Naming diversity in an evolutionary context: Phylogenetic definitions of the Roucela clade (Campanulaceae/Campanuloideae) and the cryptic taxa within

Abstract In recent times, evolution has become a central tenet of taxonomy, but nomenclature has consistently been decoupled from the tree‐thinking process, often leading to significant issues in reconciling traditional (Linnaean) names with clades in the Tree of Life. Recent evolutionary studies on the Roucela clade, a group of endemic plants found in the Mediterranean Basin, motivated the establishment of phylogenetic concepts to formally anchor clade names on the Campanuloideae (Campanulaceae) tree. These concepts facilitate communication of clades that approximate traditionally defined groups, in addition to naming newly discovered cryptic diversity in a phylogenetic framework.


| INTRODUCTION
With the advent of phylogenetic systematics (Hennig, 1950(Hennig, , 1966, tree-thinking has seized a prominent role in building classifications that more closely reflect the evolutionary history of taxa. However, regardless of how taxa are discovered and diagnosed, biodiversity knowledge is directly or indirectly linked to names. Traditional (Linnaean) names appear to work well for groups that were defined under nonevolutionary frameworks but still approximate clades. However, they may not scale satisfactorily when repurposed for large clades with many nested taxa (e.g., insufficient ranks and/or significant nomenclatural instability due to name changes, among other issues) or to name newly discovered biodiversity below species level (e.g., naming cryptic taxa without destabilizing currently accepted species names). The motivation to develop a phylogenetic system of nomenclature has been well delineated in numerous papers since the early 1990's (de Queiroz & Gauthier, 1990, 1992, 1994, including alternative approaches to naming species (Cellinese, Baum, & Mishler, 2012;Dayrat, Schander, & Angielczyk, 2004).
However, more specifically rationale for generating and making available meaningful phylogenetic definitions (phyloreferences) to (1) reconcile taxonomic names with their concepts; (2) allow data integration on the Tree of Life with confidence; (3) query the Tree of Life without ambiguity and ultimately foster discovery has been covered only in the realm of phyloinformatics (N. Cellinese et al., unpublished).
Two recent phylogenetic studies have focused on the evolution of the Roucela complex (Crowl, Myers, & Cellinese, 2017;Crowl et al., 2015). These groups include small, herbaceous, annual Campanula plant species restricted to the Mediterranean Basin and characterized by a dichotomous branching habit and unappendaged calyx lobes (Carlström, 1986;Lammers, 2007). The last available taxonomic revision by Carlström (1986) recognized 12 morphological species, and later Tan and Sorger (1986) added to this complex by describing C. lycica from Turkey. This group has historically been very challenging to disentangle morphologically, resulting in its assignment to various ranks, including its own genus distinct from Campanula (Roucela; Dumortier, 1822), and later demoted to a subgenus of Campanula (Damboldt, 1976;Lammers, 2007).
Our recent study used this group as a model for understanding historical drivers of speciation and endemism in the Mediterranean Basin because of its narrow distribution and high level of endemism in this region (Crowl et al., 2015). Most species are narrow endemics to one or few islands primarily in the eastern Mediterranean Basin, with the notable exception of C. erinus, which is widespread across the Mediterranean climate zone, from the Arabian Peninsula to Macaronesia, including some coastal regions of North Africa.
In addition to disentangling the complexity of species relationships within this group, we also attempted to understand the processes leading to the peculiar distribution of C. erinus compared to its close relatives. Our phylogenetic analyses suggested that the nonmonophyly of C. erinus may be due to hybridization and cryptic diversity within this species. In Crowl et al. (2017), we uncover evidence that supports the occurrence of a hybridization event between the tetraploid C. erinus and the tetraploid C. creutzburgii, leading to an octoploid hybrid taxon that has remained historically hidden within the tetraploid C. erinus, mainly because it is morphologically indistinct from its parent. However, not only are these taxa genetically distinct, but they also occupy different geographic ranges. The tetraploid C. erinus is exclusive to the western Mediterranean Basin, whereas the octoploid C. erinus is found from the Balkans throughout the eastern range, including many islands. The generation of meaningful phylogenetic definitions and assignment of clade nomenclature to these independent entities are important to facilitate communication and query of these taxa and build a more accurate classification.

| THE ROUCELA CLADE
Several molecular phylogenetic studies (Cellinese et al., 2009;Haberle et al., 2009;Mansion et al., 2012) have consistently recovered a highly supported clade that traditionally included 12 annual Campanula species (Campanulaceae) found in the Mediterranean Basin (Carlström, 1986). Traditionally, this group has previously been referred to as the genus Roucela (Dumort.) Damboldt, or Campanula subg. Roucela (Dumort.) Damboldt, or the Roucela complex. More recently, Crowl et al. (2015) elucidated the systematics and historical biogeography of this clade (Figure 1), motivating the establishment of a formal phylogenetic definition for this group.

| Etymology
The name Roucela has previously been used at the rank of genus (Roucela Dumort.) and, more recently, subgenus (Campanula subg. Roucela [Dumort.] Damboldt). Here, we repurpose the name Roucela as a clade name to approximate the traditionally defined group named Campanula subg. Roucela (Dumort.) Damboldt. We prefer to select the name Roucela because it is the oldest name by which this group of close relatives has been consistently referred to. Crowl et al. (2015, figure 2, page 4).

| Composition
In the most recent taxonomic revision, Carlström (1986)  Carlström, and C. veneris Carlström. More recently, one additional species, C. lycica was described and added to this group by Tan and Sorger (1986). Although cryptic diversity appears to exist within Campanula erinus L., both C. erinus lineages were found to fall within this group.
According to the phylogenetic analyses of Mansion et al. (2012) and Crowl et al. (2015), C. scutellata Griseb. does not appear to belong to this clade.

| Diagnostic apomorphies
In addition to molecular synapomorphies, members of the Roucela clade are annual Campanula species with dichotomous branches and an unappendaged calyx (Carlström, 1986, Lammers, 2007.
Within the Roucela clade, the geologic history of the eastern Mediterranean appears to have played an important role in the diversification of many species, while the climatic history-specifically, the shift from a subtropical climate-may have adversely affected diversification (see Crowl et al., 2015 for an in-depth discussion).

| THE HOLOERINUS CLADE
A genomic dataset consisting of 130 nuclear loci and near-complete plastomes across 27 populations of Campanula erinus L., spanning its distribution range, provides strong evidence for two lineages within this currently recognized species . While a subset of the nuclear genome suggested the octoploid C. erinus lineage to be sister to From a phylogenetic perspective, the octoploid lineage is, therefore, sister to both parental lineages (Figure 2a). The phylogenetic definition provided here refers to the octoploid plus tetraploid C. erinus clade but does not include C. creutzburgii (Figure 2b).

| Etymology
We choose the name Holoerinus to indicate the inclusion of both tetraploid and octoploid lineages of C. erinus. This name combines the specific epithet of the traditionally recognized species, Campanula erinus, with the Greek prefix, holo-which means "whole." Crowl et al. (2017, figure 3, page 919).

| Composition
The Holoerinus clade is composed of both tetraploid and octoploid lineages within the traditionally recognized Campanula erinus, occurring throughout the Mediterranean Basin.

| Synonyms
None, although technically the Holoerinus clade approximates to the traditionally established C. erinus. However, we choose to ignore ranks (specifically, the rank of species in this case) and apply the name Holoerinus to the clade that includes all populations of tetraploid and octoploid C. erinus.

| General comments on Holoerinus
The phylogenetic analyses of Crowl et al. (2015)  result, however, was insufficient to draw meaningful conclusions.
Recent phylogenomic analyses of Crowl et al. (2017), which increased both population and genomic sampling, verified this nonmonophyly only in a subset of the genome, but found C. erinus to be monophyletic when considering plastome data and a number of nuclear loci.
This study concluded that the observed discordance was the result of hybridization, leading to the formation of two cryptic taxa within the traditionally recognized species. Although seemingly indistinguishable on the basis of morphology, populations belonging to these lineages are recognized on the basis of geography and ploidy: western Mediterranean tetraploids and eastern Mediterranean octoploids.
The Holoerinus clade, as defined here, includes both of these lineages and approximates in content to the traditionally established C. erinus ( Figure 2).

| Etymology
We have combined the traditional specific epithet of Campanula erinus with the Latin prefix, tetra-to reflect the ploidal level of this clade. Crowl et al. (2017, figure 1, page 914). See also Crowl et al. (2017, figure 2, page 919) for results from species-tree analyses.

| Composition
The Tetraerinus clade is composed of tetraploid populations of the traditionally recognized species, Campanula erinus, occurring in the western Mediterranean Basin from the Balkans to the Azores.

| Etymology
We have combined the specific epithet of Campanula erinus with the Latin prefix octo-to reflect the octoploid level of this clade, Crowl et al. (2017, figure 1, page 914). See also Crowl et al. (2017, figure 2, page 918) for results from species-tree analyses.

| Composition
The Octoerinus clade is composed of octoploid individuals within the traditionally circumscribed species, Campanula erinus. This cryptic taxon is found primarily in the eastern Mediterranean Basin.

| General comments on Octoerinus
We generated a phylogenetic definition for Octoerinus to distinguish cryptic octoploid and tetraploid taxa within the traditionally established species, Campanula erinus. This octoploid appears to be the result of an allopolyploid event in which the tetraploid Tetraerinus and tetraploid C. creutzburgii are the parental lineages.
The phylogenetic analyses of Crowl et al. (2017) showed that populations of different ploidal levels form separate clades, consistent with distinct geographic ranges. Because of the hybrid origin of this lineage, it is sister to both C. creutzburgii and the Tetraerinus ( Figure 2). Octoerinus can be distinguished from Tetraerinus based on ploidal level: octoploid (2n = 56) and tetraploid (2n = 28), respectively.

| CONCLUSIONS
During our studies on the evolution of Mediterranean Campanula species, we encountered a number of well-supported clades that also included unknown cryptic diversity due to hybridization and allopolyploidy. Here, we have established clade nomenclature for this group in order to anchor names and concepts to specific parts of the Campanuloideae tree. Clade nomenclature is native to tree-thinking, and therefore, it is useful for communicating taxa that are discovered in an evolutionary framework and allows for unambiguous assignment of names to nodes on the Tree of Life.