Recently in TREE, Mallet [1,2], argued for an operational, concept-free definition of species as ‘genotypic clusters’, asserting ‘that species are man-made groupings’ . However, Mallet resorts to the traditional notion of ‘good’ species for final arbitration regarding what degree of variation is appropriate for the species-level taxon.
This is a poor species definition for two reasons. As an operational definition it leaves us with no means for dealing with the great complexities of biological systems of descent. Moreover, the decision as to what constitutes species-level variation is based on an essentialistic perspective offered by a ‘good taxonomist’s or naturalist’s definition’ .
If species are, in Mallet’s operational terms, ‘groups that remain recognizable in sympatry because of the morphological gaps between them’ , it is important to realize that they are nothing more than the sum of the operations that serve to identify them [3,4]. We are left with the arbitrary decision of how large these gaps must be and what frequency of intermediates would lead us to accept two species rather than one. These difficulties are problematic for the diagnosis of a ‘species’ under any definition or concept, due to the fuzzy nature of groups resulting from, or participating in, the evolutionary process (i.e. natural groups).
However, they are more severe for a concept-free definition because we have no theoretical guideline with which to sort variation into hypotheses about natural groups. A conceptfree definition of species as ‘genotypic clusters’ must also deal with the discrete morphological variation manifest, for example, between genders of many plants and animals. These are ‘genotypic clusters’ of sorts. However, neither today, nor in Darwin’s time, do biologists wittingly hypothesize different species for different genders. Without an ontological context with which to sort variation in biological systems, we find ourselves perplexed by situations as straightforward as sexual dimorphism.
In stark contrast to his purely operational definition, Mallet alludes to such an ontological framework by reference to ‘good’ species . But what are ‘good’ species? Mallet endorses the traditional position that, in the most difficult cases, the ultimate authority of the existence of ‘good’ species is the taxonomist or naturalist.
The implication of this deference to the taxonomist is that ‘good’ species exist, but that their essential nature is hidden; a taxonomist’s contribution is to reveal ‘good’ species through description, case by case. Thus, the notion that ‘good’ species can be revealed to us by taxonomic authorities is steeped in the essentialistic outlook that Mallet [1,2] (and others ) seek to condemn. Furthermore, the definition of a species becomes ‘a group of organisms that is recognized as a ‘good’ species by the taxonomist or naturalist.’
This is obviously undesirable. Although taxonomists may point to groups that they believe exist, species will only have objective value if the general properties of ‘good’ species (the species taxon) are revealed to the rest of us. Whereas a purely operational definition causes us to forego the question, ‘what is the nature of the group that we might call species?’, asserting the existence of ‘good’ species (even if we knew their properties) demands that all groups of organisms, that we might call species, exist in the same ways. A definition that results in one or both of these outcomes should be avoided, particularly in studies of speciation where we are interested in all the natural groups produced by a pluralistic process of evolution. It is a step forward for students of speciation to acknowledge that different sorts of natural groups have valid claims to the term ‘species’ [6,7].
Similarly, it is regressive to undermine the notion that the species taxon (whatever natural group we choose for it to designate) has underlying properties that make it worth studying. One possible solution is provided by a nominalistic approach [6-9], which formulates a species definition explicitly while retaining the ontological meaning that a purely operational definition leaves behind. Such a definition would embody a statement of the necessary and sufficient properties for the diagnosis of species in any particular case. The important distinction between a nominalistic definition and Mallet’s is that our avenue of inquiry would lead us to explore the nature and evolution of natural groups (as opposed to some notion of a ‘good’ species), with or without a coextensive relationship between such groups and ‘good’ species (whatever they are!).
Kerry L. Shaw
Dept of Organismic and Evolutionary Biology, Harvard University, 16 Divinity Avenue, Cambridge, MA 02138, USA
 Mallet, J. (1995) Trends Ecol. Evol. 10,294-299
 Mallet, J. (1995) Trends Ecol. Evol. 10, 490-491
 Hull, D.L. (1968) Syst. Zoo/. 17,438-457
 Baum, D.A. and Donoghue, M.J. (1995) Syst. Bot. 20, 560-573
 Mayr, E. (1982) The Growth of Biological Thought, Belknap
 de Queiroz, K. and Donoghue, M.J. (1990) Cladistics 4, 317-338
 de Queiroz, K. (1994) Syst. Biol. 43, 497-510
 Popper, K.R. (1966) The Open Society and its Enemies (5th edn), Princeton University Press
 Baum, D.A. and Shaw, K.L. (1995) in Experimental and Molecular Approaches to Plant Biosystematics
(Hoch, PC. and Stephenson, A.G., eds), pp. 289-303, Missouri Botanical Garden