“Taxonomy is classification, as well as the how-to of classification, and its rationale” (Groves 2004). That is to say, it is the study of why we classify organisms, and how best to do it; “a taxonomy” is often used to mean simply a classification. It is, in its more theoretical aspect, nearly the same as systematics, which is generally taken to encompass much more about evolution and biodiversity.

The basis of taxonomy is the species. Species are the units of the living world, but how we define and recognize these units is controversial to perhaps an unexpected degree. What we mean by species affects genetics, biogeography, population biology, ecology, and ethology, in the present-day sphere; paleontology and paleoanthropology; and, in an era in which threats to the natural world and its biodiversity are ubiquitous and accelerating, it affects conservation strategies (Rojas 1992). And so we have the flourishing field of “species concepts.” Mayden (1997) distinguished 24 different species concepts; as many of these are simply variants of each other, I will talk only about 5.

The Evolutionary Species Concept was proposed by Simpson (1961): “A lineage . . . evolving separately from others and with its own evolutionary role and tendencies.” Commentators such as Mayden (1997), Groves (2001a), de Queiroz (2007), and Groves and Grubb (2011) agree that this definition goes to the heart of the matter and encapsulates what a species really is, but of course it is little help in a practical sense. What we need, in effect, is a way of recognizing an evolutionary species when we see one.

Under the Biological Species Concept (BSC), species are defined as “groups of actually or potentially interbreeding natural populations which are reproductively isolated from other such groups” (Mayr 1963). Hence, what distinguishes a species is its reproductive isolation. The nature of reproductive isolating mechanisms (RIMs) was discussed by Dobzhansky (1937), who first coined the term; they may be either premating or postmating. Premating RIMs include mechanisms that prevent potential mates from meeting (seasonal or ecological), those that prevent mating if they do meet (ethological), and those that prevent sperm transfer if they get as far as mating (mechanical). Postmating RIMs include embryo mortality, hybrid inviability, and hybrid sterility. A widespread misunderstanding of this is that “different species cannot interbreed, or if they do their hybrids are sterile.” It should be clearly understood that this is a misunderstanding. Mayr (1963) devoted 2 entire chapters of his classic textbook to trying to explain why it is a misunderstanding— to no avail, it would seem, as far as many people are concerned, including all too many who are biologically trained!

The BSC works well if 2 taxa are sympatric (that is, they occupy the same geographic area). Then they are, or appear to be (see below), reproductively isolated. They are distinct species, and no further discussion is possible. If they are parapatric (they occupy bordering areas), then equally they have the chance to interbreed, and do not. But what if they are allopatric (they occupy discrete areas)? Several authors have emphasized that the BSC is simply not applicable in cases of allopatry (see, especially, Groves 2004), and this is perhaps the most cogent reason the BSC cannot be the species concept. Another reason is, quite simply, that mtDNA analysis has shown that there is widespread interbreeding in nature between different species. DNA work has shown very clearly that there was interbreeding between different lineages in the past, and today even widely sympatric species do often interbreed, but looking at phenotypes, one would never know it.

Under the Recognition Species Concept (RSC), proposed by Hugh Paterson (1978, 1982), species are recognized by their unique specific mate recognition system (SMRS). In such a system, one individual emits a signal, to which another responds. A species would be a population (or group of populations) in which the members share a common SMRS.

But how to recognize SMRSs in practice? One proposal that has been widely followed, and has proved extremely fruitful, has been to analyze vocalizations. Bush babies, or galagos (Galagidae), were conventionally divided into 6 species (Napier and Napier 1967), but throughout the 1980s and 1990s the number was increased, in particular because studies of vocalizations had alerted researchers to the coexistence of sympatric species pairs (Masters 1991; Nash et al. 1989). As fieldwork progressed, it became evident that allopatric populations within what had previously been deemed unitary species were also, very often, distinguished by unique vocalization types, so that Bearder et al. (1995) felt able to predict the existence of new species. One of these previously undescribed species, first identified by its vocalizations, was indeed subsequently found to be morphologically distinct. It was duly described as such, under the name Galagoides rondoensis (Honess 1996). Further studies are under way by Simon Bearder’s team at Oxford Brookes University to test the reality of other galago species predicted from their vocalizations.

There are several problems with the RSC. Would one, in every case, know what actually is a mate recognition signal? Once the signals—vocalizations, splashes of color, and so on—are identified, how different do they have to be in order for one to say that there are 2 different species involved? (We face the same problem with morphology and with genetic distances.) And there is always the problem that if 2 populations happen to have the same apparent SMRS, however different they may be in other respects, some overzealous practitioner of the RSC is going to unite them into a single species.

The RSC has more than proved its value in primatology, yet it has its problems, as I have indicated. In particular, it may fail, for practical reasons or simply because it does not go far enough, to identify the full range of biological units that we seem to mean when we speak of species.

It was because the failings of the BSC had become too obvious to ignore that Cracraft (1983) proposed the Phylogenetic Species Concept (PSC). This has gone through much discussion, and differing authors have had different views of it, but a basic definition is that a species is “an irreducible cluster of sexual organisms within which there is a parental pattern of ancestry and descent and that is diagnosably distinct from other such clusters by a unique combination of fixed characters” (Christoffersen 1995). “Diagnosably distinct” means, quite simply, that they are 100% diagnosable (given age/sex variation); they have fixed heritable differences between them; they are genetically isolated, though not necessarily reproductively isolated. The designation “phylogenetic” was given because, under this concept, species are the terminal points on a cladogram; that is to say, they are the least inclusive phylogenetic units. It has sometimes been misunderstood as implying that any species must have a long phylogenetic history, but in fact all that is necessary is that each species is absolutely distinguishable from any other.

Essentially, the BSC can be characterized as a species concept for the lumpers—it sees the wood but tends to miss the individual trees. On the contrary, the PSC is a concept for the splitters, who are more concerned about identifying the individual trees, thinking it better to leave the wood for a future state of analysis. It has often been pointed out that the PSC and RSC commonly identify the same units as species. The advantage of the PSC is that it can be used in cases where we do not know anything about SMRSs, or even can never know anything about them (as in the case of fossils), and that it identifies as species many cases that the RSC will miss because they happen not to differ in features that are known or suspected to act as SMRSs in their relatives. And it is necessary to identify the units of biodiversity; the conservation crisis reminds us of that.

The idea that species ought to be more differentiated genetically than infraspecific groups (mere subspecies or populations) has a long history going back to Ayala (1975), who compared values for genetic distance, using various proteins, for various animal groups. Using invertebrates, fish, salamanders, lizards, and rodents as test cases, he showed that species tend to differ more than do subspecies, which in turn differ more than do different local populations. There are overlaps, however. One simply cannot say that a certain degree of difference indicates species, a lesser degree indicates subspecies, and so on. In more recent times, the availability of DNA sequencing has provided an enormous amount of data on which to perform similar exercises, but unfortunately, what started off as a rule of thumb has often been used in a rather rigid way. Curiously, this idea of species remained nameless until a key paper by Bradley and Baker (2001) called it the Genetic Species Concept (GSC). Bradley and Baker tested the amount of sequence divergence in mtDNA among bats and rodents, finding, as had Ayala, overlapping values between species and infraspecific units; moreover, different species of rodents seem to be characteristically associated with higher sequence-divergence values, compared with different species of bats.

Therefore, although the average species pair differs more (at least in mtDNA) than does the average intraspecific (including subspecies) pair, it is simply not possible to draw a line at some level of sequence divergence and say that above this level they are species, below it they are not. Differences in sampling or in taxonomic practice could account for some of the variation, and high intraspecific genetic distances could be due to retention of ancestral polymorphism and unusual delays in lineage sorting; but we must simply adjust to the fact that even well-differentiated species pairs may show extraordinarily little sequence divergence in a given DNA region.

In a later paper, Baker and Bradley (2006) went on to consider the GSC in more detail. A genetic species, in their definition, is “a group of genetically compatible interbreeding natural populations that is genetically isolated from other such groups.” This makes it very close to the PSC; indeed, I maintain (above) that the quality of being genetically isolated (having “fixed heritable differences”) is part of the consequences of the PSC. Yet Baker and Bradley (2006) continue to urge the relevance of genetic distances. This, as I have explained above, is a subjective notion (how much difference does there have to be between species?); yet, it has a heuristic value, because there are cases where genetic data alone suffice to identify species. There is no doubt that the use of genetic-inspired concepts has been enormously valuable in uncovering cryptic diversity in nocturnal Malagasy lemurs. Still, when all is said and done, species identifications under the GSC, like those under the RSC, fall well within the parameters established for the PSC, and when we think about the implications of each species concept, we can see that species are objective, testable entities only under the PSC. All the other proposed species concepts involve greater or lesser degrees of subjectivity and hypothesis. The fundamental importance of the idea of the species makes it absolutely essential that these categories be as testable as possible. Hence, I recommend that primate species be recognized according to the following definition: “A species is a diagnosable cluster of individuals within which there is a parental pattern of ancestry and descent, beyond which there is not, and which exhibits a pattern of phylogenetic ancestry and descent among units of like kind” (Eldredge and Cracraft 1980). Or, more succinctly, a species is “a population or aggregation of populations which has fixed heritable differences from other populations or aggregations of populations” (Groves and Grubb 2011).

Having identified the units (the species), we then classify them in groups called genera (singular, genus). At present, the only universally agreed upon stipulation is that a genus must be monophyletic, meaning descended from an exclusive common ancestor. Unfortunately, this still leaves genera rather subjective. Which ancestor do you choose: the last common ancestor of the most closely related species, or some ancestor a little farther back in time, which was also the ancestor of other species? For example, should we include in the genus Callithrix all the descendants of the ancestor of all marmosets, or just those of the ancestor of the Callithrix jacchus group? For that matter, why not extend the genus back to include all descendants of the way-back common ancestor of all marmosets and tamarins, and why not include tamarins as well as marmosets in the genus Callithrix? In all 3 cases, the genus Callithrix is monophyletic. Which common ancestor should we take as the relevant common ancestor? To end this subjectivity, Goodman et al. (1998) revived an old idea that a genus should have a certain time depth; they suggested about 7 million years, but Groves (2001a,b) proposed to lower this limit to 4 million years.

Wouldn't this depend on the quality of the fossil record? No. We do not in point of fact need a fossil record at all to decide how long ago the common ancestor of a particular group of species lived. Used cannily, the molecular clock can give us this information.

We then go on to classify genera into families. Like a genus, a family has to be monophyletic, and exactly as in the case of a genus, how inclusive this monophyly should be is purely subjective. Goodman et al. (1998) suggested that a family, too, should have a particular time depth, somewhere around the Oligocene-Miocene boundary (23 million years ago), and Groves (2001a,b) suggested lowering this a little.

A monophyletic group of families is clustered into an order. Again, the time depth criterion was used by Goodman et al. (1998) to try to introduce some objectivity, placing it at about the Cretaceous-Tertiary boundary (65.5 million years ago), which is more or less standard practice for mammalian orders in general. The order Primates has certainly been separate since then.

With more than 400 species of primates recognized, we now need finer degrees of classification than just dividing the primates into families, the families into genera, the genera into species. By common consent, the order Primates is divided into suborders, these into infraorders, these into superfamilies, these into families, these into subfamilies, these into tribes, these into subtribes, and these into genera. Some taxonomists recognize subgenera within genera. Orders, families, genera, and species are obligatory in every classification, but all the sub-, infra-, and even super- categories are optional; for example, a family of limited diversity like the Daubentoniidae does not need subfamilies and so on, because there is only a single genus, Daubentonia.

Finally, species themselves are often divided up into geographic units with some amount of differentiation between them; these are called subspecies. By “some amount of differentiation” we mean that they are not absolutely different, they overlap in some character states, and they will differ in gene frequencies without any difference being fixed. Thus, the allocation of a particular individual to a subspecies is a matter of probability. For some, this means that subspecies are impossibly subjective, and there is a move, especially among some followers of the PSC, to junk the idea of subspecies altogether. Nonetheless, most taxonomists do see some merit in distinguishing populations within a species that differ from each other “as a whole, though not absolutely,” as long as one does not try to reify them or treat them as entities in the same way as species.

All these taxonomic entities and categories are called taxa (singular, taxon). A species is a taxon, a family is a taxon, a suborder is a taxon, and so on.

Having identified the species, and having at least plausible hypotheses of how they should be arranged into a series of nested monophyletic groups, there remains the problem of what names one should give the groups. This is nomenclature. It is vital to draw a line between taxonomy and nomenclature. Taxonomy is an attempt to represent nature, whereas nomenclature is an invention of the human mind, intended to serve the needs of taxonomy. There are rules for how one gives names; the rules are decided by the International Commission on Zoological Nomenclature and laid out in the International Code of Zoological Nomenclature. The rules are modified, according to need, every so often, so the Code has undergone 4 revisions (the latest, in 1999, took effect in 2000).

The basic rules are as follows:

These rules of nomenclature may seem cumbersome and unnecessary, but they are actually quite helpful. They are “designed to provide the maximum stability compatible with taxonomic freedom,” in the words of the preamble to the fourth edition of the Code. “The Code refrains from infringing upon taxonomic judgment, which must not be made subject to regulation or restraint.” The Code is here reminding everybody that nomenclature and taxonomy are different fields of endeavor. It also makes an important point about taxonomy: a taxonomy is a working hypothesis. There are more plausible or less plausible taxonomic hypotheses, but there can be no “official” taxonomy.

Biography

Colin Groves earned a Ph.D. from London University in 1966, spent two years as a post-doc in the University of California at Berkeley, then took up a fixed term appointment at Cambridge University, and in 1974 came to the School of Archaeology and Anthropology at the Australian National University in Canberra. He has done field work in Kenya, Tanzania, Rwanda, Indonesia and Iran. His specialty is taxonomy and phylogeny of primates, including human evolution. He has also done a great deal of work on ungulates, and some on carnivores, rodents, marsupials and monotremes. For nearly 30 years, he has been a member of the Australian Skeptics, which combats pseudoscience, especially creationism, and those aspects of "traditional medicine" regimes that use animal body parts as supposed cures and which have been so damaging to wildlife. He has published around 200 peer-reviewed scientific papers, and a number of books and monographs including the influential Primate Taxonomy in 2001; Bones, Stones and Molecules with David W.Cameron in 2004; Extended Family: Long Lost Cousins. A Personal Look at the History of Primatology in 2008; Ungulate Taxonomy with the late Peter Grubb in 2011; Climate, Fire and Human Evolution with Andrew Glikson in 2015; and Taxonomy of Australian Mammals with Stephen Jackson in 2015. He retired at the end of 2015, and during that year a Festschrift was published, Taxonomic Tapestries (edited by Alison M.Behie and Marc F.Oxenham). After his retirement, he continued to be active as an Emeritus Professor at the Australian National University. Dr. Groves passed away in November 2017.