What Is a Species? A Category Biology Cannot Quite Define

Each concept works well for some organisms and fails for others, and most working biologists use whichever suits the group they study:

• •The biological species concept, from Ernst Mayr in 1942: groups that actually or potentially interbreed and are reproductively isolated from other such groups. Useless for asexual organisms and for fossils, and strained by widespread hybridisation
• •The morphological concept: organisms that look sufficiently alike. The oldest approach, still the practical default in palaeontology and much fieldwork, and defeated by species that look identical and by ones where the sexes look nothing like each other
• •The phylogenetic concept: the smallest group sharing a common ancestor and distinguishable from other such groups. Precise and tends to split existing species into many, which has consequences for conservation law
• •The ecological concept: a set of organisms occupying the same niche and adapted to the same resources
• •The genotypic cluster concept: a group of organisms forming a distinct cluster in genetic space, with few intermediates, which works for asexual organisms and for messy hybridising ones
• •For bacteria, a working convention based on the similarity of particular gene sequences, adopted because nothing else is workable

Ring species are the classic demonstration that the boundary can be genuinely continuous. A population spreads around a barrier, such as a mountain range, with each neighbouring group able to interbreed with the next, and where the two ends meet on the far side they cannot interbreed at all, so a single continuous chain is one species locally and two species where it closes; the greenish warbler around the Tibetan plateau and the Ensatina salamanders around California's Central Valley are the standard examples, both of them disputed in the details. Hybridisation is far more common than was believed: polar bears and brown bears interbreed, as do coyotes and wolves, and many plant species arose by hybridisation followed by chromosome doubling, including bread wheat. Genomic work has shown that modern humans carry Neanderthal and Denisovan DNA, which means those groups interbred and produced fertile offspring, and whether they should therefore count as separate species is a live disagreement with no fact of the matter to settle it.

A second difficulty is that most species have never been examined. Roughly 1.7 million have been formally described, and estimates of the total on Earth range from about 2 million to over 10 million for eukaryotes, with bacteria and archaea uncountable by any current method. Insects, fungi, nematodes and deep sea organisms dominate the undescribed remainder. Formal description takes a specialist, a published paper and a type specimen deposited in a museum, and the number of taxonomists qualified to do this is declining in most groups, a situation the field calls the taxonomic impediment. DNA barcoding, which identifies organisms from a short standard gene sequence, has accelerated recognition enormously and has revealed that many supposed single species are complexes of several that look identical, which pushes the estimated total upward rather than resolving it.

This would be an academic dispute if conservation law did not attach rights and money to the category. Endangered species legislation in most countries protects species, and sometimes subspecies or distinct populations, so whether a threatened group counts as its own species can determine whether its habitat is protected and whether money is spent on it. Splitting a widespread species into several narrow ones, which the phylogenetic concept tends to do, can create several endangered species out of one secure one, and the phenomenon has a name, taxonomic inflation, and critics who argue it is sometimes motivated by conservation goals rather than by evidence. The reverse happens too: the red wolf's status has been contested for decades on genetic grounds, with its protection depending on whether it is a species, a subspecies or a wolf and coyote hybrid. Similar questions decide whether a population can be hunted, whether an import is legal under wildlife trade rules, and whether a development can proceed.

The philosophical position most biologists settle into is that species are real without being sharply bounded. They are real because the living world is genuinely clumpy: organisms are not distributed evenly across the space of possible forms, and the clusters correspond to real populations that exchange genes internally and not much externally, which is why people in every culture independently recognise and name roughly the same groups of birds and mammals as trained taxonomists do. They are fuzzy because evolution is a continuous process and any attempt to cut a continuous lineage into discrete units will find intermediate cases, which is not a failure of the definition but a prediction of the theory. A boundary problem is exactly what descent with modification should produce, and a world in which every species had a crisp edge would be evidence against evolution rather than for it.

The interbreeding definition fails for asexual organisms, fossils and the many groups that hybridise, so biologists use at least two dozen concepts, choosing by organism: morphological for fossils, phylogenetic for fine distinctions, genetic clusters for bacteria. Ring species and human interbreeding with Neanderthals show the boundary can be genuinely continuous. Around 1.7 million species have been described out of a total that may exceed 10 million, and the definition matters legally because conservation protection attaches to it.