Species Delimitation: Discrete Names in a Continuous World with Fuzzy Boundaries

Abstract

The introductory quotation by Aristotle is not only a nice encapsulation of the idea of a great chain of being or , but it also shows that, from the birth of Western science and philosophy, it has always been obvious that nature has fuzzy boundaries and that demarcation of similar entities is a difficult task. In Sect. 5.2 a potential theoretical solution to the species problem has been introduced: a hierarchical view that distinguishes between a true ontological or primary species concept and secondary operational species criteria. The ontological species concept (telling us what species taxa really are) is based on conceiving of species as population-level lineages or segments of such lineages as defined by either the Evolutionary Species Concept or the General Lineage/Unified Species Concept which are very similar. However, these concepts are all non-operational, they just provide the framework within which species can and should be identified. Identification and delimitation of species taxa is what taxonomists are doing, and it is important that they are doing it within a consistent theoretical framework. The practice of species delimitation, however, is not guided by the lineage framework beyond the condition that species must be lineages. Since the Tree of Life (and even more so the Web of Life in the prokaryotic world) is made up of lineages within lineages, the question whether there is an objective level of the species crucially depends on whether there is some (population) level within this fractal pattern that stands out. If that is not the case, then the species category, at least a single universal species category, is a myth: “What if there are theoretically important differences between the various segments of population lineages that we are identifying as species? Perhaps there are crucial differences between vertebrates, invertebrates, fungi and bacteria such that they should not all be regarded as forming the same kinds of species. What if, on our best theoretical understanding, there really do seem to be different kinds of species?” (Richards 2013, p. 60). If species taxa cannot be unified under the same kind of species category, then we would have to accept species pluralism. Yet, even if we accept, for the time being, the hierarchical solution of the species problem as far as the theoretical side of the coin is concerned, species delimitation is a fuzzy business. I will argue that completely objective, non-arbitrary species delimitation is impossible and that the approach coming closest to this unattainable ideal is not feasible and biologically irrelevant, which leaves us with the insight that taxonomy, even at its best, is only going to be an approximation of the real natural pattern and, importantly, that species delimitation resulting in biologically meaningful entities can only be done in hindsight. In other words, it is theoretically impossible to arrive at a fully consistent, completely non-arbitrary and biologically meaningful classification of life. The main reason for this is that taxonomy is essentially binary—species or no (sub-/super-)species, one or two (sub-/super-)species, etc.—while the evolutionary process is continuous and will create inherently fuzzy boundaries. The practical side of the species problem, the actual delimitation of species, is what biologists are dealing with every day—either directly as taxonomists or indirectly as “users” of taxonomy when they count species, analyse and compare conspecific and interspecific populations or manage species. It is therefore hardly surprising that species delimitation has received increasing attention, boosted undoubtedly by the methodological revolution in both genomics and bioinformatics. A Scopus search revealed that the number of published papers that had “species delimitation” in their title, abstract or keywords increased from 93 between 2001 and 2005 through 321 between 2006 and 2010 to 885 between 2011 and 2015 (see Fig. 1 in Camargo and Sites, 2013, for a similar finding). The available analytical approaches to species delimitation are diverse and comprise tree-based and non-tree-based methods. A more recent development is the implementation of coalescent theory, but there are also old ideas, such as the one regarding species as fields for recombination which originally goes back to Carson (1957) but has been revived by first Doyle (1995) and then more recently by Flot et al. (2010), which delimit pools of alleles that recombine and co-occur in individuals. An overview and discussion of existing methods can be found in the following publications (and references therein): Sites and Marshall (2003, 2004), Knowles and Carstens (2007), Wiens (2007), Petit and Excoffier (2009), Birky et al. (2010), Ence and Carstens (2011), Fujita et al. (2012), Hey and Pinho (2012), Birky (2013), Camargo and Sites (2013), Carstens et al. (2013), Zhang et al. (2013). There is even evidence that, across eukaryotic groups, a very simple correlation seems to exist between sexual compatibility and so-called compensatory base changes (CBC) in the internal transcribed spacer 2 (ITS2) transcript secondary structure such that the presence of even a single CBC seems to coincide very reliably (>90 %) with sexual incompatibility (Muller et al. 2007; Coleman 2009 and references therein). However, no matter if simple methods are based on genetic distances or morphological gaps or gene flow across hybrid zones or more complex methods such as tree-based coalescent models or cohesion tests, what all these approaches have in common is that they aim at consistently identifying lineages. What they do not do is objectively mark the level which is considered the species level: they make grouping (more) objective, but not ranking! This is why Sites and Marshall (2004, p. 220) conclude “virtually all [methods of species delimitation] will require researchers to make qualitative judgments. For example, there is no objective criterion for how much morphological divergence is enough to delimit a species, what threshold frequency of intermediates is needed to delimit species by genotypic clusters […], what proportion of unlinked loci are needed to delimit coalescent species […], or what frequency cutoff most appropriately indicates that no significant gene flow is occurring between populations”. There simply is no silver bullet for species delimitation. The available species delimitation methods become ever more sophisticated, but they only help us identify lineages; none of them can answer the question of where the line of the species lineage should be drawn. They will always entail a decision on some kind of cut-off criterion. But an (arbitrary!) cut-off criterion that can be measured or tested for objectively is not the same as an objective delimitation criterion! The sooner we recognize that some arbitrariness is naturally inherent in species delimitation, the earlier we can focus on biological phenomena, not names. The grey area during lineage divergence is a field of exciting evolutionary and ecological phenomena and for research that should not be spoiled by endless and ultimately futile debates about how to make species delimitation absolutely objective. I will not discuss concrete species delimitation algorithms here but rather focus on the more general aspects and difficulties surrounding delimitation in a world of fuzzy boundaries.