Hierarchy and monophyly.

Abstract

In a series of recent papers, Pleijel (1999) and Pleijel and Rouse (2000a,b, 2003) have proposed that systematists should disregard species, advocating a nomenclatural system in which the smallest recoverable clades (termed least-inclusive taxonomic units, or LITUs) are recognized as basic taxonomic entities. Among the arguments presented in support of their proposal, Pleijel and Rouse contend that most contemporary species definitions are inappropriate in systematics as they permit species to be non-monophyletic. As an example, they cite the finding of Talbot and Shields (1996) that Ursus maritimus has a sister-group relationship to some insular demes of U. arctos. Pleijel and Rouse (2000a; p. 628) regard arctos as paraphyletic (since it does not include all descendants of an ancestral deme1) and argue that in recognizing arctos and maritimus as separate (exclusive) taxa, systematists unacceptably neglect historical information. The issue of monophyly has figured prominently in discussion on the species category, with several earlier authors (e.g., Rosen, 1979; Mishler and Brandon, 1987; de Queiroz and Donoghue, 1988) having suggested that monophyly could (or should) be employed exclusively as a criterion for recognizing taxa. These authors often have appealed to the possibility of non-monophyletic species as an argument against the use of criteria such as reproductive isolation, diagnosability, etc. in species definitions (however, see de Queiroz and Donoghue, 1988 for an exception). Here, I argue that the opposition to non-monophyletic species is unfounded, stemming from a lack of consideration of the hierarchical organization of genealogical entities and their historical relations. As Grene (1969) noted, the term hierarchy has been used to denote several related but distinct concepts. For the purpose of this note, I regard as hierarchical entities or systems that may be considered to consist of a series of compositional levels, that is, “divisions of stuff … organized by part-whole relations, in which wholes at one level function as parts at the next (and at all higher) levels” (Wimsatt, 1994; p. 222). An example of such a series of levels is provided by the familiar somatic hierarchy of cells, tissues, organs, and organisms; entities occupying levels earlier in this series (i.e., particular cells, tissues, etc.) constitute parts of entities occupying later levels. This use of the term hierarchy corresponds with that of “constitutive hierarchy” by Mayr (1982), “scalar hierarchy” by Salthe (1985), and “incorporative hierarchy” by Ghiselin (1997). There have been several hierarchies expounded in the biological literature (for a critical review, see Salthe, 1985; pp. 175–180), differing according to authors' interests. Of concern here is the genealogical hierarchy of genes, cells, organisms, demes, and species (Eldredge and Salthe, 1984; Vrba and Eldredge, 1984; Salthe, 1985). The part-whole relations of entities at lower levels in this hierarchy, specifically those of genes, cells, and organisms, are well known; it is axiomatic that, in general, organisms have cells as parts and cells have genes as parts. The part-whole relations of demes and species, however, are more obscure. Frost and Kluge (1994) have argued cogently that species should be equated with largest population lineages, integrated by interbreeding and isolated reproductively (whether extrinsically or intrinsically) from other lineages. According to this view, species are “composed of reticulating sublineages [i.e., demes] and toko-genetically related organisms” (Frost and Kluge, 1994; p. 275) unified by interbreeding, and thus occupy a compositional level above demic and organismal levels. The part-whole relations of organisms, demes, and species conceived of are illustrated (fortuitously) by Simpson's (1955) “diagram of phylogenetic patterns” (his fig. 48). At each level in the genealogical hierarchy entities share historical (genealogical) relations as a result of reproductive events; replication or duplication, tokogenesis, and speciation (i.e., cladogenesis), for example, produce historical relations among genes, organisms, and species, respectively. As a consequence of the part-whole relations obtaining across levels, reproductive events at particular levels will necessarily produce entities that are non-monophyletic assemblages at lower levels. This is because the descendants of an ancestral entity at a lower level can not be part of both an ancestral entity and its descendants at a higher level; as Rieppel (1994; p. 94) put it, “ancestors are doomed to paraphyly”. Thus, we expect that genealogical entities will be non-monopyletic in most instances. This applies not only to species but to entities at all levels in the genealogical hierarchy. Lidén (1990; p. 183) noted that “Even if monophyly is a universal concept, it is in a particular case meaningful only in relation to the conceptual singularities [i.e., entities] in the model [i.e., system] concerned”. As hypotheses of descent, statements of monophyly necessarily implicate reproduction; the notion that a group of somatic cells is descended from an common ancestor is meaningful only insofar as it may be related to some event by which an ancestral cell gives rise to descendants, in this case mitosis or schizogony. Since reproductive events at particular levels in the genealogical hierarchy produce patterns of historical relationship incongruent with those at lower levels, reproduction at different levels must be considered independently, so that particular statements of monophyly can apply only to entities at a single level. If our concern is the historical relationships of somatic cells, any consideration of the relationships of organisms, for example, is extraneous, since tokogenesis does not necessarily produce monophyletic groups of cells.2 Similarly, if our concern is the historical relationships of species, any consideration of the relationships of demes is extraneous, as speciation does not necessarily produce monophyletic groups of demes. Accordingly, that the demes composing a species may constitute a non-monophyletic assemblage is irrelevant if our goal is to develop a system based on the descent of species. Arguments such as Pleijel and Rouse's are problematic as there is no logical basis for granting primacy to the level of demes in describing the historical relations of genealogical entities. This is evident if we consider that some of the organisms composing a deme may be more closely related to organisms of other demes than to one another. Applying Pleijel and Rouse's reasoning, such demes are non-monophyletic and thus, presumably, subject to the same criticisms as non-monophyletic species. Perhaps, then, systematists should concern themselves with the discovery of monophyletic groups of organisms rather than demes. Alternatively, they might focus on discovering monophyletic groups of cells, since many organisms are non-monophyletic at the cellular level (see, for example, Wiley and Mayden, 2000). The point is that entities at all levels in the genealogical hierarchy compose monophyletic groups. Moreover, as reproductive events at particular levels necessarily produce entities that are non-monophyletic assemblages at lower levels, patterns of historical relationship at different levels will not coincide. Consequently, a single description of the historical relations of genealogical entities, encompassing all levels (or even two levels), is unattainable; depictions of historical relationships, and hence concerns for monophyly, must be restricted to a single level in any particular instance. This note is modified from part of a thesis submitted in fulfillment of a Master of Science degree at The University of Adelaide. The ideas presented derive from those of Frost and Kluge (1994), and a debt to their writing is acknowledged. I thank Mike Lee, Greg Rouse, Steve Donnellan, Mark Hutchinson, and an anonymous reviewer for helpful comments and discussion.