1. Introduction

Recently in this journal Coleman and Wiley (Reference Coleman and Wiley2001) presented a new defense of the species-are-individuals thesis (hereafter SAI-thesis). In the debate over the SAI-thesis the issue is usually construed as a straightforward dilemma: either species have the ontological status of individuals, or of classes, without a third option (e.g., Ruse Reference Ruse1987). Although both sides in the debate are still represented in the literature, most biologists and philosophers nowadays hold the view that species are individuals with organisms as their parts, rather than classes having organisms as members. Those authors who find convincing arguments on both sides of the dilemma usually subscribe to an ontology of species as individuals and attempt to construct an epistemological account of how such individuals can function as kinds (e.g., Griffiths Reference Griffiths1999; Millikan Reference Millikan1999). Coleman and Wiley defend the SAI-thesis by attempting to disprove the view that species are classes.

Construing the issue as a yes-or-no dilemma is too simple in two respects. Firstly, there is the assumption (which is by no means self-evident) that the issue has but two horns. Traditionally, species have been understood as all belonging to the same category of things and the species problem has been interpreted as the quest for the one true definition of the associated concept. In the last decades several authors have abandoned this monistic stance in favor of so-called “species pluralism.” As with most “-isms,” “species pluralism” denotes a heterogeneous collection of positions, which however share the view that there are different sorts of species taxa. This view allows a third option: the SAI-thesis is both true and false, depending on the situation considered. Given that there are species taxa of different sorts, not all of them need to be individuals (or classes for that matter). From this perspective the term “species” is seen to be a homonym, denoting different kinds of things associated with different concepts in different domains of biological research. If this perspective is correct, it follows that the SAI-thesis should be evaluated independently for each distinct concept; findings for one concept need not automatically apply to the other concepts at stake.

Secondly, choosing one of the horns of the original dilemma does not resolve the problem of the ontological status of species. For if species are indeed always individuals, then what kind of individuals are they and are they always of the same kind? There are several feasible options, e.g., interactors (dynamic processes entities) and lineages (branches on phylogenetic trees). And if species are not individuals, then what are they? Again there are several options, e.g., collections of organisms (Ruse Reference Ruse1987; Mahner and Bunge Reference Mahner and Bunge1997) or groups of populations (as in Mayr's Biological Species Concept; cf. Boyd Reference Boyd1989). On this level as well the issue has more than two horns and more than one may simultaneously be right.

The homonymic nature of the term “species” manifests itself in at least two ways. Firstly, “species” is used by biologists to denote both the dynamic units of evolution and the static units of biodiversity (Reference LidénLidén and Oxelman 1989; Ereshefsky Reference Ereshefsky1991, Reference Ereshefsky1992; Mahner and Bunge Reference Mahner and Bunge1997). The architects of the Modern Synthesis emphasized the dynamic nature of species: “the taxonomic categories in general, and species in particular, are not static but dynamic units” (Dobzhansky Reference Dobzhansky1935); and:

From this dynamic perspective species are composite entities that actively participate in evolutionary processes and interact as cohesive wholes with their environment and the entities therein, similarly to soccer teams participating in matches and tournaments (“interactors”: Hull Reference Hull1980). Recently Gould (Reference Gould2002, 703ff.) extensively discussed and defended this perspective on species.

By contrast, units of biodiversity are static entities resulting from rather than participating in evolutionary processes: “Taxa (including species) are … atemporal components of a historical pattern. They are the products of evolution, not its determinants.” (Reference LidénLidén and Oxelman 1989); and: “species are generally considered to be products of evolution (lineages), but not the units participating in processes.” (Kluge Reference Kluge1990). Understood in this manner species constitute the building blocks of the “tree of life” that phylogenetic systematics investigates. “Unit of evolution” and “unit of biodiversity” in the sense above are both individual-concepts, i.e., concepts associated with entities rather than classes. The fundamental difference between these two types of entities becomes clear when considering the allocation of organisms from the past: whereas dead organisms no longer constitute parts of any coherent entity partaking in evolutionary processes, they still belong to some branch on the tree of life.

Secondly, biological science consists of several distinct (sometimes partly overlapping) research contexts like ecology, ethology, and phylogenetic systematics, all employing the concept of species in investigating their own particular kinds of questions. Since the demands posed on this concept generally are incompatible between different research contexts, no single concept will suit all contexts of biological research (Kitcher Reference Kitcher1984; Kornet Reference Kornet1993; Shaw Reference Shaw, Howard and Berlocher1998). The term “species” is thus best understood as the common denominator for a number of distinct scientific concepts, each tailored to the idiosyncratic demands of a particular research context. In some cases these concepts will be individual-concepts, in others class-concepts. Many discussions of the ontology of species consider only one context, resulting in divergent conclusions: Ruse (Reference Ruse1987) analyzed evolutionary biology and concluded that species are kinds, while Lidén and Oxelman (Reference Lidén1989) considered phylogenetic systematics and decided that species are individuals. Both conclusions may well be valid within the considered research contexts, but their validity does not automatically extend beyond the boundaries of these contexts. Kitcher (Reference Kitcher1984), Falk (Reference Falk1988), and Dupré (Reference Dupré1993) have made similar points.

2. Criticisms of Coleman and Wiley's Defense

Coleman and Wiley defend the SAI-thesis by analyzing biologists’ use of Latin species-denoting binomials. They start from the straightforward dichotomy discussed above: in biological discourse binomials can be understood to function as singular terms, i.e., terms denoting entities rather than classes, or as categorical terms to be analyzed as predicate expressions (Reference Coleman and Wiley2001, 500). Whereas the former entails an ontology of species as individuals (an “objectual account” of species in Coleman and Wiley's terminology), the latter entails an ontology of species as classes whose names are interchangeable with predicate expressions that reflect the necessary and sufficient conditions for species membership (a “predicative account” of species). My focus is now on the two main steps in Coleman and Wiley's defense. First (Section 2) they present some examples from biological practice in which binomials are used as singular terms. These examples serve to show that an objectual account of species fits some cases, but cannot rule out the possibility that other cases require a predicative account of species. In their Section 3 they attempt to rule out this possibility by refuting the so-called eliminability thesis, which says that binomials in biological discourse are “substitutes for complex predicate expressions that do function as categorical terms” (Reference Coleman and Wiley2001, 506). Lacking a third option, a refutation of the eliminability thesis would directly entail that only an objectual account of species is adequate to biological practice.

Although I consider their first example, “Fundulus nottii is a species” (Reference Coleman and Wiley2001, 503), somewhat unfortunately chosen,Footnote ¹ I do not deny that cases in which binomials denote objects rather than classes are abundant in biology. Clear examples can be drawn from e.g., ecology or phylogenetic systematics. What I do dispute is that biology consists exclusively of such cases. Biology also includes many cases in which binomials are used as class terms supporting generalized knowledge statements:

Inferences from the observed properties of organisms to their unobserved properties (Mayr Reference Mayr1961, Reference Mayr1969) or to properties of other organisms belonging to the same species (Millikan Reference Millikan1999) must be ontologically founded, i.e., there must be an ontological guarantee that the organisms of one species are the same in relevant respects (Griffiths Reference Griffiths1999; Millikan Reference Millikan1999). Whereas ontologies of species as classes automatically provide such a guarantee (due to the intensionality of class definitions—see below), ontologies of species as individuals with organisms as their parts do not, since an entity's parts need not be the same in any respect. Species-as-individuals ontologies must therefore explicitly state how such inferences are supported. No such accounts have been provided so far; moreover, ontologies of species as lineages are in principle unable to provide them.Footnote ²

By refuting the eliminability thesis Coleman and Wiley purport to establish that predicative accounts of species are unsuitable to scientific practice. The eliminability thesis however allows for a strong and a weak reading. In its strong reading the thesis entails that Latin binomials can be substituted by predicate expressions in all scientific statements in which they occur. In its weak reading it only implies that this can be done at least in some statements. Since they hold that species are always individuals, Coleman and Wiley must refute the eliminability thesis in its weak reading.

Coleman and Wiley attack two ways of conceptualizing binomials as disguised categorical terms: by understanding them as placeholders for predicate descriptions of organism morphology, or as placeholders for specific instances of general properties thought to define the species category (e.g., potential interbreeding or common descent). I do not discuss the latter point. Against the former conceptualization Coleman and Wiley hold that sets of organism characteristics necessary and sufficient for species membership are virtually impossible to obtain. This is however only the case if some other criterion for allocating organisms to species is already at work, i.e., if we attempt to reconcile organism morphology with a classification based on another preconception regarding species membership. Coleman and Wiley's remarks (Reference Coleman and Wiley2001, 507) on the suitability of characters for diagnosing but not for defining species taxa suggest that they argue from just such a biased perspective.Footnote ³ Organism morphology as criterion for species membership is bound to conflict with this phylogenetic perspective. Once this perspective is abandoned, however, the conflict disappears for lack of a second party to conflict with. Although morphology-based classes of organisms generally will be nonreal, nominalistic classes and will show little overlap with the species used in phylogenetic systematics, this does not prevent them from being practically useful and theoretically significant classes that can support generalized knowledge statements in other contexts of biological research. Consider for example fields that address “how”-questions rather than “why”-questions (Mayr Reference Mayr1961) by studying organisms with respect to structural similarities and classifying them according to analogies rather than homologies (e.g., functional morphology and ecology; see the discussion in Griffiths Reference Griffiths1999). Since not all of biological investigation rests on assuming a phylogenetic perspective, arguments against the eliminability thesis based on phylogenetic considerations cannot serve to refute the thesis for the whole of biological science.

Another problem in this part of Coleman and Wiley's analysis is that it rests on the traditional essentialistic conception of natural kinds as classes of entities that all share a set of (usually microstructural) properties necessary and sufficient for class membership (Reference Coleman and Wiley2001, 502), an account that for many reasons fails with respect to species. This renders their analysis incapable of addressing alternative accounts of species as natural kinds, such as Boyd's (Reference Boyd1989, Reference Boyd1999) account in which natural kinds are defined by the natural mechanisms underlying the repetitive co-occurrence of particular properties, rather than by these properties themselves. Hence, even if the eliminability thesis were refuted in its weak reading, this does not imply the rejection of such alternative ontologies of species as natural kinds.

3. Species as Sets

In their Section 4 Coleman and Wiley criticize the conceptualization of species as sets. Although many of their arguments seem to indicate real problems for this understanding of species, Coleman and Wiley are misguided in thinking that their attack supports the SAI-thesis. The problem lies in the formal difference between sets, which are defined extensionally, and classes, which are defined intensionally (Mahner and Bunge Reference Mahner and Bunge1997; Muller Reference Muller2001). The debate on the SAI-thesis hinges on the issue whether species are classes, not whether they are sets (but see Kitcher Reference Kitcher1987). Both species as individuals and species as classes can be reconstructed as sets, implying that arguments for or against the view of species as sets cannot help to establish the ontological status of species as individuals or classes. Such reconstructions take place by way of associating an abstract set of organisms with a particular species taxon, which is different from identifying the taxon with the set. The reconstruction of a species as a set thus does not imply the ontological statement that this taxon is a set. Kornet (Reference Kornet1993) performed this reconstruction on the prototypical example of species as individuals: Hennig's phylogenetic species defined as chunks of the genealogical network. Coleman and Wiley's argument, saying that on the interpretation of species as (extensionally defined) sets of organisms a species ceases to exist when its extension changes (Reference Coleman and Wiley2001, 511), does not hold for (intensionally defined) classes: whereas a different set becomes associated with a species when its composition changes, the species itself remains in existence. Since the reconstruction of species as sets is compatible with the interpretation of species as individuals, judgments regarding the validity of the construction of species as sets do not reflect on the validity of the SAI-thesis.

4. Conclusion

Coleman and Wiley conclude that “biological discourse contains an ineliminable reference to individual things called ‘species’” (Reference Coleman and Wiley2001, 516). Taken literally, as a denial of the strong reading of the eliminability thesis, this statement is correct: in many cases Latin binomials indeed refer to individuals. However, this does not amount to saying that in biological science species are always to be conceptualized as individuals, i.e., it does not amount to the validity of the SAI-thesis for the whole of biology.

I believe the SAI-thesis is valid for some but certainly not all contexts of biological research. In their defense of the SAI-thesis Coleman and Wiley fail to take into consideration the heterogeneity of the species category. The fact that the term ‘species’ denotes different concepts in different contexts of biological research implies that in some contexts species might best be understood as individuals, whereas in others their proper ontology is that of classes or even natural kinds.