Global species diversity in numbers and what they mean



Briefly reviewed are species concepts, discussing the problems caused by asexual organisms (criteria based on interbreeding do not apply) and evolution (species recognition in spite of change), emphasising bacteria as a group where both problems are prominent. The problems caused by taxonomists for different groups, and different evolutionary times, working on the basis of different species concepts, are discussed, and special scenarios where taxonomy fails to provide satisfactory answers, namely ring and sibling species, presented. Various attempts at quantifying the Earth's biodiversity are described, based on empirical data and various routes of extrapolation. The conclusion argues for various motives for conserving this biodiversity.

To discuss global diversity, it is first of all necessary to show how biodiversity is amenable to counting in spite of its continuous (diverse) nature. Therefore I discuss species concepts in section 1.1. I seek to describe what species concepts have been offered, which ones are actually being used and what the problems are in defining a species. Next, it is important to realise how these concepts are presently applied, which is the focus of section 1.2. I emphasise the major troubles of taxonomy, and their implications for our interpretation of the species counts reviewed in the following section. Current studies on species numbers are presented and discussed that section 2, and I close the essay with the all-important question of what species mean to us, and in particular, what use it is to be counting species in any of the ways currently proposed.

1.1Discussion of species concepts
1.2Difficulties in implementing biological species in taxonomy
2Estimates of global diversity
3Why should we value the diversity of species?

1.1 Discussion of species concepts

Whilst the first species were defined on morphological similarity, the evolutionary thinking pervading the 20th century has precipitated the biological species concept as an ideal. Proposed initially by Dobzhansky (1935), Mayr has offered a good definition:

Species are groups of actually or potentially interbreeding natural populations that are physiologically reproductively isolated from other such groups.

(Mayr 1942)

The biological species concept cannot be applied to asexual organisms, where no interbreeding between individuals takes place. This, however, is achieved by the phenetic species concept. The phenetic species is a set of organisms that resemble one another and are distinct from other sets (Ridley 1996). If we define phenetic species by application of parameters (such as a minimum distance in phenetic space between sets recognised as species, and, by the same token, a maximum distance between one individual and at least one other within the set), we can create reasonably unambiguous (depending on how many dimensions of phenetic space we assume) and universal species. On the other hand, the phenetic species concept does not easily allow for the change of characters as populations evolve, this being a fundamental philosophical problem. This problem is addressed by the recognition species concept, where only those characters that are used by individuals of that species to recognise each other. These characters are necessary for reproduction, individuals lacking them not being recognised as potential mates, thus not reproducing and falling outside the biological species. If the specific mate-recognition system can be characterised, this should be an excellent marker of species (Paterson 1985). Like the phenetic species concept, the recognition concept cannot allow speciation, but at least a limited amount of change to one species through time.

Another species concept relevant especially to the subject of diversity is the ecological species concept:

A species is a lineage that occupies an adaptive zone minimally different from that of any other lineage in its range and which evolves separately from all lineages outside its range.

(van Valen 1976)

The latter could be simplified to apply to asexual species as well, defining a species as a group of similar individuals occupying the same adaptive zone, or showing a Normal distribution across a range of adaptive zones.

One could also look at the proposals of various species concepts as converging on the same answer in time. For instance, all invoke monophyly, meaning they formulate some kind of more or less arbitrary cut-off point along a lineage leading to the individuals under investigation. This is true especially of the phylogenetic species concept, in which monophyly is the only criterion, though not arbitrarily defined. Instead, it disposes of the need for reproductive isolation other than by geography:

A species is the smallest monophyletic group of common ancestry.

(de Queiroz & Donoghue 1990)

Other species concepts that have received some attention are the cohesive species concept (Templeton 1989), which especially considers the problem of lateral gene transfer, and the internodal species concept (Kornet 1993).

The worries about asexual species may be unfounded at least for the taxa of higher eukaryotes that human researchers tend to be most concerned with, since parthenogenesis is rare and in cases where there is only one population of asexual individuals coherent in space, the use of the word population implying a recent common ancestor shared by all asexual individuals, but not shared with any sexual species, a strong case for the delineation of just one asexual species in the genus can be made.

One area where a solution to the problem of asexual species had to be found, was in the biology of microorganisms. Bacteria, for instance, pose a problem as they evolve rapidly, and morphologically similar strains may have very different effects on the environment. The preferred mean of delineating species has become the use of ribosomal RNA and other genetic sequences to establish the genetic distance between two species. If there is more than a 3% difference between two specimens in their ribosomal RNA sequence, they are considered different species (Madigan et al. 2000). The chosen value of 3% is an arbitrary delineation of species, and also disposes of the requirement for species to be monophyletic. Due to the difference in evolutionary rates and species definition (for instance, human and chimpanzee are less than 3% different under the same criterion), it may be better to treat the question of how many bacterial species there are as distinct from the question that is after the number of sexual eukaryotes in particular. The spirit of taxonomy for these groups is so radically different that adding the two values up simply does not give any useful information. In addition, the problem of loss of species may be orders less severe for bacteria very much due to their fast evolutionary rates: They should be able to fill abandoned ecological niches much faster than, for instance, vertebrates, whom we seem much more concerned about. Nonetheless, the question of how diverse microbes are is an interesting one, and I shall address it further on.

However, even between animal phyla, different species concepts, or mixtures thereof, have been used by taxonomists describing those groups, mostly for reasons of convenience: Arthropods, for instance, because their exoskeleton resists decay, can be easily kept as dead specimens in collections, species being described based almost exclusively on morphological characters. This is not possible, on the other hand, with other phyla, for example, vertebrates. These are often studied alive, with an opportunity to record behaviour, of which mating is particularly important, since it often makes it possible to define true biological species. These differences mean that we must interpret numbers of species in various phyla with an appreciation of the criteria on which the delineation of species in each was based.

Another problem is measuring species diversity through time. The fossil record affords no evidence of genealogy at any level up to and including biological species, and little basis even for distinguishing species based on anatomy, since only a certain set of characters will be preserved, with differences, again, between, say, arthropods and vertebrates. Mollusc shells, for instance, are well preserved, but vertebrates poorly in comparison. Necessarily, the distinctions made between fossil species is even cruder, with the added difficulty of delineating species in time in view of their continuous breeding and morphological changes. Thus, we find a rather morphological species concept being applied to fossils (the morphological changes through time necessitating the definition of separate species) which means fossil and living diversity are not easily compared. The criteria presently imposed on fossil species is that the specimens must be morphologically similar, must coincide in space and time (within the limits of accuracy afforded by the record), and show a Normal distribution in morphometric characters. Where dimorphic sexes are recognised, their numbers are expected to be approximately equal (Kemp 1999).

1.2 Difficulties in implementing biological species in taxonomy

Describing species is the art of taxonomy, and a crude art at that, for the time being, at least. Apart from the fact that no rules of consensus between molecular, behavioural, geographic and morphological data have been defined, taxonomists continue to struggle with the phenomena of ring and sibling species.

First of all, ring species are species whose spatially spread out populations are able to interbreed throughout the range, whilst individuals from some extremes of the range may be incompatible. In some cases, these extreme populations may expand to meet. This happens for instance around the Mediterranean Sea. The greater the geographic range of a species, the greater the risk of it being described twice. This, however, is not usually a problem once it comes to attention, although it may influence the estimate of the total of described species significantly. In the case of ring species, there seem to be good biological species, though, because they do not interbreed in some areas of their range, while they melt together entirely in other parts of the range, with individuals exhibiting intermediate phenotypes, and no two species being recognisable. This problem cannot be resolved with traditional taxonomy, although it seems useful to classify "species complexes" in these cases while resisting the desire to distinguish species.

While ring species may lead to more species being described than could actually be said to exist, sibling species are somehow the diametrically opposite phenomenon. This is where specimens are described, based on some characters (usually on morphology), as one species, but are subsequently found to differ in other characters, such as behaviour, not obvious in the specimens. This is a surprisingly widespread phenomenon, such that species numbers may be underestimated as a result, and some of the species found to be sibling species are similar to the point of identity in all characters except those relating to reproduction. Adis (1990) finds N. scandens to be comprised of two species, based on ecological, phenetic and molecular characters, with the notable absence of morphological differences.

But even if species concepts leave considerable ambiguity and taxonomy suffers the inherent problems described before, there is no denying that their application to the natural world gives useful pointers as to the loss of genetic diversity at (and following) the detriment of ecosystems everywhere.

2 Estimates of global diversity

Raven (1985) estimated that there may be between 3 and 5 million species, since for such well-characterised groups as mammals, birds and other macrofauna, there are roughly twice as many species in tropical as in temperate regions. The total of species described at the time was about 1.5 million, of which two thirds were found in temperate regions, with a majority of temperate insects. More species have been discovered since that time, which might tend to increase the estimate, but Raven did not consider the problem of species being given two names by two independent taxonomists, and so whilst there are now more than 1.8 million species described, an inclusion of doubly-described species would pull this estimate closer to 1.5 million. Raven went on to explain that if for every temperate insect, there were two in the tropics, there should be between 3 and 5 million species. However, his paper neglects bacteria, protozoa and helminthes of parasitic disposition, which tend to be studied only when they beset economically important host species. If every animal species has one parasite (Toft 1986), Raven's estimate should fall short by half of the real value. And it seems likely that some animals have more than one specific parasite (Toft 1986). The Acarina are one of a number of numerous, badly described taxa, whose inclusion may increase the estimate significantly (May 1988).

Within many taxa, it is true to say that there are consistently more species of smaller size than large species. Van Valen (1976) and May (1978) describe the data supporting this, and May (1978) also briefly reviews the theory, according to which

S @ L-y

with y being a value between 1.5 and 3 (May 1988).

Applying this size relationship to the data, extrapolating down to about 0.2mm (since the linear relationship between log body size and log species number will break up at very small sizes), we find there should be between 10 and 50 million species (May 1986, 1988).

Another stab at the problem was by Erwin and coworkers (Erwin & Scott 1980, Erwin 1982, 1983a, b), who sprayed insecticidal fogs into the canopies of the tree Luehea seemannii in Panama to collect beetles from that canopy, and finally extrapolated under application of several dubitable assumptions to an overall number of arthropods globally of 30 million. I will discuss his four assumptions and criticisms thereof next.

Of the species Erwin found, he classified 682 as herbivores, 296 as predators, 69 as fungivores and 96 as scavengers. He then estimated how many of each category might be specific to the tree species (assumption 1), concluding that 160 species should be specific. He then said that beetles pose 40% of the known arthropods (assumption 2), and hence there should be 400 arthropod species unique to the canopy of each tree species. As canopies, he further reasoned, are twice as rich in species as the forest floor (assumption 3), 600 species in total should be specific to each tree species. Based on an estimate of 50 000 species of tropical trees (assumption 4), there should be 30 million tropical arthropods.

First of all, I see a problem with Erwin's count in that there may be sibling species present that he and his coworkers might not have been able to distinguish on morphological grounds. Then there may well be other species with discrete polymorphisms or great sexual dimorphism, inflating his count. Then his various assumptions for the extrapolation can be criticised. I will now tackle the assumptions in turn, reviewing them in the light of more recent work:

Assumption 1: Erwin believes 20% of the herbivore, 5% of the predator, 10% of the fungivore and 5% of the scavenger beetles to be specific to the tree species. Some general criticisms can be made: Some of the species sampled from any tree by insecticidal fogging may be tourist species (Moran & Southwood 1982), meaning those that do not actually use the tree as a food source (this would inflate the estimate relative to the actual number of specialists). Another theoretical consideration important here is that due to the great tree diversity of tropical forests, individual trees of the same species are often a long way away. This should favour generalist species of herbivorous insects much more than in temperate woodlands. For lack of tropical evidence, May (1990) analyses data provided by Southwood to come to an estimate of 10% or less specific herbivore beetles on Luehea seemannii, halving Erwin's total for species. However, Erwin (1988) has also suggested that the level of specificity may be higher, based on recent work of his in Amazonia, leading him to a new estimate of 50 million species.

Assumption 2: Stork (1988), in analysing data obtained by Southwood et al. (1982) and himself, suggests that there may not be 40%, but 30% beetle species among the faunas of forest canopies, but the data also show a great variation among different regions as to the composition of the arthropod fauna, leaving much scope for further investigation. It is also regretful that no information is given on whether the 100% arthropods exclude Crustacea, which surely have no relevance for tropical tree canopies, but could significantly influence the estimates.

Assumption 3: Erwin suggests that the canopies of tropical trees are twice as rich in species as all other parts of the tree combined. Stork (1988) samples various parts of trees, and finds tenfold more individuals in the lower parts, including the topsoil, of the tree than the canopy. Ants in the canopy and springtails in the soil and leaf litter are likely to be few species with many individuals. Studies by Adis and coworkers (Adis & Albuqueque 1990, Adis & Schubert 1984) show the same pattern in Amazonia, albeit with many more individuals for each part of the tree. May (1990) concludes that the evidence suggests a ratio of canopy to ground species of at least 1, giving a larger estimate of total species number. May also suggests that mites may be as diverse as canopy beetles, leading to the suspicion that there might be even more species. However, one question is left unanswered in the papers I have reviewed, and that is, is it realistic to suppose the same level of specificity for soil species as for canopy species? I suspect that what Erwin holds as "ground" arthropods will be less specific than canopy beetles, since canopy species have to fight against secondary defences which will have evolved slightly differently in each tree, whilst sap-sucking species on the trunk are up against defences that will be similar among tree species. By the time the leaf litter reaches the mites, its content of secondary metabolites will be reduced and there will be no active response to predation, as in the canopy.

Assumption 4: May (1990) outlines two problems with scaling up by multiplying by the number of tree species: One species of tree may be associated with different sets of arthropods in different parts of its range. Some insect species may also be specially adapted to other tree species in other parts of their range. There are examples of each, and so there's a chance that they will cancel each other out.

Another piece of work worth mentioning is Hodgkinson & Casson's (1990) estimate based on the proportion of undescribed species in a sample taken in Sulawesi. Of 1700 Hemiptera on 500 species of tree, assumed to be an exhaustive list, only 37% were described. If insects generally were also known to the same degree of accuracy, the extrapolation would lead to 2.7 million species of insect. The authors have a second argument prepared. Extrapolating to 50 000 tree species, there should be 170 000 bugs in the tropics. Bugs pose 7.5% of described insects (Southwood 1978), leading to an estimate of 2.3 million insects. Once again, many assumptions are involved that could be called into question, in particular of how representative the site is of the tropics generally (which applies to most similar studies, including Erwin's and Stork's) and whether the Hemipteran fauna will be represented completely in one year's sampling.

It seems reasonably, then, to conclude based on the estimates here described, that the number of species on the planet may lie anywhere between 3 and 50 million. There are also several areas, or rather volumes, within the biosphere in which we may yet expect to discover many new species. One of these has already been covered in this essay: The tree tops, which retain some secrets as to their arthropod fauna. Then there are other inaccessible areas that humans have patchily explored, such as the deep sea, which appears to bear a surprising diversity of bacteria in particular, but also several taxa of large vertebrates recently discovered (Diamond 1985). Other extreme environments may be hot springs, saline lakes or the Arctic environments, for instance, which collectively may hold many new taxa of small organisms. Bacteria of great diversity are also found to a depth of 2.8km in the Earth's crust (Fredrickson & Onstott 1996). In the higher levels of the biosphere, such as the stratosphere, there may exist species of airborne microorganisms that no human individual has ever glanced evidence of, whose stochastic decline towards the Earth's surface, or alternatively, ascent into the skies, influenced by air movements such as the jetstreams, may be offset by their rate of reproduction. So besides inaccessibility of habitat by humans, there is also the criterion of inconspicuousness favouring species to remain undiscovered. This usually entails small size and drab patterning and colour (Diamond 1985). A limited geographic range and lack of migratory behaviour also contribute. Whilst we may look forward to such discoveries, we must also realise that this opportunity may be vanishing unexplored.

3 Why should we value the diversity of species?

At all times since the evolution of life, species - however defined - have gone extinct, while new species formed. Species are presently disappearing at a rate exceeding that at which new species come into being by several orders of magnitude. But why should this bother us? It is because human beings derive certain benefits from at least some of the species with which we share Earth.

Crudely, we can distinguish direct use values (consumptive use values and productive use values), nonconsumptive use values, option values and existence values (Primack 1998).

Direct use values derive from exploiting natural resources for fuel, vegetables, fruit, meat, medicine, cordage and building materials, fish and shellfish, skins, fibers, rattan, honey, beeswax, natural dyes, seaweed, animal fodder, natural perfumes, plant gums and resons. For instance, many thousands of plants are used for medicinal purposes in China and the Amazon basin alone. Besides those animals that appear in the gross national products of countries due to their export, many are consumed within the country in less developed regions such as equatorial Africa, among them a diversity of wild species and invertebrates. Timber products have a value of $75 billion per year, with other forest products contributing a sum of the same magnitude. Clearly, species richness influences this figure.

It is inevitable that humans will carry pests from one place to another, where they may run out of control, causing economic damage. Biological control by the introduction of natural predators of pests has often been the solution to such scenarios. The 20 most frequently prescribed pharmaceuticals in the US are worth $6 billion.

Like many of the direct use values, the nonconsumptive use values rely on functioning ecosystems. We know that often some species are especially important in ecosystems. These are the dominant and keystone species. Dominant species are often primary producers and large animals, and keystone species may fulfil any role within the system, usually occur in small numbers and are vital to the functioning of the ecosystem. Examples of such indirect uses are the protection that forests provide from floods and droughts, the way in which terrestrial and marine ecosystems globally affect the climate and ecotourism. Some ecosystems can be used for degradation of human wastes, and in others, species harvested by humans depend on other species. Some species are indicator species whose disappearance can give an early warning of the slow demise of the environment. Finally, bordering on existence values, species are useful for education and science, though arguably, if the species went, both of these areas would be obsolete.

More interesting to me, though, seems the possibility of new products for consumption, such as microbes (e.g. quorn) and seaweed, some of which may prove to be more energy-efficient in production. Similarly, many tree species look destined for liquid fuel production. It can be hoped that many plants will turn out to possess extraordinary medicinal properties, such as the rosy periwinkle (Catharanthus roseus). These hopes must be summarised as option values, which are the values of potential uses of species that will be lost by the loss of those species before beginning exploitation. It is difficult to predict what options will become open in the future, arising out of the depths of molecular research, for instance, and so I will not attempt this any further here, except to say that I believe we are only beginning to use our natural environment for our benefit.

Existence values are those hardest to put a dollar label on. The pleasure that human beings derive from knowing that we share the planet with certain fascinating creatures is what defines existence values. I would include in that the value of knowing that we lost only a small proportion of all species in our destructive exploitation of the planet, rather than worse. This we have yet to put into practice.

Finally, there are certain ethical questions that we have to face if we assume the right to extinguish other forms of life, if accidentally. As we are becoming increasingly aware that few things distinguish us from other species, it becomes harder to argue that we do have a special place on Earth that allows us to trash the place rather than assume good stewardship.

The main justification, I believe, for diverting energy to the studies outlined in section 2 and those proposed in the references, is that a number is required so that we can estimate the magnitude of the threat to biodiversity. The global total is useful for getting a feel of how widespread most species are, and knowledge of how many species are lost in a period of time can be used to raise public awareness and concern about the loss of biodiversity. In utilitarian terms, the number of species is also an indicator of the magnitude of biodiversity, and an indicator of the economic values we may yet draw from it.

However, I also wonder whether having established a number of species in the millions, and given that no matter how many studies of beetle diversity we undertake, we are still only extrapolating figures, it is morally justifiable to pour more research money into studies of species diversity rather than devoting such money to the cause of saving our biosphere.


Adis, J. 1990. Thirty million arthropod species - too many or too few? Journal of Tropical Ecology 6, 115-8.

Adis, J. & Albuqueque, M. O. 1990. Impact of deforestation on soil invertebrates from Central Amazonian inundation forests and their survival strategies to long-term flooding. Water Quality Bulletin 14, 88-98.

Adis, J. & Schubert, H. O. R. 1984. Ecological research on arthropods in Central Amazonian forest ecosystems with recommendations for study procedures. In: Cooley, J. H. & Golley, F. B. (eds.), Trends in ecological research in the 1980s, pp. 111-44. Plenum Press, New York.

Diamond, J. M. 1985. How many unknown species are yet to be discovered? Nature 315, 538-9.

Dobzhansky, Th. 1935. A critique of the species concept in biology. Philosophy of Science 2, 344-55.

Erwin, T. L. 1982. Tropical forests: their richness in Coleoptera and other arthropod species. Coleopterists' Bulletin 36, 74-82.

Erwin, T. L. 1983a. Tropical forest canopies: the last biotic frontier. Bulletin of the Entomological Society of America 29, 14-9.

Erwin, T. L. 1983b. Beetles and other insects of tropical forest canopies at Manaus, Brazil, sampled by insecticidal fogging. In: Sutton, S. L., Whitmore, T. C. & Chadwick, A. C. (eds.), Tropical rain forest: ecology and management, pp. 59-74. Blackwell, Oxford.

Erwin, T. L. 1988. The tropical forest canopy: the heart of biotic diversity. In: Wilson, E. O. (ed.), Biodiversity, pp.123-9. National Academy Press, Washington, D. C.

Erwin, T. L. & Scott, J. C. 1980. Seasonal and size patterns, trophic structure and richness of Coleoptera in the tropical arboreal ecosystem: the fauna of the tree Luehea seemannii in the Canal Zone of Panama. Coleopterists' Bulletin 34, 305-35.

Fredrickson, J. K. & Onstott, T. C. 1996. Microbes deep inside Earth. Scientific American 275 (4), 42-7. October 1996.

Hodgkinson, I. D. & Casson, D. 1990. A lesser predilection for bugs: Hemiptera (Insecta) diversity in tropical rain forests. Biological Journal of the Linnean Society 43, 101-9.

Kemp, T. S. 1999. Fossils and evolution. Oxford UP, Oxford.

Kornet, D. J. 1993. Permanent splits as speciation events: a formal reconstruction of the internodal species concept. Journal of theoretical biology 164, 407-35.

Madigan, M. T., Martinko, J. M. & Parker, J. 2000. Biology of microorganisms. 9th ed. Prentice Hall, Upper Saddle River, New Jersey.

May, R. M. 1978. The dynamics and diversity of insect faunas. In: Mound, L. A. & Waloff, N. (eds.), Diversity of insect faunas, pp. 188-204. Blackwell, Oxford.

May, R. M. 1986. How many species are there? Nature 324, 514-5.

May, R. M. 1988. How many species are there on Earth? Science 241, 1441-9.

May, R. M. 1990. How many species? Philosophical Transactions of the Royal Society of London, Series B 330, 293-304.

Mayr, E. 1942. Systematics and the origin of species. Columbia UP, New York.

Moran, V. C. & Southwood, T. R. E. 1982. The guild composition of arthropod communities in trees. Journal of Animal Ecology 51, 289-306.

Paterson, H. E. H. 1985. The recognition concept of species. In: Vrba, E. S. (ed.), Species and Speciation, pp. 21-9. Transvaal Museum Monograph No. 4, Pretoria, South Africa.

Primack, R. B. 1998. Essentials of conservation biology. 2nd ed. Sinauer Associates, Sunderland, Massachusetts.

de Queiroz, K. & Donoghue, M. J. 1990. Phylogenetic systematics or Nelson's version of cladistics? Cladistics 6, 61-75.

Raven, P. H. 1985. Disappearing species: a global tragedy. The Futurist 19, 8-14.

Ridley, M. 1996. Evolution. 2nd ed. Blackwell Science, Cambridge, Massachusetts.

Southwood, T. R. E. 1978. The components of diversity. In: Mound, L. A. & Waloff, N. (eds.), Diversity of insect faunas, pp. 19-40. Blackwell, Oxford.

Stork, N. E. 1988. Insect diversity: facts, fiction and speculation. Biological Journal of the Linnean Society 35, 321-37.

Templeton, A. R. 1989. The meaning of species and speciation: a genetic perspective. In: Otte, D. & Endler, J. A. (eds.), Speciation and its consequences, pp. 3-27. Sinauer Associates, Sunderland, Massachusetts.

Toft, C. A. 1986. In: Diamond, J. M. & Case, T. J. (eds.), Community ecology, pp. 445-63. Harper & Row, New York.

van Valen, L. 1976. Ecological species, multispecies, and oaks. Taxon 25, 233-9.