Living organisms are classified through a hierarchical system of ranked groups, from broad domains down to individual species, based on shared ancestry inferred primarily from genetic data. The framework most people encounter in school, with its familiar kingdom-phylum-class-order-family-genus-species ladder, dates back to the eighteenth century, but the rules for sorting organisms into those ranks have changed dramatically. What began as a system built on physical appearance has become one driven by DNA sequences, and the shift has redrawn boundaries across the entire tree of life in ways that are still unfolding.
The Naming System That Started It All
Modern biological classification traces to the Swedish naturalist Carl Linnaeus, who in the mid-1700s formalized the practice of giving every organism a two-part Latin name: a genus and a species. That naming convention, called binomial nomenclature, is still in use today. Your house cat is Felis catus, the common daisy is Bellis perennis, and every newly discovered species gets the same treatment. Linnaeus also grouped organisms into nested ranks, placing similar species into a genus, similar genera into an order, and so on up to the kingdom level.1Europe PMC / EMBO reports. There shall be order. The legacy of Linnaeus in the age of molecular biology.
For most of the system’s history, those groupings relied on what organisms looked like. Animals with feathers went together. Plants with similar flower structures were placed in the same family. Linnaeus himself recognized two kingdoms: plants and animals. By the mid-twentieth century, biologists realized this was too crude. Fungi, for instance, were lumped with plants despite being metabolically and genetically closer to animals. In 1969, Robert Whittaker proposed a five-kingdom system that separated fungi, added a kingdom for single-celled organisms with nuclei, and created one for bacteria. The revision reflected a growing consensus that physical similarity alone could not capture how organisms were actually related to one another.2PubMed. New concepts of kingdoms of organisms
When Molecules Rewrote the Tree of Life
The biggest upheaval came in the late 1970s, when Carl Woese and George Fox compared ribosomal RNA sequences across a wide range of organisms. Ribosomes are molecular machines found in every living cell, and the genes encoding them change slowly enough to serve as a kind of evolutionary clock. Woese’s analysis revealed that life falls into three fundamentally distinct lineages: the eubacteria (ordinary bacteria), the archaebacteria (a group of microbes that had been lumped in with bacteria but turned out to be genetically as different from them as humans are), and the eukaryotes (everything with complex cells, from amoebas to oak trees).3PubMed Central. Phylogenetic structure of the prokaryotic domain: the primary kingdoms This three-domain system, now the standard framework, could not have been discovered by examining organisms under a microscope. It took molecular data to see it.
The reason molecular data proved so powerful is that physical features can be deeply misleading. Organisms facing similar environments often evolve similar body shapes and structures independently, a phenomenon called convergent evolution. Cave-dwelling arachnids in eastern North America, for example, share a suite of traits like reduced eyes and elongated legs, but molecular analysis has shown that these features evolved independently in at least three separate lineages rather than being inherited from a shared cave-adapted ancestor.4PubMed. Molecular systematics of eastern North American Phalangodidae (Arachnida: Opiliones: Laniatores), demonstrating convergent morphological evolution in caves A similar story plays out among Brazilian worm lizards, where the physical characters traditionally used to distinguish genera turned out to be so riddled with convergence that the entire taxonomy built on them was misleading.5PubMed. Molecular phylogenetics reveals extreme morphological homoplasy in Brazilian worm lizards challenging current taxonomy
A large-scale study on mammals helps explain why physical traits mislead so often. The researchers found that about three-quarters of informative morphological characters are simple two-state traits (present or absent, large or small), and characters with only two possible states are inherently prone to looking the same by chance once multiple evolutionary changes have occurred. Molecular characters, by contrast, typically have five or more possible states, making accidental matches far less likely.6PubMed Central. Morphological and molecular convergences in mammalian phylogenetics It is not that physical features are inherently worse indicators of relatedness; it is that they have fewer possible forms, so unrelated lineages stumble into the same appearance more often.
What Counts as a Species
Ask a biologist to define “species” and you will get a longer answer than you expected. There are over two dozen species concepts in use, but the two that matter most in practice are the Biological Species Concept (BSC) and the Phylogenetic Species Concept (PSC). The BSC defines a species as a group of organisms that can interbreed and produce fertile offspring. The PSC defines it as the smallest group of organisms that shares a unique evolutionary history. These two approaches do not always agree, and the disagreements are not trivial.
A comparison of shrike-babblers, a group of small Asian birds, illustrates the gap. Five species were traditionally recognized under the BSC. Under the PSC, researchers split them into 19. A reanalysis incorporating vocal data alongside genetic and morphological evidence settled on 9 species under the BSC, and the study’s authors noted that the PSC total was likely inflated because traits that appeared unique to certain populations were artifacts of small sample sizes.7Zootaxa. Species limits in Pteruthius (Aves: Corvida) shrike-babblers: a comparison between the Biological and Phylogenetic Species Concepts The choice of species concept can double or triple the number of species counted in a group, with direct consequences for conservation funding and legal protection.
Some organisms seem designed to break any species definition. Ring species form when a population expands around a geographic barrier, like a mountain range, with neighboring populations able to interbreed along the way but the two ends of the ring unable to do so when they meet. Greenish warblers encircling the Tibetan Plateau are a classic case.8PubMed Central. Evolution and stability of ring species A California lily complex shows a similar pattern along a south-to-north range expansion, with populations that remain connected yet have differentiated from each other along the way.9Journal of Biogeography. Phylogeography and population genetics reveal ring species patterns in a highly polymorphic California lily Are the two ends one species or two? The BSC says two (they cannot interbreed), but there is no clean point along the ring where you can draw the line.
Then there are cryptic species, organisms that look identical to the eye but are genetically distinct and reproductively isolated. Bornean fanged frogs are a striking example: what was considered a single species has been split into 18 genetically divergent but morphologically indistinguishable species.10Systematic Biology. A Genomic Perspective on Cryptic Species Reveals Complex Evolutionary Dynamics in the Gray Zone of the Speciation Continuum Cryptic species are not rare curiosities. The routine use of molecular methods over the past two decades has uncovered enormous numbers of them, particularly in aquatic environments, and their existence matters beyond academic classification. Ecotoxicology studies, for instance, can produce misleading results when researchers unknowingly test a single morphospecies that actually encompasses multiple cryptic species with different tolerances to pollutants.11Environmental Pollution. Cryptic species complex shows population-dependent, rather than lineage-dependent tolerance to a neonicotinoid
Classifying the Invisible World
If defining species is difficult for birds and frogs, it becomes almost philosophically fraught for microbes. Bacteria reproduce asexually, which means the “can they interbreed” test does not apply. Worse, bacteria routinely swap genes with unrelated species through horizontal gene transfer, blurring the lines between lineages in ways that have no real counterpart among animals or plants.12Genome Biology and Evolution. Reconstructing the Network of Horizontal Gene Exchange in Bacteria to Differentiate Direct and Indirect Transfers
To cope with this, microbiologists rely heavily on genome-level comparison tools. Average nucleotide identity, or ANI, compares the overall DNA similarity between two genomes. A widely used threshold places organisms in the same species if their genomes share roughly 95 to 96 percent identity.13PubMed Central. Using average nucleotide identity to improve taxonomic assignments in prokaryotic genomes at the NCBI But those thresholds are not universal. In the genus Streptomyces, a group of soil bacteria that produce many of the world’s antibiotics, detailed analysis revealed that the standard cutoffs needed adjustment: a threshold closer to 96.7 percent ANI was more appropriate, and borderline cases still required additional lines of evidence to resolve.14PubMed Central. New Insights Into the Threshold Values of Multi-Locus Sequence Analysis, Average Nucleotide Identity and Digital DNA-DNA Hybridization in Delineating Streptomyces Species There is no single magic number that cleanly separates every pair of bacterial species.
Viruses pose an even stranger challenge. They are not cells, they have no ribosomes, and whether they count as “living” depends on your definition. The International Committee on the Taxonomy of Viruses (ICTV) has overhauled its classification system in recent years, expanding from five ranks to fifteen to accommodate the enormous genetic diversity of the virosphere. The shift has moved virus classification from an older approach based on physical characteristics and disease symptoms toward one based primarily on gene sequences and evolutionary relationships.15Veterinary Science Today. Nature of viruses and the radical change in viral taxonomy Even so, the ICTV acknowledges that strict hierarchical classification may not work for all viruses, and some groupings may need to sit outside the traditional tree structure entirely.
Ancient DNA Corrects the Record
Paleontology has traditionally classified fossils by their shapes, because shape is often all a fossil preserves. Ancient DNA technology has made it possible to check that work, and the corrections have sometimes been dramatic. A study of Eurasian wild asses found that ancient DNA from fossil bones simplified their family tree considerably, revealing that populations with very different body shapes shared the same genetic lineage while some populations that looked alike were genetically distinct.16PubMed. Eurasian wild asses in time and space: morphological versus genetic diversity The physical diversity of these animals turned out to reflect adaptability within a species rather than the existence of separate species.
Ancient DNA has also confirmed identifications that bone shape alone could not settle. Giant salamanders are notoriously difficult to tell apart from their skeletons, and their fossil record is incomplete. Researchers who extracted ancient DNA from giant salamander fossils found in a cave on Shikoku Island, Japan, confirmed that the fossils belonged to the extant Japanese giant salamander, Andrias japonicus, resolving what had been an open question based on bones alone.17PubMed Central. Ancient DNA integrates fossil and modern giant salamander taxonomy In another case, a deer specimen found during construction of the Toronto subway decades ago had been described as a unique species based on its distinctive antler shape. Ancient DNA analysis showed it belongs to the same genus as modern white-tailed deer but with enough genetic divergence to potentially represent a distinct, now-extinct species, adding to the list of Ice Age megafauna lost from North America.18bioRxiv. Ancient DNA of the Toronto Subway Deer Adds to the Extinction List of Ice Age Megafauna
The Modern Toolkit
Today’s classification work draws on a range of molecular tools that would have been unimaginable a generation ago. DNA barcoding uses a short, standardized gene region to identify species, much like a supermarket scanner reads product codes. For animals, the most commonly used barcode region is a stretch of the mitochondrial COI gene, chosen because verified reference sequences exist for an enormous number of species.19PubMed Central. DNA metabarcoding and the cytochrome c oxidase subunit I marker: not a perfect match Environmental DNA, or eDNA, takes this a step further. Instead of catching or collecting organisms, researchers can scoop up water or soil samples and sequence all the DNA present, identifying organisms from the genetic traces they leave behind. In principle, eDNA metagenomics can detect virtually any organism whose DNA is in the sample.20PubMed Central. Improving Whole Biodiversity Monitoring and Discovery With Environmental DNA Metagenomics
Whole-genome sequencing adds another layer. Comparing entire genomes rather than single gene regions can resolve relationships that barcoding cannot, particularly among closely related species where a single gene might not carry enough information to distinguish them. Even whole genomes, however, sometimes tell conflicting stories: different parts of the genome can suggest different family trees, a phenomenon known as gene-tree discordance. Teasing apart the causes of that discordance, whether it reflects ancient hybridization, incomplete sorting of ancestral variation, or something else, is an active area of research.21Systematic Biology. Whole Genomes Reveal Evolutionary Relationships and Mechanisms Underlying Gene-Tree Discordance in Neodiprion Sawflies
Hybridization and Plants That Break the Rules
Plants introduce their own classification headaches. Many plant lineages have undergone polyploidy, the duplication of entire genomes, sometimes combined with hybridization between different species. When two species hybridize and their offspring end up with both parents’ complete genomes, the result is a new lineage that does not fit neatly into a branching tree. European woodrushes in the genus Luzula are a case study: their evolutionary history is a tangle of polyploidy, hybridization, and a specialized form of chromosome fragmentation, all wrapped in species that look nearly identical to one another.22Systematic Biology. Hybridization and Polyploidy Shaped the Evolutionary History of a Complex of Cryptic Species in European Woodrushes (Luzula sect. Luzula) Standard tree-building methods assume evolution is a branching process, but when lineages merge as well as split, the result is less a tree than a tangled web.
The Synonym Problem and Why Classification Affects Conservation
One practical consequence of classification’s messy history is the staggering number of synonyms, different scientific names that have been applied to the same species over the centuries. Among butterflies alone, each of the roughly 19,000 recognized species has an average of six synonyms, and individual species can have far more. The common Palearctic butterfly Plebejus argus has accumulated about 160 different names.23ScienceDirect (Trends in Ecology & Evolution). A globally integrated structure of taxonomy to support biodiversity science and conservation This is not just a bookkeeping nuisance. When different names are used for the same species in different countries or databases, conservation assessments can accidentally count one species twice or miss it entirely.
Taxonomic decisions also directly shape conservation priorities. When a population that was previously considered a subspecies gets elevated to full species status, it often gains access to legal protections and funding that were not available before. Researchers have documented numerous cases where a change in taxonomy led to increased conservation efforts, in groups as varied as plants, birds, frogs, dolphins, and giraffes.24Biological Conservation. The impact of taxonomic change on conservation: Does it kill, can it save, or is it just irrelevant? This can cut both ways. A thorough understanding of the taxonomic basis of species lists is critical to conservation planning, because merging species can remove protections just as splitting them can create new ones.25Bird Conservation International. Taxonomy is important in conservation: a preliminary reassessment of Philippine species-level bird taxonomy
Machine Learning Steps In
Identifying species has traditionally required expert knowledge that takes years to develop, and there are far fewer trained taxonomists in the world than there are species to catalog. Machine learning, particularly deep learning applied to images, is helping close that gap. Computer vision systems can now classify images of organisms rapidly, automatically, and with high accuracy, making large-scale biodiversity surveys more practical.26PubMed. A gentle introduction to computer vision-based specimen classification in ecological datasets Apps that let you photograph a plant or insect and receive a species-level identification in seconds are consumer-facing versions of this technology.
Research in this area has focused on training models with diverse morphological traits rather than single features, because a system trained only on leaf shape will struggle with species that differ mainly in flower color or bark texture.27Ecological Informatics. Advances in machine learning models for plant species identification: A scoping review Deep learning breakthroughs over the past several years have pushed accuracy high enough that automated identification is now a genuine complement to human expertise for many groups of organisms.28Methods in Ecology and Evolution. Machine learning for image based species identification These systems do not replace taxonomists, though. They are trained on datasets that taxonomists built, and they struggle with species that are rare, visually cryptic, or underrepresented in reference databases. When an AI model encounters something genuinely new, a human expert still needs to decide where it belongs.
How Everyday People Already Classify Life
Scientists are not the only ones who sort organisms into categories. Every human culture develops its own folk taxonomy, and these systems are far from arbitrary. A classic study of plant names used by Tzeltal-speaking people in Chiapas, Mexico, found that about a third of their native plant names corresponded one-to-one with a single botanical species, while others grouped multiple species under one name or subdivided a single species into finer categories. The direction of the mismatch was not random: culturally important plants were more finely subdivided, while less significant plants were lumped together.29Science. Folk taxonomies and biological classification This pattern suggests that human attention shapes classification. We carve nature at the joints that matter to us, and those joints are not always the same ones that evolutionary history carved.
This tendency persists even in scientific practice, where well-funded, charismatic groups like birds and mammals are classified in far greater detail than invertebrates or fungi. The gaps in coverage matter because the organisms we classify least carefully are often the ones that play the most critical ecological roles. Soil bacteria, deep-sea invertebrates, and tropical fungi remain poorly cataloged compared to the vertebrates that fill field guides, even though molecular tools have made cataloging them more feasible than ever.