I had visited every state but Idaho. A few months ago, I finally got my opportunity to complete the roster of 50 by driving east from Spokane, Washington, into western Idaho. As I crossed the state line, I made the same feeble attempt at humor that so many of us try in similar situations: Gee, it doesn’t look a bit different from easternmost Washington. We make such comments because we feel the discomfort of discord between our mental needs and the world’s reality. Much of nature (including terrestrial real estate) is continuous, but both our mental and political structures require divisions and categories. We need to break large and continuous items into manageable units. Many people feel the same way about species as I do about Idaho-- but this feeling is wrong. Many people suppose that species must be arbitrary divisions of an evolutionary continuum in the same way that state boundaries are conventional divisions of unbroken land. Moreover, this is not merely an abstract issue of scientific theory but a pressing concern of political reality. The Endangered Species Act, for example, sets policy (with substantial teeth) for the preservation of species. But if species are only arbitrary divisions in nature’s continuity, then what are we trying to preserve and how shall we define it? I write this article to argue that such a reading of evolutionary theory is wrong and that species are almost always objective entities in nature. Let us start with something uncontroversial: the bugs in your backyard. If you go out to make a complete collection of all the kinds of insects living in this small discrete space, you will collect easily definable packages, not intergrading continua. You might find a kind of bee, three kinds of ants, a butterfly or two, several beetles, and a cicada. You have simply validated the commonsense notion known to all: in any small space during any given moment, the animals we see belong to separate and definable groups--and we call these groups species. In the eighteenth century this commonsense observation was translated, improperly as we now know, into the creationist taxonomy of Linnaeus. The great Swedish naturalist regarded species as God’s created entities, and he gathered them together into genera, genera into orders, and orders into classes, to form the taxonomic hierarchy that we all learned in high school (several more categories, families and phyla, for example, have been added since Linnaeus’s time). The creationist version reached its apogee in the writings of America’s greatest nineteenth-century naturalist (and last truly scientific creationist), Louis Agassiz. Agassiz argued that species are incarnations of separate ideas in God’s mind, and that higher categories (genera, orders, and so forth) are therefore maps of the interrelationships among divine thoughts. Therefore, taxonomy is the most important of all sciences because it gives us direct insight into the structure of God’s mind. Darwin changed this reverie forever by proving that species are related by the physical connection of genealogical descent. But this immensely satisfying resolution for the great puzzle of nature’s order engendered a subsidiary problem that Darwin never fully resolved: If all life is interconnected as a genealogical continuum, then what reality can species have? Are they not just arbitrary divisions of evolving lineages? And if so, how can the bugs in my backyard be ordered in separate units? In fact, the two greatest evolutionists of the nineteenth century, Lamarck and Darwin, both questioned the reality of species on the basis of their evolutionary convictions. Lamarck wrote, In vain do naturalists consume their time in describing new species; while Darwin lamented: we shall have to treat species as . . . merely artificial combinations made for convenience. This may not be a cheering prospect; but we shall at least be freed from the vain search for the undiscovered and undiscoverable essence of the term species (from the Origin of Species). But when we examine the technical writings of both Lamarck and Darwin, our sense of paradox is heightened. Darwin produced four long volumes on the taxonomy of barnacles, using conventional species for his divisions. Lamarck spent seven years (1815-1822) publishing his generation’s standard, multivolume compendium on the diversity of animal life--Histoire naturelle des animaux sans vertèbres, or Natural History of Invertebrate Animals--all divided into species, many of which he named for the first time himself. How can these two great evolutionists have denied a concept in theory and then used it so centrally and extensively in practice? To ask the question more generally: If the species is still a useful and necessary concept, how can we define and justify it as evolutionists? The solution to this question requires a preamble and two steps. For the preamble, let us acknowledge that the conceptual problem arises when we extend the bugs in my backyard example into time and space. A momentary slice of any continuum looks tolerably discrete; a slice of salami or a cross section of a tree trunk freezes a complexly changing structure into an apparently stable entity. Modern horses are discrete and separate from all other existing species, but how can we call the horse (Equus caballus) a real and definable entity if we can trace an unbroken genealogical series back through time to a dog-size creature with several toes on each foot? Where did this dawn horse, or eohippus, stop and the next stage begin; at what moment did the penultimate stage become Equus caballus? I now come to the two steps of an answer. First, if each evolutionary line were like a long salami, then species would not be real and definable in time and space. But in almost all cases large-scale evolution is a story of branching, not of transformation in a single line--bushes, not ladders, in my usual formulation. A branch on a bush is an objective division. One species rarely turns into another by total transformation over its entire geographic range. Rather, a small population becomes geographically isolated from the rest of the species--and this fragment changes to become a new species while the bulk of the parental population does not alter. Dawn horse is a misnomer because rhinoceroses evolved from the same parental lineage. The lineage split at an objective branching point into two lines that became (after further events of splitting) the great modern groups of horses (eight species, including asses and zebras) and rhinos (a sadly depleted group of formerly successful species). Failure to recognize that evolution is a bush and not a ladder leads to one of the most common vernacular misconceptions about human biology. People often challenge me: If humans evolved from apes, why are apes still around? To anyone who understands the principle of bushes, there simply is no problem: the human lineage emerged as a branch, while the rest of the trunk continued as apes (and branched several more times to yield modern chimps, gorillas, and so on). But if you think that evolution is a ladder or a salami, then an emergence of humans from apes should mean the elimination of apes by transformation. Second, you might grasp the principle of bushes and branching but still say: Yes, the ultimate products of a branch become objectively separate, but early on, while the branch is forming, no clear division can be made, and the precursors of the two species that will emerge must blend indefinably (figure 1). And if evolution is gradual and continuous, and if most of a species’ duration is spent in this state of incipient formation, then species will not be objectively definable during most of their geologic lifetimes. Fair enough as an argument, but the premise is wrong. New species do (and must) have this period of initial ambiguity. But species emerge relatively quickly, compared with their period of later stability, and then live for long periods--often millions of years--with minimal change (figure 2). Now, suppose that on average (and this is probably a fair estimate), species spend one percent of their geologic lifetimes in this initial state of imperfect separation. Then, on average, about one species in a hundred will encounter problems in definition, while the other 99 will be discrete and objectively separate--cross sections of branches showing no confluence with others (C, figure 1). Thus, the principle of bushes, and the speed of branching, resolve the supposed paradox: continuous evolution can and does yield a world in which the vast majority of species are separate from all others and clearly definable at any moment in time. Species are nature’s objective packages. I have given a historical definition of species--as unique and separate branches on nature’s bush. We also need a functional definition, if only because historical evidence (in the form of a complete fossil record) is usually unavailable. The standard criterion, in use at least since the days of the great French naturalist Georges de Buffon (a contemporary of Linnaeus), invokes the capacity for interbreeding. Members of a species can breed with others in the same species but not with individuals belonging to different species. This functional criterion is a consequence of the historical definition: distinct separateness of a branch emerges only with the attainment of sufficient evolutionary distance to preclude interbreeding, for otherwise the branch is not an irrevocably separate entity and can amalgamate with the parental population. Exceptions exist, but the reproductive criterion generally works well and gives rise to the standard one-liner for a textbook definition of a species: a population of actually or potentially reproducing organisms sharing a common gene pool. Much of the ordinary activity of evolutionary biologists is devoted to learning whether or not the groups they study are separate species by this criterion of reproductive isolation. Such separateness can be based on a variety of factors, collectively termed isolating mechanisms: for example, genetic programs so different that an embryo cannot form even if egg and sperm unite; behaviors that lead members of one species to shun individuals from other populations; even something so mundane as breeding at different times of the year, or in different parts of the habitat--say, for example, on apple trees rather than on plum trees- -so that contact can never take place. (We exclude simple geographic separation--living on different continents, for example--because an isolating mechanism must work when actively challenged by a potential for interbreeding through spatial contact. I do not belong to a separate species from my brethren in Brazil just because I have never been there. Similarly, reproductive isolation must be assessed by ordinary behavior in a state of nature. Some truly separate species can be induced to interbreed in zoos and laboratories. The fact that zoos can make tiglons--tiger-lion hybrids--does not challenge the separate status of the two populations as species in nature.) Modern humans (species Homo sapiens) fit these criteria admirably. We are now spread all over the world in great numbers, but we began as a little twig in Africa (the historical criterion). We may look quite different from one another in a few superficially striking aspects of size, skin color, and hair form, but there is astonishingly little overall genetic difference among our so-called races. Above all (the functional criterion), we can all interbreed with one another (and do so with avidity, always, and all over the world), but not with any member of another species (movies about flies notwithstanding). We are often reminded, quite correctly, that we are very similar in overall genetic program to our nearest cousin, the chimpanzee--but no one would mistake a single individual of either species, and we do not hybridize (again, various science fictions notwithstanding). I do not say that these criteria are free from exceptions; nature is nothing if not a domain of exceptions, where an example against any clean generality can always be found. Some distinct populations of plants, for example, can and frequently do interbreed with others that ought to be separate species by all other standards. (This is why the classification of certain groups--the rhododendrons, for example--is such a mess.) But the criteria work in the vast majority of cases, including humans. Species are not arbitrary units, constructed for human convenience, in dividing continua. Species are the real and objective items of nature’s morphology. They are out there in the world as historically distinct and functionally separate populations with their own historical role and tendency (as the other textbook one-liner proclaims). Species are unique in the Linnaean hierarchy as the only category with such objectivity. All higher units--genera, families, phyla, et cetera--are human conventions in the following important respect. The evolutionary tree itself is objective; the branches (species) emerge, grow, and form clusters by subsequent branching. The clusters (figure 3) are clearly discernible. But the status we award to these so-called higher taxa (clusters of branches with a single root of common evolutionary ancestry) is partly a matter of human decision. Clusters A and B in the figure are groups of species with a common parent. Each branch in each cluster is an objective species. But what are the clusters themselves? Are they two genera or two families? Our decision on this question is partly a matter of human preference constrained by the rules of logic and the facts of nature. (For example, we cannot take one species from cluster A and one from cluster B and put them together as a single genus--for this would violate the rule that all members of a higher taxon must share a common ancestor without excluding other species that are more closely related to the common ancestor. We cannot put domestic cats and dogs in one family while classifying lions and wolves in another.) The taxonomic hierarchy recognizes only one unit below species-- the subspecies. Like higher taxa, subspecies are also partly objective but partly based on human decision. Subspecies are defined as distinctive subpopulations that live in a definite geographic subsection of the entire range of the species. I cannot, for example, pluck out all tall members of a species, or all red individuals, wherever they occur over the full geographic range, and establish them as subspecies. A subspecies must be a distinct geographic subpopulation--not yet evolved far enough to become a separate species in its own right but different enough from other subpopulations (in terms of anatomy, genetic structure, physiology, or behavior) that a taxonomist chooses to memorialize the distinction with a name. Yet subspecies cannot be irrevocably unique natural populations (like full species) for two reasons: First, the decision to name them rests with human taxonomists, and isn’t solely dictated by nature. Second, they are, by definition, still capable of interbreeding with other subpopulations of the species and are, therefore, impermanent and subject to reamalgamation. This difference between species and subspecies becomes important in practice because our Endangered Species Act currently mandates the protection of subspecies as well. I do not dispute the act’s intention or its teeth, for many subspecies do manifest distinctly evolved properties of great value and wonder (even if these properties do not render them reproductively isolated from other populations of the species). We would not, after all, condone the genocide of all Caucasian human beings because members of other races would still exist; human races, if formally recognized at all, are subspecies based on our original geographic separations. But since subspecies do not have the same objective status as species (and since not all distinct local populations bear separate names), argument over what does and does not merit protection is inevitable. Most of the major ecological wrangles of recent years--rows over the Mount Graham red squirrel or the Northern spotted owl--involve subspecies, not species. These taxonomic issues were once abstract, however important. They are now immediate and vital--and all educated people must understand them in the midst of our current crisis in biodiversity and extinction. I therefore close with two observations. By grasping the objective status of species as real units in nature (and by understanding why they are not arbitrary divisions for human convenience), we may better comprehend the moral rationale for their preservation. You can expunge an arbitrary idea by rearranging your conceptual world. But when a species dies, an item of natural uniqueness is gone forever. Each species is a remarkably complex product of evolution--a branch on a tree that is billions of years old. All the king’s horses and men faced an easy problem compared with what we would encounter if we tried to reconstitute a lost species. Reassembling Humpty-Dumpty is just an exceedingly complex jigsaw puzzle, for the pieces lie at the base of the wall. There are no pieces left when the last dodo dies. But all species eventually die in the fullness of geologic time, so why should we worry? In the words of Tennyson (who died exactly 100 years ago, so the fact is no secret): From scarped cliff and quarried stone She cries, A thousand types are gone: I care for nothing. All shall go. (From In Memoriam.) The argument is true, but the time scale is wrong for our ethical concerns. We live our lives within geologic instants, and we should make our moral decisions at this proper scale--not at the micromoment of thoughtless exploitation for personal profit and public harm; but not at Earth’s time scale of billions of years either (a grand irrelevancy for our species’ potential tenure of thousands or, at most, a few million years). We do not let children succumb to easily curable infections just because we know that all people must die eventually. Neither should we condone our current massive wipeout of species because all eventually become extinct. The mass extinctions of our geologic past may have cleared space and created new evolutionary opportunity--but it takes up to 10 million years to reestablish an interesting new world, and what can such an interval mean to us? Mass extinctions may have geologically distant benefits, but life in the midst of such an event is maximally unpleasant-- and that, friends, is where we now reside, I fear. Species are living, breathing items of nature. We lose a bit of our collective soul when we drive species (and their entire lineages with them), prematurely and in large numbers, to oblivion. Tennyson, paraphrasing Goethe, hoped that we could transcend such errors when he wrote, in the same poem: I held it truth, with him who sings To one clear harp in divers tones That men may rise on stepping-stones Of their dead selves to higher things.