One species or four? Yes!...and, no.
Or, Arbitrary assignment of lineages to species obscures the diversification processes of Neotropical fishes

Species are fundamental units in many biological disciplines, but there is continuing disagreement as to what species are, how to define them, and even whether the concept is useful. While some of this debate can be attributed to inadequate data and insufficient statistical frameworks in alpha taxonomy, an equal part results from the ambiguity over what species are expected to represent by the many who use them. In many cases the species “concept” acts as a red herring with regard to the identification of lineages and their properties, serving to distract from our actual goal, which is to understand the processes of diversification. One domain where this appears to be prevalent is in resolution of the most lineage dense assemblage of vertebrates on Earth, Neotropical freshwater fishes, which continue to be described under a paradigm presuming the structure of meta-population lineages can be adequately described by discrete taxa, despite ample empirical evidence that nature exhibits nested endemism and reticulate lineages that are incongruent with the operational expectations of the “species” concept. As a case study, I examined the delimitation of lineages in a “species complex” of “peacock bass” cichlid fishes from the Amazon, the Cichla pinima sensu lato group, using extensive molecular data. Although described in 2006 as four species based on highly variable coloration patterns and overlapping meristics, these populations were later shown to exhibit two non-sister mtDNA lineages that did not correspond to the described species and which were haphazardly distributed geographically. In contrast, microsatellite data showed extensive areas of both admixture between two clusters as well as areas of no admixture. Finally, sequences from 17 nuclear genes recovered all the localities of these species as monophyletic, in contrast with the mtDNA genealogy.

I re-examined these data in an integrated and quantitative manner to resolve their contemporary and historical structure and their correspondence to “species”. The power of the strategy applied here is that each of these apparent conundrums was largely decipherable using model-based approaches that allowed testing of the apparent conflict within and between datasets. A spherical phylogeographic analysis revealed that the admixture represented in the microsatellite data result from a recent joint colonization of the eastern Amazon by two non-sister lineages (“Southern” and “Western”) that evolved allopatrically but that apparently did not develop intrinsic reproductive isolation. The populations in the area recently colonized exhibit a hybrid ancestry that manifests itself in their nuclear sequences and which is strong enough to affect a common ancestry (monophyly) for the Southern and Western lineages under multi-species coalescent models. Analyses of the microsatellite data with approximate Bayesian computation showed that the degree and timing of admixture differs across these sub-populations, while FST-analogs implied that this has been preserved by the low rates of contemporary gene exchange between these eastern localities. Despite this extensive admixture, multi-species coalescent-based delimitation with Bayes Factors not only recognized the separate evolutionary history of the Southern and Western lineages even when these lineages were reconstructed as monophyletic, but also emphasized the distinction of several nested Western lineages that exhibit unique mtDNA sub-lineages. However, the mtDNA and microsatellite data both confirmed that at least two of these western sub-lineages have experienced recent gene exchange.

Although these analytical results provide a cohesive narrative for these lineages in terms of the divergence, dispersal, and admixture processes that produced the current pattern, it suggests that reconciliation of the evolutionary patterns of Cichla pinima sensu lato with all but the most arbitrary definitions of “species” cannot be done without ambiguity. Any of several renderings (one, two, or four species) would be accurate with respect to divergence history and contemporary population structure, but each would require subjective preference of criteria for conferring the species rank and would obscure the processes that have contributed to this diversity. For example, though the Southern and Western lineages had a separate evolutionary origin, they apparently did not develop reproductive isolation. This characteristic would not prevent their recognition as species under some concepts, but the result has been that there are now a comparable number of admixed populations of these lineages as unmixed and that they may now be functionally inseparable. On the other hand, recognizing these as a single, hybrid species would ignore the ongoing diversification in the western lineage. It would also treat each of the admixed populations as equal replicates of the parental species, despite evidence that they are unique in the degree and age of admixture, with presumably unique solutions for local adaptation. More broadly, if these lineages are synonymized because of admixture or incomplete reproductive isolation, what does this imply about the remaining species of Cichla, which exhibit degrees of hybridization between both the closest and most distantly related species?

These results suggest that the historical narrative for these populations is more complex than can be portrayed by recognizing them as one, two, or four species: their history and contemporary dynamics cannot be unambiguously rendered as discrete units (taxa) at any level without both choosing the supremacy of one delimitation criterion and obscuring the very information that provides insight into the diversification process. This calls into question the utility of species as a rank, term, or concept, and suggests that while biologists may have a reasonable grasp of the structure of evolution, our methods of conveying these insights need updating. The lack of correspondence between evolutionary phenomena and “species taxa” should serve as a null hypothesis, and researchers should focus on quantifying the diversity in nature at whatever hierarchical level and in whatever form it occurs without the distraction of making a subjective determination as to whether it can be captured by discrete taxa or rises to the rank of “species”.