Current issues in species delimitation

In practice, species are defined taxonomically based on phenetic characters, and delimitation tends to be uncontroversial for species that have diverged substantially from their closest relatives.¹ We can thus consider the tacit criterion for recognizing species taxonomically to be diagnosability.

However, in some situations where two groups have diverged very little, or where they are genetically but not phenotypically distinct, a more objective, less ad hoc criterion — a species concept — may be required. This is also necessary for management efforts to conserve not only species richness but also genetic diversity. Several defining characteristics of species have been proposed (e.g. possibility of interbreeding or forming phenotypic clusters), though none to universal satisfaction.

The increasing availability of molecular data, and particularly of direct DNA sequences, has brought the hope of resolving the disagreements between species concepts, as the extent of molecular divergence is thought to represent the extent of evolutionary divergence between two taxa (which is proportional to the time since speciation if a perfectly neutral mode of evolution holds). As soon as two populations have become fixed for different alleles, they can be distinguished as being descendants of different lineages, a situation referred to as lineage sorting. Coalescent theory can be used to determine the time since the most recent common ancestor of all the gene copies present in each of those populations (which may or may not postdate the separation of the two populations).²

However, it has been observed that there are significant differences in the rates of mutation and allele fixation between different genomes (organellar and nuclear) within an individual, as well as between different loci in the same genome. This can lead to situations of incomplete lineage sorting, where two populations may have become fixed for different alleles of one gene, but not for another. This will lead to discordance between the different gene trees, as one will show reciprocal monophyly between the two populations, while the other will show a polyphyletic or paraphyletic relationship of one population with respect to the other. This makes it impossible to unequivocally assign populations that were only recently separated to different species.

Baum & Shaw 1995³ address this issue by defining species as “basal, exclusive groups of organisms,” thus ensuring that species show divergent relationships amongst each other, but reticulate relationships within each species (due to interbreeding of sexual organisms). This approach provides an objective basis for delimiting species, as it rests on the criterion of genetic concordance: If the time to coalescence of a population of genes is shorter than the coalescence time of genes in that and any other population, then those populations will be exclusive and therefore divergent. Given sufficient time since separation, this situation will obtain for all the gene loci in the organism.

In practice, this can be evaluated by constructing a consensus of several different gene trees: (Monophyletic) clades supported in all gene trees are genealogical species; individuals found not to belong to basal, exclusive groups form “metaspecies”.³ This counterintuitive concept of metaspecies reveals the limited utility of the genealogical species definition, as many organisms — possibly the majority of organisms — could turn out not to belong to a species at all. This method, depending as it does on comparing a number of gene trees, is sensitive to the number of genes sampled; a particular species definition is likely to be invalidated by increased sampling. Also, any speciation event will cause the parent population to lose its species status under the criterion of exclusivity. In fact, a single occurrence of parthenogenesis (or even a case where two siblings happen to have the same alleles for all gene loci examined) will form a new exclusive group and cause the rest of the species to be demoted to metaspecies. Finally, although Baum & Shaw claim to be focusing on the gene rather than organismal level, this species concept cannot be applied to asexual organisms any more than the Biological Species Concept, and thus seems like an undesirable way to define species, despite making use of nucleotide data and coalescence theory, which apply universally.

Kevin de Queiroz’s “unified” species concept⁴ is a more flexible way of approaching the species problem that is more sound both theoretically, as it applies to all organisms and is methodology-independent, and practically, as it can accept different types of evidence, depending on what is obtainable or informative for a given organism. It resolves the conflict between previously proposed species concepts by noting that they all agree on the fundamental conception of species as “separately evolving metapopulation lineages,” but by reconsidering the secondary criteria as different possible lines of evidence, which together provide support for a species hypothesis, though no single one is mandatory.⁴ Although this cannot discriminate between diverging populations where lineage sorting is not yet complete as they are not yet evolutionarily separate, the multiple-lines-of-evidence approach can identify situations of ongoing speciation if some support can be found for differentiation, even if it is currently contradicted (not yet supported) by other types of evidence.

For the purposes of conservation and management, however, the above method may be too burdensome in terms of investment of time, resources, and expertise. Clearer but less strict criteria have been proposed for identifying units of genetic diversity to conserve, such as that Evolutionarily Significant Units should have diverged in their mitochondrial DNA, but need not be distinct in their nuclear DNA.⁵ Although easier to investigate, the expectation that all groups of individuals have reciprocally monophyletic mtDNA sequences at a useful species-level may be unrealistic, as well as open to the accusation of arbitrariness.

Jeff Doyle has proposed a framework for defining a gene pool, which is biologically meaningful as it defines both the gene population within which recombination (i.e. interbreeding) takes place and within which the genetic potential of individuals is constrained.⁶ It is the level of hierarchical organization just on the divergent side of the divergence–reticulation boundary, as sought by Baum & Shaw 1995;³ it is unified by gene flow, but there is no gene flow between gene pools (by definition). Allele frequencies are examined, as opposed to genealogies, as this eliminates the problem of incomplete lineage sorting leading to poorly resolved or contradictory gene trees. This has the added advantage that any locus can be used for tracking the histories of organisms (involving processes like migration and selection), while being insensitive to molecular-level processes like mutation and recombination.⁶ Allele pools are those alleles which cooccur in heterozygotes, and gene pools are groups of alleles which are found across individuals sharing allele pools. By using recombination to define breeding groups while ignoring the confusion it causes in gene trees, this method can usefully be applied to identify groups with different genetic potentials. Even if these groups will usually be smaller (and at most the same size) as taxonomically-defined species, this is the level of concern for the preservation of genetic diversity, in addition to involving a simpler and less contentious method of delimitation, and thus seems commendable for use in conservation.