (PDF) Testing Species Boundaries in an Ancient Species Complex with Deep Phylogeographic History: Genus Xantusia (Squamata: Xantusiidae) - ResearchGate, accessed January 11, 2026, https://www.researchgate.net/publication/30126698_Testing_Species_Boundaries_in_an_Ancient_Species_Complex_with_Deep_Phylogeographic_History_Genus_Xantusia_Squamata_Xantusiidae

21. Comparative approaches to understanding speciation: a case study in Xantusia lizards | Request PDF - ResearchGate, accessed January 11, 2026, https://www.researchgate.net/publication/385077589_Comparative_approaches_to_understanding_speciation_a_case_study_in_Xantusia_lizards

22. Comparative approaches to understanding speciation: a case study in Xantusia lizards - Oxford Academic, accessed January 11, 2026, https://academic.oup.com/evolinnean/article-pdf/3/1/kzae024/59907525/kzae024.pdf

23. genomic revolution and species delimitation in birds (and other organisms): Why phenotypes should not be overlooked - Oxford Academic, accessed January 11, 2026, https://academic.oup.com/auk/article-abstract/138/2/ukaa069/6154985

24. Systematics, biogeography, and diversification of Scytalopus tapaculos (Rhinocryptidae), an enigmatic radiation of Neotropical montane birds - LSU Scholarly Repository, accessed January 11, 2026, https://repository.lsu.edu/biosci_pubs/494/

25. Divergent selection and drift shape the genomes of two avian sister species spanning a saline–freshwater ecotone - PubMed Central, accessed January 11, 2026, https://pmc.ncbi.nlm.nih.gov/articles/PMC6912898/

26. Genomic landscapes of divergence among island bird populations: Evidence of parallel adaptation but at different loci? - uu .diva, accessed January 11, 2026, https://uu.diva-portal.org/smash/get/diva2:1918722/FULLTEXT01.pdf

27. The genomic revolution and species delimitation in birds (and other organisms): Why phenotypes should not be overlooked - BioOne Complete, accessed January 11, 2026, https://bioone.org/journalArticle/Download?urlId=10.1093%2Fornithology%2Fukaa069

28. Improved phylogenetic resolution within Siphonophora (Cnidaria) with implications for trait evolution | bioRxiv, accessed January 11, 2026, https://www.biorxiv.org/content/10.1101/251116v1

29. A Chromosome-Scale Genome of Nanomia septata Reveals Extensive Rearrangement But No Clear Driver of the Unique Colony-Level Organization of Siphonophores | bioRxiv, accessed January 11, 2026, https://www.biorxiv.org/content/10.1101/2025.05.01.651713v1.full-text

30. Population genomics of a sailing siphonophore reveals genetic structure in the open ocean | Request PDF - ResearchGate, accessed January 11, 2026, https://www.researchgate.net/publication/392849093_Population_genomics_of_a_sailing_siphonophore_reveals_genetic_structure_in_the_open_ocean

31. The bacterial species dilemma and the genomic–phylogenetic species concept - PMC - NIH, accessed January 11, 2026, https://pmc.ncbi.nlm.nih.gov/articles/PMC1857736/

32. Exclusivity offers a sound yet practical species criterion for bacteria despite abundant gene flow - PubMed Central, accessed January 11, 2026, https://pmc.ncbi.nlm.nih.gov/articles/PMC6171291/

33. Transmission in the Origins of Bacterial Diversity, From Ecotypes to Phyla - PMC - NIH, accessed January 11, 2026, https://pmc.ncbi.nlm.nih.gov/articles/PMC11687548/

34. Taxonomy of Rhizobiaceae revisited: proposal of a new framework for genus delimitation, accessed January 11, 2026, https://pmc.ncbi.nlm.nih.gov/articles/PMC9558580/

35. Recombination-Driven Genome Evolution and Stability of Bacterial Species - PMC - NIH, accessed January 11, 2026, https://pmc.ncbi.nlm.nih.gov/articles/PMC5586378/

36. Coalescent-based species delimitation meets deep learning: Insights from a highly fragmented cactus system | bioRxiv, accessed January 11, 2026, https://www.biorxiv.org/content/10.1101/2020.12.23.424219.full

37. Unsupervised Machine Learning for Species Delimitation, Integrative Taxonomy, and Biodiversity Conservation | bioRxiv, accessed January 11, 2026, https://www.biorxiv.org/content/10.1101/2023.06.12.544639v1

38. A demonstration of unsupervised machine learning in species delimitation, accessed January 11, 2026, https://repository.cshl.edu/id/eprint/38392/

39. Species Delimitation Using Genomic Data: Options and Limitations - PMC - NIH, accessed January 11, 2026, https://pmc.ncbi.nlm.nih.gov/articles/PMC11974488/

40. Comparative approaches to understanding speciation: a case study in Xantusia lizards | Evolutionary Journal of the Linnean Society | Oxford Academic, accessed January 11, 2026, https://academic.oup.com/evolinnean/article/3/1/kzae024/7777004

41. Genomic divergence during speciation: causes and consequences - The Royal Society, accessed January 11, 2026, https://royalsocietypublishing.org/rstb/article/367/1587/332/21763/Genomic-divergence-during-speciation-causes-and

42. A Genomic Perspective on Species Delimitation - R Discovery - Researcher.Life, accessed January 11, 2026, https://discovery.researcher.life/article/a-genomic-perspective-on-species-delimitation/fb8db7fba3af3bac9e695c91d47ed225