Typification and taxonomic status re-evaluation of 15 taxon names within the species complex Cymbella affinis/tumidula/turgidula (Cymbellaceae, Bacillariophyta)

Abstract Specimens belonging to the Cymbella affinis / Cymbella tumidula / Cymbella turgidula species complex have many taxonomic problems, due to their high morphological variability and lack of type designations. Fifteen taxon names of this complex, distributed in five species, were re-evaluated concerning their taxonomic status, and lectotypified based on original material. In addition to light microscopy, some material was analyzed by electron microscopy. Four new combinations are proposed in order to reposition infraspecific taxa.


Introduction
Th e history of the genus Cymbella C.Agardh is replete with taxonomic complexities. Within these complexes many species are similar in valve morphology. Much of the confusion in these complexes was caused by poor species descriptions including specimen images and the lack of designation of types, which has been required by the International Code of Nomenclature for algae, fungi, and plants (ICN) only since 1958 (McNeill et al. 2012).
In the most recent revision of the genus Cymbella, Krammer (2002) characterized and emended species descriptions within the complex C. affi nis Kütz. / C. tumidula Grunow / C. turgidula Grunow. Th is has generated confusion in the taxonomy of the group, specifi cally concerning the typifi cation of C. affi nis. Th e concept of C. affi nis as proposed by Krammer (2002) involved the synonymization of C. affi nis and C. tumidula Grunow, and the restoration of C. excisa Kütz. Cymbella excisa has previously been treated by some authors as a synonym of C. affi nis, at the same or at an infraspecifi c rank (e.g. Cleve 1894, Grunow 1882, Patrick and Reimer 1975. According to Krammer (2002), the specimens treated as C. affi nis have cymbelloid outlines, with the axial area straight to slightly curved and a prominent central area. In these specimens the valvar ends are more protracted and more densely striated and areolated than in C. excisa, which has the axial area curved and an indistinct central area. Moreover, C. excisa commonly has an excision in the ventral middle part of the valve. Krammer in Krammer and Lange-Bertalot (1986) also designated a neotype for C. turgidula, a species with an outline very similar to C. affi nis but with wider, fewer and uniformly distributed striae, and less-dense punctae in the striae. In the same contribution, Krammer also described the new species C. subturgidula Krammer, which he distinguished from C. turgidula by its narrower breadth, higher length/breadth ratio, and the size and shape of the central area. Tuji (2007) designated a lectotype for C. affi nis from an original illustration provided by Kützing (1844, Pl. 6, Fig. 15). Moreover, he assigned an epitype to a single specimen in preparation BM 18530 (Tuji 2007, Fig. 9) which was made from the original sample, which is Kützing's packet 333. Tuji (2007) observed that the designation of a neotype for C. turgidula by Krammer in Krammer and Lange-Bertalot (1986) was inappropriate. Th is designation was made because Krammer was unable to locate the original samples or preparations (Krammer and Lange-Bertalot 1986). Tuji (2007) found the original slide (syntype) used by Grunow in Schmidt (1875) to describe C. turgidula, and designated an original illustration as a lectotype and one specimen on Grunow's slide as an epitype.
Despite the changes, the revision of Tuji (2007) did not resolve the status of taxa such as C. tumidula or C. excisa. Moreover, he overlooked some requirements by the ICN with respect to the validity of the name C. uenoi.
Th e aim of the present study was to elucidate the current taxonomic status of C. tumidula, C. excisa and C. subturgidula, as well as of C. uenoi and C. uenoi f. nipponica [≡ C. turgidula var. nipponica; ≡ Cymbella rheophila]; to revisit the infraspecifi c taxa encompassed in this complex of species; and to lectotypify the ambiguous taxa.
Scanning Electron Microscopy (SEM) analyses were carried out only for C. turgidula var. nipponica. Samples were deposited on cover slips and attached to aluminum stubs using LeitSilver® (Sigma-Aldrich, Berlin, Germany). Th e material was coated with 150-200 Å of gold in an Emitech K 550 sputter coater (Quorum Technologies Ltd., Kent, UK). Th e material was analyzed in a Jeol JSM-6390 scanning electron microscope (Jeol, USA), operated at 6-8 kV, spot size 10-30, in the electron microscopy laboratory in the Museu Nacional, Rio de Janeiro, Brazil.
Th e term "degree of dorsiventrality" is used here to defi ne how dissimilar the sides of the valvae are on the apical axis. Th e symbols "≡", "=" and "-" preceding  Epitype. An individual on preparation BM18530, from Kützing packet 333, in the Natural History Museum (BM), designated by Tuji (2007, Fig. 9).
Remarks. Although Krammer (2002, p. 41) used material sampled by the same collector and from the same locality and related his new preparation to the type locality, his eff ort did not constitute a typifi cation of C. affi nis as ruled by the ICN (Art. 7.10, McNeill et al. 2012). On the other hand, Tuji (2007) was the fi rst to consider the original material of this taxon and performed a lectotypifi cation and epitypifi cation, and as such, must be followed according to Articles 9.19 and 9.20 (McNeill et al. 2012).
Th e lectotype valve of C. affi nis is similar to the lectotype valve of C. excisa, except for the excision in the middle part of the valve present in the majority of specimens . Populations of this complex from diff erent parts of the world may or may not have excisions, but this character is present in the majority of specimens from the populations examined (Krammer 2002). Th erefore, we consider C. excisa and C. affi nis as belonging to the same species with diff erences at the varietal rank based on phenotypic expression and ecological modifi cations. Genetic studies are still to be completed.
Th e two taxa were both proposed by Kützing (1844, p. 80) and therefore have the same priority. In similar cases, Article 11.5 of the ICN rules that "the fi rst such choice to be eff ectively published establishes the priority of the chosen name".
Grunow (1882) proposed the new combination and the new status of C. excisa for C. affi nis var. excisa (Kütz.) Grunow. Th is was the fi rst publication that defi ned  the priority of the epithet affi nis over excisa at the specifi c level. Th erefore, the name C. affi nis must be considered to be the name of the species when C. excisa and C. affi nis are considered to be the same species, in conformity with Article 11.5 of the ICN (McNeill et al. 2012).
Cymbella affi nis valves sensu Patrick and Reimer (1975, p. 57) have similar outlines as C. affi nis in the type population, but higher range of length and breadth values (length: 20-50 vs. 22.5-26.5 μm; breadth: 7-12 vs. 7.0-8.3 μm, respectively). However, these authors (Patrick and Reimer 1975), included representatives of diff erent localities that could encompass diff erent varieties subscribed to this taxon. Th e density of striae in the material from USA was similar in the middle part of the valve compared to the type material (9-11 vs. 9-12 striae in 10 μm, respectively) and lower when comparing them close to the ends of the valvae (12-14 vs. 11-18 striae in 10 μm, respectively) (Patrick and Reimer 1975).

Cymbella affi nis var. excisa (Kütz.) Grunow
Remarks. Similarly to C. affi nis, the lectotypifi cation of Cymbella excisa designated by Krammer (2002, p. 26) cannot be considered according to Article 7.10 of the ICN (McNeill et al. 2012), because the phrase "designated here" or equivalent is required from 1 January 2001 and it was not included by Krammer (2002). Th erefore, the lectotype designated by us cannot be considered a replacement of Krammer's (2002) "lectotype"; rather, it is the fi rst lectotypifi cation of this taxon.
Th e main diff erence between C. affi nis var. excisa and the nominate variety is the presence of an excision in the middle portion of the ventral side of the valve, a characteristic common to populations of this taxon around the world (Krammer 2002). Th erefore, we consider that the presence of excisions in all populations, not present in the type material of C. affi nis, constitutes suffi cient grounds to consider C. affi nis var. excisa diff erent from the nominate variety, which conforms to the statement by Grunow (Schmidt 1875); and not a diff erent species. Krammer (2002) recorded populations in the isotype material of C. affi nis var. excisa with length 17-41 μm, and breadth 6.0-10.7 μm, which were higher than populations in the lectotype material. Krammer (2002) was able to observe initial and post initial cells which were similar to the minimum and maximum length and breadth of this taxon. However, Krammer (2002) included specimens of the variety excisa and the nominate variety in his description. Valvae dorsiventral, dorsal margin broadly convex, ventral margin straight; ends not protracted, rounded, or subrostrate to rostrate; length 21.5-41.0 μm, breadth 8.0-11.0 μm, L/B ratio 2.7-4.0; axial area narrow, linear-arched, indistinct central area; striae 9-13 in 10 μm, becoming 11-14 toward ends, one isolated pore at end of central striae on ventral side; 22-27 punctae in 10 μm.
Remarks. Th e combination of C. excisa var. procera with Cymbella affi nis would be illegitimate unless it was given a new name, because it would be a later homonym of C. affi nis var. procera Krammer. Specimens designated by Krammer (2002), i.e., the preparation 212A IOK (BRM), were found to belong to more than one taxon therefore to clarify the taxonomy we designated a lectotype as established in Art. 9.11 of the ICN (McNeill et al. 2012).
Similar to C. affi nis var. neoprocera, specimens designated by Krammer (2002), i.e., the preparation 752 IOK (BRM), were found to belong to more than one taxon. Th erefore we designated a lectotype for this taxon as established in Art. 9.11 of the ICN (McNeill et al. 2012 Type locality. Hungary, Balaton. Valves dorsiventral, dorsal margin broadly convex, ventral margin straight, usually with excision in middle portion; ends barely protracted, rounded, to broadly protracted, subcapitate; length 21.0-31.7 μm, breadth 7.4-9.0 μm, L/B ratio 2.7-4.0; axial area linear-arched, indistinct central area; striae 8-12 in 10 μm, becoming 9-15 towards ends, one isolated pore at end of central striae of ventral side; 26-31 punctae in 10 μm. Remarks. Krammer (2002) distinguished this variety from the variety excisa based on the shape of the ends of the valvae. However, he (Krammer 2002, p. 28) did not provide metric characterizations of this the variety subcapitata, which we observed to agree with the characterizations of C. excisa sensu Krammer (2002), except for the density of striae that was slightly lower in the variety subcapitata.
Specimens designated by Krammer (2002), i.e., the preparation 717A IOK (BRM), were found to belong to more than one taxon. Th erefore we designated a lectotype as established in Art. 9.11 of the ICN (McNeill et al. 2012 Remarks. Th e lectotypifi cation of C. affi nis allowed us to consider C. affi nis and C. tumidula [-C. affi nis sensu Krammer], lectotypifi ed and epitypifi ed here, as independent species. C. tumidula has a more lanceolate outline, subcapitate ends, and a lower degree of dorsiventrality than C. affi nis. Th e striae in the middle part of the valve are shorter and unevenly distributed in C. tumidula, forming a distinct central area , in contrast to C. affi nis where the central area is indistinct and the striae are uniformly distributed. In addition, all specimens of the type material of C. affi nis have only one stigma, whereas in C. tumidula 1-5 stigmata can be observed.
Specimens from the population of the holotype material of C. affi nis var. procera were very similar in outline but larger and wider than C. tumidula var. tumidula, resulting in higher maximum length/breadth ratios. However, all metric characteristics of C. affi nis var. procera intergraded with C. tumidula var. tumidula, and therefore this taxon was transferred to C. tumidula var. procera.

Remarks. According to
Specimen designated by Krammer (2002), i.e., the preparation 714 IOK (BRM), were found to belong to more than one taxon. Th erefore we designated a lectotype as established in Art. 9.11 of the ICN (McNeill et al. 2012).

Holotype. Preparation 1603 in the Grunow Collection in the Naturhistorisches Museum Wien (W).
Lectotype (designated here). An individual on preparation 1603, in the Grunow Collection in the Naturhistorisches Museum Wien (W), represented by the illustration in Krammer (2002, Fig. 25: 13).
Remarks. Th is taxon presents morphometric characteristics similar to C. tumidula var. tumidula, except it has wider valves. Cleve (1894) recorded specimens of C. tumidula var. salinarum with 27-40 μm length, 8-10 μm breadth, and 11 or 12 striae in 10 μm, and considered that the only diff erence between this taxon and C. tumidula var. tumidula was the shape of the ends. Although in poor condition, in preparation 1603 we did not fi nd diff erences between the shape of the valvar ends of the variety salinarum and the nominate variety. However, C. tumidula var. salinarum has higher breadth values compared to the type population of C. tumidula var. tumidula, even in populations of this taxon as recorded by Krammer (2002) from Falaise where initial and post initial cells were found. Moreover, the occurrence of C. tumidula var. salinarum has been restricted to brackish waters. Krammer (2002, Fig. 25: 13) provided the illustration of an individual of the type of C. salinarum. Th e individual represented by him (Krammer 2000) was similar to C. tumidula var. tumidula. However, it was larger and had only one isolated pore, diff ering from C. tumidula, which has more than two isolated pores (Figs 50-56). Th us, in contrast to Krammer (2002), who treated C. salinarum at the specifi c level, we consider this taxon at the infraspecifi c rank as did Cleve (1894).
Remarks. Krammer (2002) described C. tropica and recorded diff erences in the size, length/breadth ratio, and the presence of only one stigma as consistent diagnostic diff erences between this species and C. turgidula. Tuji (2007) recorded the occurrence of specimens of C. turgidula with 1-3 isolated pores, which was also observed by Krammer (2002) in his material. We observed a continuum between the metric characteristics of C. tropica and C. turgidula, even in those characters that were considered by Krammer (2002) as diff erentiating. However, the outline was more lanceolate, the ends more protracted and slightly defl ected to the ventral side, and the degree of dorsiventrality was lower in C. tropica compared to C. turgidula. Lectotype (designated here). An individual on preparation 1046e IOK, in the Alfred-Wegener-Institut für Polar-und Meeresforschung (BRM), Bremerhaven, Germany, represented by Fig. 73.
Type locality. Korea, Ulchin County, Kyungsang Pukdo, Kwangchun River. Valvae slightly lanceolate to lanceolate, dorsiventral, dorsal margin broadly convex and ventral margin straight to convex; ends barely protracted, subrostrate to broadly subcapitate; length 26.3-41.0 μm, breadth 9.0-13.5 μm, L/B ratio 2.3-3.6; axial area linear to linear-lanceolate, arched, central area indistinct to slightly rounded; striae 9-13 in 10 μm, becoming 12-15 toward ends, 1-3 isolated pores; 21-26 punctae in 10 μm. In SEM, the striae showed lineolate punctae externally and internally, the striae are composed by an alveolus internally, surrounded by thick costae; the isolated pores are rounded externally; internally, the alveoli of the isolated pores are irregularly obovate and connected to intercostae, the margins with tooth-like structures (brocca); one apical pore fi eld (APF) not divided by the external terminal fi ssure of the raphe can be observed on each pole of the valvae; the terminal nodule extends to the dorsal side, under the APF and has a short branch that penetrates the APF apically; the helictoglossae lie under the terminal nodule and are defl ected to the dorsal side.
Remarks. Krammer (2002) described C. subturgidula based on preparation 1046c, which he designated as the holotype. Th is preparation was sought in the Alfred-Wegener-Institut für Polar-und Meeresforschung (BRM), where the entire Krammer Collection was transferred. However, preparation 1046c IOK is from Argentina in South America, and not from the holotype designated from Korea. In the protologue of C. subturgidula, Krammer (2002, p. 278 and 279) illustrated three specimens from preparation 1046E IOK, from Korea. Th us, the existence of slide 1046c IOK from Argentina, which is incongruent with the type locality, and the existence of slide 1046E which was used by Krammer to illustrate C. subturgidula, led us to consider the indication of preparation 1046c IOK as a typographical error.
Cymbella subturgidula and C. turgidula are closely related species. However, C. turgidula is more lanceolate and has a higher degree of dorsiventrality than C. subturgidula. Moreover, C. turgidula is slightly broader than C. subturgidula, with a more prominent ventral side of the valve. Th e ends in the two species are diff erent, being subrostrate-rounded in C. turgidula and slightly subrostrate-truncate in C. subturgidula. Th e central area is more distinct in C. turgidula than C. subturgidula. Although the number of punctae in 10 μm is the same in both species, the striae in C. subturgidula seem to be more coarsely punctuated than in C. turgidula. Cymbella turgidula var. nipponica was described by Skvortzow (1936). He considered that this taxon diff ered from the nominate variety due to the elongated valve, slightly undulate ventral margin, and broad rostrate ends. Ohtsuka and Tuji (2002) proposed that maintaining this taxon as a variety of C. turgidula was not appropriate. Th ey based their arguments on the co-occurrence of the nominate variety and the variety nipponica. Th erefore, they proposed the name C. rheophila T.Ohtsuka for this taxon at the specifi c rank.
Skvortsov and Noda (1971) described C. uenoi Skvortsov, but did not indicate any type. According to Article 40.1 of the ICN, names of new genera or taxa of lower ranks published after 1958 are valid only when the type is indicated (McNeill et al. 2012), and therefore C. uenoi is invalid. Tuji (2007), however, indicated a type for C. uenoi, fulfi lling the conditions required by the ICN. Th erefore, the author of the name becomes C. uenoi Skvortsov ex Tuji. Tuji (2007) also transferred C. turgidula var. nipponica [≡ C. rheophila] to that species, resulting in the name C. uenoi f. nipponica, considering erroneously that the name C. uenoi had priority under the name C. rheophila.  Tuji (2007, p. 54) suggested the conspecifi city of C. subturgidula and C. uenoi. Th e observations of the type material of C. uenoi provided by Tuji (2007) and C. uenoi f. nipponica [≡ C. rheophila], compared with the type material of C. subturgidula (i.e. morphometric characteristics), led us to agree with Tuji (2007). Since C. uenoi was validated only in 2007, the valid names of this species are either C. rheophila or C. subturgidula, both published in 2002, and not C. uenoi as stated by Tuji (2007 mer on 28 January 2002 (Koeltz Scientifi c Books, pers. comm.), while C. rheophila was not published before 29 July 2002, the date of acceptance of the paper. Th erefore, the epithet subturgidula has priority over the epithet rheophila.

Discussion
Few studies have discussed the criteria to delimit infraspecifi c ranks in diatoms. Cox (1997) reviewed this issue and found contradictions among the criteria adopted by diff erent researchers over time. In order to resolve this question, she proposed a pragmatic solution to delimit infraspecifi c taxonomic ranks. According to Cox (1997), the term variety should be used "for populations (within the same species) which are ecologically and morphologically distinct, in which there is no evidence of morphological intergrading under intermediate conditions".
However, taxonomic analyses of some species complexes in diatoms using molecular data have demonstrated that Cox's (1997) suggestions about the delimitation of varieties do not apply to every situation. Sellaphora pupula (Kütz.) Mereschk. sensu lato and Nitzschia palea (Kütz.) W.Sm. sensu lato, for example, have been shown to be an assemblage of pseudocryptic species that correspond, in the majority of instances, to populations with intergrading morphological characteristics (Mann et al. 2008;Trobajo et al. 2009). We have found a similar situation in populations that were clearly identifi ed as Cymbella tumida (Bréb.) Van Heurck (unpublished data).
Th e criteria adopted by Krammer (2002Krammer ( , 2003 to delimit several specifi c and infraspecifi c taxa of Cymbella take into account variations between the type and similar individuals from other populations. It is important to note that, for Krammer (2002Krammer ( , 2003, types can be individuals marked or represented in some illustration, as well as a group of individuals mounted in a preparation, as defi ned in the ICN (McNeill et al. 2012).
Because of the lack of taxonomic studies on the Cymbella affi nis/tumidula/turgidula species complex using molecular data, we opted to use Krammer's concept in order to attempt to organize this confusing group. Th is criterion is usual as a reference to circumscribe groups with similar morphologies, even if individuals with these morphologies sometimes overlap. Some authors prefer to defi ne similar groups as "morphodemes", which have no nomenclatural status. However, in some cases, this defi nition sounds more similar to the old taxonomy when any taxonomic unit could be denominated by diff erent names, but scientifi cally by a long sentence that is more similar to the current "diagnosis". Although the concept of varieties used here implies the publication of nomenclatural novelties, none of them are new proposals, but rather are simply adjustments of already existing names that are in confusing combinations, because of a misinterpretation by Krammer (2002) as well as the history of the taxa and the evolution of nomenclatural rules.
Th e criteria of delimitation of taxa and the weight of characters in diatoms are variable from group to group (Mann 1999). While in some groups the density of striae is a good morphological indicator of diff erent species, in other groups this can be irrelevant (Abarca et al. 2014). Diff erent sizes and valve outlines are not always good characters to delimitate taxa such as in the complex Encyonema silesiacum/minutum/ ventricosum (unpublished data). Th us, it is common to fi nd taxa with polythetic defi nitions, that is, in which a set of characteristics, sometimes interweaving with other taxa, are taken into account (Needham 1975). Th e delimitation of Cymbella taxa is a clear example of polythethism (Krammer 2002).
In this context, several characters must be considered in their characterization and identifi cation of taxa. Th e degree of dorsiventrality of the valve, for example, is slightly higher in C. affi nis than in C. tumidula, and even higher in C. schilleri Krammer (2002, Fig. 26: 7, 8) than in C. orientalis Lee (Krammer 2002, Fig. 26: 1-6). Another example is the thickness of the striae. Th e striae of C. tumidula are narrower than in individuals of C. affi nis in LM, even in specimens of the same size. Th is is often the result of thick intercostae and the size of areolae, which can only be observed in SEM (Krammer 2002, Figs 5: 1, 23: 18). Th us, these characters, alone, seem to be unimportant but in combination with further features can provide a better concept of the taxa.
Th e economic and ecological uses of diatoms require a refi ned taxonomy, which is more detailed than simply species complexes. Th is is especially true for bioassessments using diatoms (European Committee for Standardization 2003). Gomphonema lagenula Kütz. has for a long time been treated as synonym of G. parvulum Kütz., since the variability in the shape of the ends were considered insuffi cient to distinguish the taxa (VanLandingham 1971). However, diff erences in the ecological preferences of these two taxa were recorded and the independence of the two species has been confi rmed by molecular data (Kermarrec et al. 2013, Abarca et al. 2014. Diff erences between the ecological preferences of Nitzschia palea (Kütz.) W.Sm. var. palea and N. palea var. debilis (Kütz.) Grunow have also been recorded, and simply morphological characters are insuffi cient to separate the series of phenotypic expressions subscribed to N. palea (Trobajo et al. 2009). Th us, the effi ciency in the use of diatoms in activities such as bioassessments needs accuracy and taxonomic harmonization (Manoylov 2014). But this will only be possible if important characteristics such as ecological preferences can be permanently attached to a taxon or an accessible designation, which is facilitated by the establishment of correct types and by the knowledge of these types.
Naturally, beside the knowledge of types, supplementary studies are necessary to record phenotypic plasticity resulting from diff erent ecological conditions or by life cycles. Such studies should be carried out in natural or cultivated samples in order to observe a more realistic concept of the species (Mann 1999). Th ese studies will allow us to observe slight changes in the morphology of the valves (i.e., outline, measures, etc.) such as some of those observed in the type population of the taxa discussed here. Moreover, these studies can possibly verify the relationship among morphological characters, which are associated in a polythetic way during the establishment of specifi c concepts and in their use in the identifi cation of taxa of similar groups.

Conclusion
Th e process of lectotypifi cation can markedly infl uence the identity of some taxa and can sometimes substantially change the relation to other taxa. Th e designation of a type for C. affi nis resulted in a profound restructuring of C. affi nis, C. excisa and C. tumidula. C. excisa has been shown to be the same taxon at the species level as C. affi nis, but because of its specifi c morphology is treated herein at a diff erent rank. Th e epithet affi nis has priority over the epithet excisa, as defi ned by the criterion of the fi rst eff ective publication. Th us, four infraspecifi c taxa of C. excisa were transferred to C. affi nis.
Th e lectotypifi cation of C. tumidula Grunow and comparisons with the lectotype of C. affi nis allowed us to conclude that the two species are independent. C. affi nis var. procera was treated as a new species, which is closer to C. tumidula than C. affi nis because of morphological similarities. Infraspecifi c taxa described by Krammer (2002) within C. affi nis had small diff erences in relation to the type of C. tumidula, and they are recombined herein.
Th e analysis of the type and the history of taxa such as C. subturgidula Krammer, C. rheophila Ohtsuka, and C. uenoi Skvortsov ex Tuji allowed us to conclude that these taxa are conspecifi c, and to determine that the epithet subturgidula has priority.