Phylogenetic analysis places Spicaticribra within Cyclotella

A strong consensus has emerged that taxonomic classifications should be based on an underlying phylogenetic hypothesis. According to this view, named groups should be monophyletic, ensuring that a name uniquely matches the evolutionary history and biological attributes of a group of taxa. As originally conceived, the diatom genus Cyclotella is a large and morphologically diverse assemblage of taxa that we now know consists of several distantly related lineages. Considerable progress has been made in placing these lineages into different monophyletic genera. The genus Spicaticribra was originally described as monotypic and has features that suggest a close relationship to Cyclotella, but it has also retained some ancestral features that appear to differentiate it from Cyclotella. We sequenced two nuclear and two plastid genes to resolve the phylogenetic position of Spicaticribra and show that it is embedded within a clade that includes the type species of Cyclotella and, further, that maintaining Spicaticribra renders Cyclotella non-monophyletic. We transfer Spicaticribra species into Cyclotella, resolve related nomenclatural issues, and caution against using ancestral characters and character states for taxonomic classification.


Introduction
Among the 820 accepted species names in the Thalassiosirales, nearly half of them have been classified as either Thalassiosira Cleve (193) or Cyclotella Brébisson (170) (AlgaeBase, accessed 12 February 2021, Guiry & Guiry 2010). These numbers reflect similar and longstanding taxonomic challenges presented by these two genera. Both of them encompass a broad range of morphological diversity, and their 'defining' morphological features include characters that, in some cases, appear to have been present since the origin of the Thalassiosirales. The use of ancestral characters for classification can lead to the bloating of taxonomic groups over time due to the inclusion of distantly related species. This situation applies to Thalassiosira, a large genus that phylogenetic analyses have shown to be polyphyletic (Alverson et al. 2007). Efforts to diagnose and name clades within the broadly and loosely defined Thalassiosira (Alverson et al. 2006, Stachura-Suchoples & Williams 2009) have left the remaining species polyphyletic, which we view as a transitional state of progress toward establishing a natural classification of this genus.
The challenge of Cyclotella is somewhat simpler because although it, too, has accumulated a large set of morphologically diverse species, decades of research *Corresponding author. E-mail: aja@uark.edu Associate Editor: Bank Beszteri (Received 10 December 2020;accepted 7 April 2021) have revealed an information-rich set of characters that have made it easier to first delineate phenetic morphological groups (e.g. Lowe 1975) and later to distinguish between ancestral and derived character states. As a result, many extant species can now be classified into one of several monophyletic genera that formerly fell under the For example, it can be hard to classify fossil taxa that have a mix of ancestral and derived character states (Stone et al. 2020), and the failure to make this distinction can lead to descriptions of new genera that make existing ones non-monophyletic. Spicaticribra Johansen, Kociolek & Lowe is one such genus. It was first described as monotypic and was defined by the following diagnostic characters: continuous 'spicate' cribra on the interior valve face, absence of central strutted processes, and absence of external extensions of the marginal strutted processes (Johansen et al. 2008). Johansen et al. (2008) suggested that the lack of distinct regions on the valve face appeared to place it outside Cyclotella, whereas the presence of continuous cribra indicated a closer relationship to Thalassiosira, another polyphyletic genus. Without a home for the newly discovered type species, S. kingstonii Johansen, Kociolek & Lowe, it was placed into its own genus. Other species were later placed into Spicaticribra, and the online nomenclatural database, AlgaeBase, lists a total of 10 Spicaticribra species (accessed 12 February 2021, Guiry & Guiry 2010).
We collected and cultured S. kingstonii from its type locality. Phylogenetic analyses of four DNA markers place S. kingstonii within Cyclotella sensu stricto. In an effort to preserve the phylogenetic integrity of Cyclotella, we transfer S. kingstonii and other Spicaticribra into Cyclotella.

Methods
We collected near-surface phytoplankton with a 10 μM mesh plankton net from the Tuckasegee River, North Carolina, USA, (35.439933, − 83.55145) on 24 April 2017. We isolated individual cells with a micropipette and grew them at 21°C in WC medium (Guillard & Lorenzen 1972). We cleaned clonally cultured cells with nitric acid and rinsed them with deionized water until the solution reached a neutral pH. Cleaned cells were transferred onto coverslips and allowed to evaporate overnight before permanently affixing them onto microscope slides with Naphrax ® . We identified and photographed cells at 600 × magnification with a Zeiss compound microscope. For scanning electron microscopy, cleaned cells were dried onto 12 mm diameter coverslips and coated with 15 nm of iridium with a Cressington 208 Bench Top Sputter Coater (Cressington Scientific Instruments, Watford, UK). Scanning electron micrographs were taken with a Zeiss SUPRA 40 VP scanning electron microscope (Carl Zeiss Microscopy, Thornwood, NY, USA).
We collected live cells from a single clonal culture by centrifugation, vortexed them with 1.0 mm glass beads, and extracted DNA with a Qiagen DNeasy Plant Kit. We sequenced two nuclear (SSU and partial LSU rDNA) and two plastid (rbcL and psbC) genes. Primers, PCR conditions, and Sanger sequencing followed Alverson et al. (2007). We added Spicaticribra sequences to gene alignments from Alverson et al. (2011), using SSU-ALIGN version 0.1.1 (Nawrocki et al. 2009) to align rDNA sequences with the covariance models included in the program for the SSU alignment, and a heterokont-based covariance model for the LSU alignment (Nakov et al. 2014). We removed poorly aligned sections of the alignment with SSU-MASK using the default settings, which retains columns with a Bayesian Posterior Probability of 0.95 of being correctly aligned. Multiple sequence alignment of the psbC and rbcL plastid genes was performed manually in AliView version 1.25 (Larsson 2014). We used trimAl version 1.4 (Capella-Gutiérrez et al. 2009) to remove alignment columns with gaps in more than 20% of the sequences. We concatenated sequences for all four genes into a single alignment with AMAS (Borowiec 2016) and used IQ-TREE version 1.6.4 (Nguyen et al. 2015) to reconstruct phylogenetic relationships. The concatenated sequences were partitioned by gene, and the best-fit substitution model for each partition was inferred using the ModelFinder algorithm implemented in IQ-TREE. We inferred the maximum-likelihood tree using the edge-linked partition model in IQ-TREE and applied the TIM2 + F + R3 model to the LSU partition, TN + F + R3 to SSU, GTR + F + I + G4 to psbC, and the GTR + F + R4 model to the rbcL partition. Branch support was assessed with 100,000 bootstrap replicates using Ultrafast Bootstrap Approximation (Hoang et al. 2018) and the setting '-bnni' to guard against overestimation of branch support.
Newly generated DNA sequences are available from the National Center for Biotechnology Information's Gen-Bank database under accession numbers MW327042, MW327043, MW326755, and MW326756. Multiple sequence alignments and tree files are available from a Zenodo online repository (10.5281/zenodo.4313346).

Results
We collected and cultured S. kingstonii from the Tuckasegee River, USA, which is part of the same riverine/reservoir system as Fontana Lake, the type locality for this species and for the genus (Johansen et al. 2008). Light microscope images confirmed the identity as S. kingstonii (Figs 1-2). Phylogenetic analyses of two nuclear and two plastid genes placed S. kingstonii as sister to a clade that includes Cyclotella distinguenda Hustedt, the type species of Cyclotella, and other Cyclotella species (Fig. 3). Cyclotella nana Hustedt was sister to the Spicaticribra + Cyclotella clade (Fig. 3). By separating C. nana from the remaining Cyclotella, Spicaticribra renders Cyclotella non-monophyletic.

Discussion
A strong consensus has emerged that taxonomic classifications should be natural, meaning that named groups are monophyletic, which ensures that a name corresponds to a shared, unique evolutionary history among the taxa bearing that name (Kociolek et al. 1989, Williams & Kociolek 2011, Kociolek & Williams 2015. A great deal of progress has been made in using phylogenetic trees and phylogenetic character interpretations to subdivide the Thalassiosirales into monophyletic genera. Most of this work, and much of the work that remains, concerns the two largest genera, Cyclotella and Thalassiosira. Molecular phylogenies have facilitated these efforts, but a phylogenetic tree is a hypothesis that can change, along with the classification that accompanies it, as new taxa are discovered (Kociolek & Williams 2015). A natural genus-level classification can also be disrupted in unintended ways by the discovery of new taxa without a clear generic affiliation. This is especially problematic when character interpretations of the new taxa are not made with respect to phylogenetic relationships (Kociolek & Williams 2015). New and challenging taxa are often placed into monotypic genera, either because of ambiguity about their relationship to other genera (Johansen et al. 2008), because they exceed some arbitrary threshold of difference to known genera (Williams 2009(Williams , 2013, or to intentionally communicate uncertainty about their relationship to other taxa (Williams 2013, Stone et al. 2020). Without 'evaluation of all available evidence in terms of monophyly and synapomorphy' (Williams & Kociolek 2011, p. 51), a newly described monotypic genus can make existing ones non-monophyletic, thereby undermining the primum non nocere principle that seems to underlie many of the guiding principles laid out by Kociolek & Williams (2015), an approach we see as having greatly improved our approach to diatom systematics and classification.
Spicaticribra was defined originally by the presence of a 'spicate' pattern of continuous cribra internally, absence of central strutted processes, and absence of external extensions of the marginal strutted processes (Johansen et al. 2008). Unless phylogenetic analyses demonstrate derived secondary loss of a character, we agree that character absences generally do not provide compelling evidence in support of a genus (Kociolek & Williams 2015), so we focus here on the internal cribra. The diagnosis of Spicaticribra was later broadened by Khursevich & Kociolek (2012) to include the following characters: loculate areolae with continuous or semicontinuous cribra and external foramina; plicated valve face, or not; one or more labiate processes that extend outwardly from the frustule, or not; strutted processes with 2-4 satellite pores that extend outwardly from the frustule, or not; absence of strutted processes on the valve face, or rarely not. By relaxing the required absence of central strutted processes and external extensions of the strutted processes, and allowing for presence or absence of other characters, the amended definition captures genera and species from across nearly all of the Thalassiosirales. Several species were identified as candidates for transfer to Spicaticribra under this new definition (Khursevich & Kociolek 2012), including T. lacustris (Grunow) G.R.Hasle, T. gessneri, C. guillardii, C. weissflogii (Grunow) Stachura-Suchoples & Williams, and C. nana -a group of taxa whose common ancestor traces back almost to the root node of Thalassiosirales (Fig. 3). Several of these (C. guillardii, C. weissflogii, and C. nana) were later transferred (Khursevich & Svirid 2013) despite known phylogenetic positions which showed them to be polyphyletic (Alverson et al. 2007). These transfers rendered Conticribra, Cyclotella, and Spicaticribra non-monophyletic (Fig. 3).
A large body of research, dating back to early applications of the scanning electron microscope for describing frustule morphology, has highlighted the importance of strutted process ultrastructure in understanding the phylogeny (Theriot & Serieyssol 1994, Shiono 2000 and classification of the Thalassiosirales (Fryxell & Hasle 1979, 1980. Spicaticribra has a strutted process ultrastructure not found outside of the cyclotelloid and cyclostephanoid lineages of the Thalassiosirales (Alverson et al. 2007). The strutted processes of Spicaticribra, which feature robust cowlings and broad satellite pore covers (Fig. 4), resemble those of many Cyclotella species (e.g. C. distinguenda) (Fig. 5) and clearly indicate a closer relationship with Figs 4-7. Strutted process ultrastructure of two species in the Cyclotella clade, Spicaticribra kingstonii (4) and C. distinguenda (5), and two Thalassiosira species, T. pacifica Gran & Angst (6) and T. nordenskioeldii Cleve (7). Scale bar = 250 nm. these species than any Thalassiosira (Figs 6-7). Johansen et al. (2008)  shown that S. kingstonii falls in the middle of the grade between C. nana, C. distinguenda (the generitype), and the rest of Cyclotella -a result that we again recovered with our analyses of S. kingstonii from its type locality (Fig. 3). In short, Spicaticribra has rendered Cyclotella non-monophyletic, a result that closely matches one of the hypothetical consequences of describing monotypic genera without reference to a phylogenetic hypothesis (Kociolek & Williams 2015).
Two options are available for preserving monophyly of Cyclotella: retain Spicaticribra and place C. nana into a monotypic genus, or include Spicaticribra in Cyclotella (Fig. 3). The nomenclatural history of C. nana has been highly unstable, which has led to confusion about the biology and natural history of this important model species (Alverson et al. 2011). The placement of C. nana in Cyclotella dates back to its original description by Hustedt (1957), who correctly diagnosed its placement in Cyclotella despite the limited information available to him at the time. A solution to the conflict resulting from the creation and later expansion of Spicaticribra should prioritize historical precedence, taxonomic continuity, and maintain recent progress in stabilizing the taxonomy of Cyclotella. To this end, we recommend that Spicaticribra should not be recognized as a separate genus, and we therefore transfer the following species to Cyclotella: