A taxonomic headache due to a ‘Frankenstein type’: solving the identity and typification of a Mimosa species

Abstract A nomenclatural type is the element to which a name is permanently linked and must consist of a gathering, or part of a gathering, of a single species or infraspecific taxon. However, sometimes more than one species may be mounted on a single herbarium sheet (‘Frankenstein type’), causing a nomenclatural confusion (admixture) in the use of a name, which creates difficulties on the delimitation of the species and application of its name (‘taxonomic headache’). This is the case of the assigned holotype to Mimosa asperoides Izag. & Beyhaut that is composed also of other Mimosa species. To solve the ‘Frankenstein type’ of Mimosa asperoides, we analyzed the original material, the protologue, and performed a morphological comparative analysis to distinguish the species involved in the admixture. These analyses allowed us to identify which of the fragments of the ‘Frankenstein type’ correspond to M. asperoides, enabling its lectotypification and providing new diagnostic features to amend its description. Finally, we reported the first citation of M. asperoides for Brazil and discussed the link between nomenclature and taxonomy to solve a taxonomic headache due to admixture.


Introduction
Names are what we use to communicate about organisms and by extension the rest of their biology (Knapp et al. 2004), thereby any change is disruptive to an enormous community outside of taxonomy (McNeill 2000).Regarding this, the nomenclature regulates how names are used to communicate taxonomic hypotheses and also establishes the rules (Codes) to ensure the least degree of ambiguity in their application (Thomson et al. 2018), avoiding scientific confusion (Turland et al. 2018).For example, the International Code of Nomenclature for algae, fungi, and plants (ICN) governs the scientific naming of all organisms traditionally treated as algae, fungi, or plants (Turland et al. 2018), being composed of a number of principles, rules and recommendations arranged in articles.
In the ICN, the application of names of taxa at the rank of family and below is determined by means of nomenclatural types, a process known as typification (Art.7.1, Turland et al. 2018).A nomenclatural type is the physical element to which the name of a taxon is permanently linked, whether

Materials and methods
To solve the 'Frankenstein type' of Mimosa asperoides, we analyzed the original material, the protologue, and conducted a morphological comparative analysis to distinguish the species involved in the admixture.Considering these analyses, we chose the fragment that represents M. asperoides, performing its lectotypification.We also proposed an amended description and reported the first citation of Mimosa asperoides for Brazil.
We consulted the original material of Mimosa asperoides by physical review or image analysis of specimens deposited in the Uruguayan herbaria, MVFA and MVM (acronyms according to Thiers 2023; Tab.1).From the MVM, we analyzed by image two paratypes; one of them lacks the diagnostic feature of this species, the fruit.The designated paratype collected by Osorio 688 (MVM No.13297) does not correspond to Mimosa asperoides due to the size and venation of leaflets and indumentum of branches; features that match with the description of Mimosa riverensis Izag.& Beyhaut.The exsiccate collected by Legrand 4990 (MVM s/n) corresponds to Mimosa asperoides.From the MVFA, we reviewed and photographed all the original material of M. asperoides (three exsiccates).This analysis showed that the designated paratype collected by Berro 5528 is free of admixture.Nevertheless, the designated holotype (MVFA barcode 0000095) and the isotype (MVFA barcode 0000096) consist in two different species mounted on the same sheet, M. asperoides and M. axillarioides.
To identify other potential diagnostic features of M. asperoides beyond the fruit shape, we analyzed separately 17 features related to leaf, flower and fruit that were available in the original material deposited on MVFA herbarium.We also examined specimens available from Brazilian collections (HUCS, ICN, MBM, MPUC, and PACA, herbaria acronyms according to Thiers 2023).These features were: arrangement of leaves on secondary branches; leaf length; leaflet number, indumentum, margin, nervures, shape, and size; indumentum and length of peduncles; calyx and corolla length; calyx proportion to corolla; calyx and corolla shape; corolla and fruit indumentum.We also consulted the protologue of M. axillarioides (in Izaguirre & Beyhaut 2009) and analyzed its original material deposited in the MVFA (MVFA barcode 26745 holotype!, Rosengurt B-1028 paratype!), photographing its leaflets, flower, indumentum of peduncles and fruits.Then, we summarized the data of Box 1. Solving a 'Frankenstein type': nomenclatural and taxonomic approaches to deal with admixture.Typification is the main nomenclatural act to solve a nomenclatural admixture.According to Shenzhen Code (Turland et al. 2018), if a type is found to belong to more than one taxon, the name must remain attached to the part (specimen as defined in Art.8.2) that corresponds most nearly to the original description or diagnosis (Art.9.14) by the designation of a lectotype (Art.9.3 and Art.9.11).
When the original description/diagnosis is itself taxonomically mixed, as in the case of Mimosa asperoides, the use of the protologue to guide the lectotypification may be compromised.A solution would be to conduct a comparative taxonomically analysis of the fragments to find features that underline the differences among the species involved in the admixture.This analysis may result in an amended circumscription (emend., Rec.47A.1), in which it is possible to perform the corrections and to include relevant information previously not cited in the description, such as new diagnostic features.If after this analysis, the original material is demonstrably ambiguous and cannot be critically identified for purposes of the precise application of the name to a taxon, an epitype should be proposed under the Art.9.9 to help the interpretation of that name; considering at least if the specimen has a similar locality, habitat, and morphological details to those mentioned in the protologue (Turland 2019;Lendemer 2020).On the other hand, if the lectotype is unambiguous, but incomplete, it is not appropriate to designate an epitype (see Box 1, scheme below).
diagnostic and overlapping morphological features between M. asperoides and M. axillarioides in a comparative table and in three illustrative figures with morphological details.Afterwards, these data and analyses were used to support which fragments correspond to M. asperoides, allowing its lectotypification based on the Articles 9.11 and 9.14 of the Shenzhen Code (Turland et al. 2018).Furthermore, these data also supported the amended circumscription of M. asperoides.The terminology of the description was adapted from Barneby (1991), Beentje (2010) and, for the trichomes, from Jordão et al. (2020).Finally, we reported the first citation of M. asperoides in Brazil by the analysis of recent collections and the review of national and international herbaria (BHCB, BLA, BOTU, CESJ, CORD, CRI, CTES, E, ESA, F, FURB, HAS, HDCF, HPBR, HUCS, HUEM, HURG, HVAT, IAC, ICN, JOI, K, MBM, MO, MPUC, MVM, NY, P, PACA, PEL, R, RB, SJRP, SMDB, SPSF, and USZ, acronyms according to Thiers 2023).We provided a distribution map, first field images of the species, and a comparative analysis between Brazilian and Uruguayan populations regarding their habitat and flowering phenology.

Identification tags highlighted the historical nomenclatural and taxonomic confusion
The analysis of identification tags on the original material of M. asperoides illustrates the taxonomic confusion involving its type specimens, indicating the presence of admixture (Fig. 1).The designated holotype of M. asperoides was collected by Mariano B. Berro in 1899 in Uruguay and firstly identified as Mimosa marginata Lindl., one of the synonyms of M. schleidenii Herter.In 1948, Arturo Burkart suggested that this specimen was a putative hybrid between M. axillaris Benth.and M. lindleyi Burkart, this last one being another synonym of M. schleidenii.Thus, annotations made by Burkart on the original material in 1948 provided the historical evidence that these exsiccates contain two distinct morphotypes related to two Mimosa species, helping us to detect and fix the admixture problem.
However, these two morphotypes identified in the original material of M. asperoides do not correspond to any of the species highlighted by Burkart in 1948, M. axillaris and M. schleidenii.In fact, they correspond to the two new species described by Izaguirre and Beyhaut (2009), M. asperoides and M. axillarioides.Both species suggested by Burkart to compose the admixture are morphologically distinct from M. asperoides, mainly by the articulate fruits, and from M. axillarioides by indumentum or flower features.For example, M. axillaris differs from both species of the admixture by its subglobose to ellipsoid inflorescences, antrorse-strigose indumentum of branches and peduncles, and longer leaves and petioles.While M. schleidenii differs by having a pappiform calyx, leaflets glabrous on both faces, and a corolla completely sericeous.

Looking for new diagnostic features to solve the admixture
The morphological analysis of the original material and the discovery of Brazilian populations (Tab.2) helped us to support which fragment corresponds to M. asperoides and which corresponds to M. axillarioides, enabling the lectotypification of M. asperoides.In addition, the comparative analysis highlighted other five diagnostic features to better circumscribe M. asperoides.These features were: the shape and size of leaflets (Fig. 2D, E, F, Tab.2); the arrangement of leaves on the secondary branches (Fig. 3F); the indumentum of peduncles (Fig. 3H) and indumentum of fruits (Fig. 3 J, K).However, we noticed that most of the features analyzed overlapped with M. axillarioides (71%), hampering their distinction due to high variance among M. asperoides populations or by sharing similar morphology (Tab.2).
For example, the shape and length of calyx were highly variable between Brazilian (BRA) and Uruguayan (URY) populations of M. asperoides, which has compromised their use as diagnostic features.While M. axillarioides has a laciniate calyx covering half of corolla length (Fig. 3A),   in M. asperoides it covers the half (URY, Fig. 3B, C) or more than half of corolla's length (BRA, Fig. 3D, E), being more laciniate (URY, Fig. 4B, C) or less laciniate (BRA), so the rim is fimbriate (BRA, Fig. 3E).Furthermore, this variation was also noticed regarding the phenological state and age of the specimens analyzed.The calyx is less laciniate and longer in flower buds and young inflorescences (Fig. 3D, E), while it is more laciniate and shorter in old inflorescences (Fig. 3B, C).Nevertheless, the relationship between calyx shape and inflorescence age merits further investigations beyond of the goals of the present study.
Another example of morphological overlap between the species is the indumentum of flowers.Both species have two types of indumentum in corolla, one antrorse in the apex of lobes and other retrorse covering the lobes besides tubular corolla and paleaceous-laciniate calyx (Fig. 3A-E, Tab. 2).Moreover, both species are covered by setiformbarbellate trichomes, mainly in peduncles and fruits (Fig. 3. I-K).However, despite being classified as the same type of trichome, this structure seems to differ regarding robustness, width, and length.Those micromorphological differences remarked for the trichomes should be Acta Botanica Brasilica, 2023, 37: e20220313  Acta Botanica Brasilica, 2023, 37: e20220313 investigated concerning an anatomical approach to allow a better delimitation of the species based on trichomes.In addition, features such as the leaf length and the number of leaflets may be of interest to delimit M. asperoides despite the overlapping measures (in Tab. 2), if more material would be available, and if analyzed considering a morphometric framework.Mimosa asperoides differs from M. axillarioides by the arrangement of leaves on secondary branches with several congested leaves per branch; the shape and size of the leaflets, which are narrowly oblong from asymmetric base and obtuse apex with 2-5 × 0.5-0.9(1.3) mm; peduncles covered by a retrorse-strigose setiform-barbellate trichomes, and a craspedium non-articulate, oblong, inflated, with valvar dehiscence, covered by retrorse-strigose and setiformbarbellate trichomes.
Acta Botanica Brasilica, 2023, 37: e20220313 The first record of M. asperoides for Brazil highlighted the admixture and improved its description The nomenclatural admixture was only detectable due to the discovery of Brazilian populations of M. asperoides.Firstly, the Brazilian collections were considered a probably new species morphologically similar to M. asperoides.However, the analysis of MFVA herbarium allowed us to confirm its occurrence in Brazil and to detect the admixture in the original material, enlightening the circumscription of M. asperoides.For example, beyond the new diagnostic features, we noticed that this species has: (1) nodules on the roots; (2) paraphyllidia; (3) spicule; (4) substipited ovary; (5) shortly monadelphous stamens at base of ovary; (6) variable flower features according to development stage; (7) two flowering periods.
In Brazil, M. asperoides is found in the remaining native vegetation of four localities of the coastal plain of Rio Grande do Sul (RS) State (32°S 1' 60", 52°W 5' 55"), growing on sandy soils and wet grasslands (Fig. 5 and 6).Most recent records were only reported from Brazil, because this species has not been collected in Uruguay for almost 70 years.Brazilian populations of this species showed differences mainly regarding its phenology and habitat beyond disjoint distribution.For example, Brazilian populations habit sandy soils in coastal plain, and they bloom twice (during April-June and August-September), having fruits at November-January.While Uruguayan populations occur on hills and prairies slopes, having only one and shorter period of flowering and fructification (November to March).Furthermore, Brazilian populations occur in areas of intense land conversion (e.g.urban expansion, agriculture and silviculture), being limited to a few individuals and small patches.Regarding this, urgent efforts are required to: (1) map all its populations, (2) evaluate its conservation status, and (3) elaborate a conservation plan to avoid its extinction in the field.

Discussion
Nomenclature rules are independent from taxonomic methods (Turland 2019).However, these disciplines are subordinated to each other regarding the typification of a name.While taxonomy provides the methods and concepts to delimit a taxon, nomenclature determines the rules that fix and link a taxon to a name.A taxonomic confusion in species delimitation will imply in instability on naming a species, and on nomenclatural problems.For example, a delimitation based only on a unique or high variable feature or mixed material can difficult the identification of a taxon, and prompt a taxonomic confusion when this feature is not available or; when used to distinguish species morphologically similar, to result in misidentification.Thereafter, if not solved, the taxonomic confusion will generate, such as a domino effect, difficulties in the species identification and application of a name in taxonomy and outside of taxonomy A nomenclatural admixture ('Frankenstein type') is an example of a taxonomic confusion in species delimitation that may be generated by the superficial similarity between the taxa, the little progress in a particular taxonomic group (Sennikov & Tikhomirov 2019), besides the simultaneous collection of a species in sympatry and the absence of field notes to distinguish them (Brummitt et al. 2004).In the case of the 'Frankenstein type' of M. asperoides, the phenological stages among the fragments contributed to the mixed material, because while M. asperoides presented Acta Botanica Brasilica, 2023, 37: e20220313 very distinctive fruits, M. axillarioides was in flowering or with immature fruits that have similar trichomes, thereby being confused with M. asperoides.The lack of the fruit may also explain why Izaguirre and Beyhaut (2009) did not consider the exsiccate from Brazil deposited in MVFA herbarium (MVFA 0000117) belong to this species.We also hypothesize that the morphological similarities, availability of only herbaria material instead of recent collections and fieldwork may have contributed to the admixture because if the authors had have observed this species in field, they may have highlighted previously the fragments of M. axillarioides, avoiding admixture in the original material of M. asperoides.
A 'Frankenstein type' is not restricted to Mimosa asperoides, it has been reported for different groups of plants and fungi, mainly for the Linnean species (Sennikov & Tikhomirov 2019).For example, the revised lectotypifications of Lycopus europaeus L. (Lamiaceae) and Daucus mauritanicus L. (Apiaceae) were needed due to their current lectotypes consisting of plant fragments representing two different species (Sennikov & Melnikov 2018;Martínez-Flores & Crespo 2019).These examples show that admixture seems to be a common taxonomic headache, although it has been underestimated and reported, being an interesting topic to be investigated regarding its causes and consequences.Other cases of 'Frankenstein types' are more complex and involve mixed material in the protologue and/or illustrations (e.g.Wu et al. 2012;Crespo et al. 2019) In the case of Mimosa asperoides, we perfomed a morphological comparative analysis to find differences among the species involved in the admixture, highlighting new morphological diagnostic features to this species.Most of these new diagnostic features helped to distinguish M. asperoides when the fruit was lacking, such as the leaflets shape and the kind of indumentum of peduncles, features that have been successfully used to delimit other species in the Mimosa genus.For example, leaf morphology was a good feature to distinguish species from Mimosa subser.Dolentes-Brevipedes (Morales et al. 2020) and M. ser.Quadrivalves (Flores-Cruz et al. 2004).Furthermore, the indumentum has been considered a relevant feature to describe new species in Mimosa (e.g.Atahuachi & Hughes 2006;Santos-Silva & Tozzi 2012;Dutra & Garcia 2013;Jordão et al. 2014;Schmidt-Silveira et al. 2019).We would also benefit from other source of evidence such as DNA, morphometry, and cytogenetic data to support the species hypothesis in the perspective of the integrative taxonomy (Dayrat 2005;Pante et al. 2015) to solve a 'Frankenstein type'.However, in the case of the Mimosa asperoides, we had limited data, being the morphological approach fundamental to detect and to solve the admixture.
Here, we have illustrated that a nomenclatural problem can be a raw material to develop and improve our understanding of plant taxonomy.We exemplified the 'taxonomic headache' due to a 'Frankenstein type' and solved the admixture on the original material of Mimosa asperoides.Firstly, the discovery of new records in Brazil, the review of original material and the protologue, besides the identification tags, helped us to detect the admixture.Secondly, we found more diagnostic features to distinguish the species involved on the admixture by a careful morphological analysis that allowed the lectotypification of Mimosa asperoides and its amended circumscription.Thirdly, we extended the occurrence of Mimosa asperoides in South America, reporting its first citation for Brazil and ecological differences among the populations.Finally, we discussed the implications of a taxonomic confusion for naming and delimiting species, probably reasons that may prompt a 'Frankenstein type' and the role of morphology to clarify the M. asperoides identity.

Figure 1 .
Figure 1.Detail of the identification tag of M. asperoides Izag.& Beyhaut isotype, illustrating the historical evidence of taxonomic confusion caused by specimen admixture; which was highlighted by Burkart in 1948, when he considered this specimen as a putative hybrid (Photo from F. Schmidt-Silveira).
Hispid with setiform-barbellate trichomes Retrorse-strigose with setiform-barbellate trichomes Diagnostic * Features based on description of Izaguirre & Behyaut (2009) and the holotype of M. axillarioides.# Diagnostic feature highlighted in the protologue.§ Data from the analysis of herbaria exsiccates (holotype and paratype of MVFA) and Brazilian collections (see Materials and methods for more details).

Figure 2 .
Figure 2. Comparative morphological analysis of Mimosa asperoides and M. axillarioides leaflets, showing that some leaflets from the fragments of the designated holotype and isotype of M. asperoides (A, B) have consistent morphology with M. axillarioides (C); while other leaflets are indeed from M. asperoides (D) when comparing its morphology with non-mixed material from the M. asperoides paratype (E) and the Brazilian collections (F). A. Leaflets from a fragment of the designated holotype of M. asperoides (MVFA barcode 0000095), whose morphology corresponds to M. axillarioides.B. Leaflets from a fragment of M. asperoides isotype (MVFA barcode 0000096) that correspond to M. axillarioides.C. Leaflets from the holotype of M. axillarioides (MVFA barcode 26745).D. Leaflets from a fragment of M. asperoides holotype (MVFA barcode 0000095), whose morphology corresponds to M. asperoides.E. Leaflets of M. asperoides paratype (Berro 5528).F. Leaflets of M. asperoides from Brazil (F.Schmidt-Silveira, 1094).Upper leaflets are from adaxial face, being usually nerveless and brighter.Lower leaflets are from abaxial face, usually 1-3 nerved and non-brighter.Scale bars of 1 mm.

Figure 3 .
Figure 3. Comparative morphological details between Mimosa asperoides and M. axillarioides, showing that the morphology and indumentum of calyx and corolla are similar in both species (A, B, C, D, E).However, they differ by the indumentum of peduncles (G, H); the arrangement of leaves on branches (F, L), by fruit type, and indumentum (I, J, K).M. asperoides: B, C, D, E, F, H, J, K, M. axillarioides: A, G, I, L. (Photos A, G, I, L from M. axillarioides holotype; Photo B from a fragment of M. asperoides holotype; Photo C from M. asperoides paratype; Photos D, E, F, H, I, J, K from F. Schmidt-Silveira, 1094).Scale bars of 0.5 mm (A-E & H) and 1mm (F, G, I-L).

Figure 4 .
Figure 4. Designated holotype of Mimosa asperoides Izag.& Beyhaut and details of admixture (mixed collection): the 'Frankenstein type'.This original material is composed of fragments of M. asperoides (A dotted & C, here lectotypified) that presents its mainly diagnostic feature, the fruit; and M. axillarioides Izag.& Beyhaut (A not dotted & B), which does not present this feature.(Photos from F. Schmidt-Silveira).

Figure 5 .
Figure 5. Current distribution of Mimosa asperoides in South America, considering the new records for Brazil, in Rio Grande do Sul State.

Figure 6 .
Figure 6.Mimosa asperoides details. A. In sandy soils of Coastal Plain in Rio Grande do Sul State, Brazil.B. Inflorescences.C. Fruits.(Photos from S. Bordignon).

Table 1 .
Review of original material of Mimosa asperoides and M. axillarioides highlighting which of them is composed of admixture.

Table 2 .
Comparative morphological analysis between Mimosa axillarioides and M. asperoides, indicating the diagnostic features of M. asperoides and overlapping features with M. axillarioides.