Nomenclature Survey of the Genus Amaranthus (Amaranthaceae): 12 Questions about Amaranthus caudatus

Nomenclatural and taxonomic issues concerning Amaranthus caudatus and the related taxa are presented. Types are designated for names A. caudatus var. albiflorus (neotype at RO), A. caudatus var. atropurpureus (neotype at GH), A. caudatus var. gibbosus (neotype at RO), A. dussi (neotype at NAP), and A. edulis (lectotype at LP). Holotypes are indicated for the names A. caudatus var. pseudopaniculatus f. oblongipetalus (EA), A. caudatus var. pseudopaniculatus f. pseudopaniculatus (EA), A. caudatus subsp. saueri (PR), and Amaranthus edulis var. spadiceus (CORD). The names A. caudatus var. albiflorus, A. caudatus var. atropurpureus, A. caudatus subsp. saueri, A. dussi, and Amaranthus edulis var. spadiceus are considered as hererotypic synonyms of A. caudatus. On the basis of morphological, cytological, and molecular data, the taxa caudatus, mantegazzianus, and gibbosus are here proposed to be treated as different species. A new name—Amaranthus baileyanus—is proposed for A. caudatus var. gibbosus because of a previous and validly published Amaranthus gibbosus.


Introduction
Amaranthus L. (Amaranthaceae Juss.) is a genus comprising 70-75 species, of which approximately half are native to the Americas [1,2]. Several American species are used as ornamentals, food, and medicines, and some of them are able to escape from cultivation, mainly impacting agricultural systems economically with reductions in productivity and crop quality [1][2][3][4].
Amaranthus is a critical genus from a taxonomical point of view because of its high phenotypic variability, which led to nomenclatural disorders and misapplication of names [1,2,5,6]. No comprehensive molecular study has been published at present yet, and, on the basis of the more recent classification [5], three subgenera were recognized: subgenus Acnida (L.) Aellen ex K.R. Robertson with three sections, subgenus Albersia (Kunth) Gren. & Godr. with four sections, and subgenus Amaranthus, with three sections and two subsections. Note, however, that the most recent molecular investigation [7] showed that the classification proposed by Mosyakin and Robertson [5] cannot be retained at the current state of knowledge.

Material and Methods
This work is based on field surveys, analysis of relevant literature (protologues are included), and checking/examination of specimens preserved in the following herbaria:

Material and Methods
This work is based on field surveys, analysis of relevant literature (protologues are included), and checking/examination of specimens preserved in the following herbaria: BH, BM, BR, CFL, CORD, EA, FI, GH, HAL, HOH, K, LINN, LP, MEL, M, MO, NAP, NY, P, PH, RO, SI, UCBD, and US (acronyms according to THIERS [19]).
The articles cited throughout the text are referred to the Shenzhen Code (hereafter reported as "ICN" [20]).

Nomenclatural Notes
3.1.1. Amaranthus caudatus subsp. saueri Amaranthus caudatus subsp. saueri was described by Jehlik in 1990 [21] to distinguish forms characterized in having rose-or white-coloured seeds with an obtuse margin (diagnosis: "Semina rosacea usque fere albicantia, margine obtusa"), whereas the autonymic species was recognized in displaying dark seeds. The holotype is preserved at PR (barcode PR615740). I had the opportunity to examine this specimen ( Figure 1) and observed the colour of the seeds, which varies from rose (light brownish in exsiccatum) to white. However, the colour of the seeds cannot be considered at present as a character which allows to distinguish infraspecific ranks and this variation in colour is currently included in the variability of A. caudatus as reported by several authors (see, e.g., [1,2,6,22,23]). A. caudatus subsp. saueri does not deserve to be considered as a separate taxon from A. caudatus s.lat. and it is a synonym of A caudatus s.s., having the pendulous terminal florescence.

Amaranthus caudatus var. alopecurus
Amaranthus caudatus var. alopecurus was validly published by Moquin-Tandon as part of his treatment of Amaranthus in Candolle's Prodromus [24] (p. 256). The lectotypification   [6] and Iamonico [12]. Both of the lectotype designations refer to the same specimen deposited at P (barcode P00482809). Isolectotypes were also listed by both of these authors on specimens preserved at GH (barcode GH00037040), HOH (barcode HOH009263), and MO (barcode MO357985). Furthermore, an isolectotype was also reported at BR (barcode BR000832631) and HAL (barcode HAL0110480) by Iamonico [12], or at K (barcode K000223569) by Bajón [6]. Since Bayón's paper was published before Iamonico's one (29 December 2015vs. 7 September 2016, the former designation [6] must be followed according to Art. 9.19 of ICN. Finally, further isolectotypes were found during the present research: they are preserved at K (barcode K000223570) and MEL (barcode MEL2459428). According to the treatment proposed in the present paper, Amaranthus caudatus var. alopecurus is considered as a synonym of A. caudatus s.s., having the pendulous terminal florescence.

Amaranthus caudatus var. pseudopaniculatus
Suessenguth [25] (p. 71) proposed to describe the var. pseudopaniculatus to distinguish plants of Amaranthus caudatus with shortly aristate tepals and highly dense branches of the synflorescence; a collection was also cited ("Tanganyika-Territor., Amani, leg. GREENWAY nr. 993 (Herb. Nairobi)"). A new f. oblongipetalus Suesseng. (reported "oblongopetalus", here corrected according to Art. 60.10 of ICN) was also described (according to Art. 26.3 of ICN the f. caudatus was automatically established) by a short diagnosis ("Tepala oblonga vel anguste oblonga") and citing the following collection: "Tanganyika-Territor., Amani 2900 ft., leg. GREENWAY nr. 6155 (Herb. Nairobi)". Townsend [26] (p. 26) listed a specimen deposited at EA (acronym of the herbarium of the National Museums of Kenya which corresponds to "Herb. Nairobi" as reported by Suessenguth and Merxmüller [25]) as the holotype of the var. pseudopaniculatus, and a specimen at K as the isotype of the f. oblongipetalus ("Type of var.: Tanzania, Lushoto District, Amani, Greeway 993 (EA, holo.!) of forma: Tanzania, Lushoto District, Amani, Greeway 6155 (K, iso.!)"). I traced the following three specimens: (1) Greeway's specimen no. 6154 (herbarium EA; collection number is indicated in the label at the base of the plant Based on the protologue, Suessenguth [25] clearly indicated for var. pseudopaniculatus s.s. and var. pseudopaniculatus f. oblongipetalus both the number of collections and the herbarium in which they were deposited. I here considered this quotation as an indication of holotypes. Townsend [26] correctly stated that EA no. 6154 is the holotype of the var. pseudopaniculatus s.s. (Figure 2), whereas the K specimen is the istoype of var. pseudopaniculatus f. oblongipetalus because of, as discussed above, the occurrence of the printed label "FROM THE HERBARIUM OF THE EAST AFRICAN RESEARCH INSTITUTE, AMANI". I traced the holotype of the f. oblongipetalus, cited by Suessenguth [25] as the specimen EA no. 6155 ( Figure 3).      Concerning the original material used by Bailey [27,28] to describe these three varieties, he did not mention any herbarium in which specimens could be deposited. Stafleu and Cowan [29]  . This BH exsiccatum can be identified as Amaranthus caudatus (see, e.g., [1,2,6,22,23]) but it cannot be referred to the var. gibbosus on the basis of the protologue [28], since it displays continuous synflorescences (not interrupted as indicated in the diagnosis of the var. gibbosus). As a consequence, it cannot be considered for the lectotypification purpose of the var. gibbosus. No further original material was traced for Bayley taxa and, as a consequence, neotypifications are required under Art. 9.8 of ICN as follows: (1) Amaranthus atropurpureus: since the colour of the leaves often change after exsiccation of amaranths and the diagnostic character of this variety is "Foliage blood-red" [27,28], the designation of a neotype was not simple since colours of amaranths usually change during the drying process. So, a coloured illustration (e.g., no. 227 published by Step & Bois [30]) would be desirable. Fortunately, I found just one specimen at GH (GH01928945) bearing the terminal part of a plant of A. caudatus with two leaves, of which one is clearly red-coloured. This plant was collected in America. GH01928945 is here designated as the neotype of Amaranthus atropurpureus. (2) Amaranthus caudatus var. albiflorus: the diagnostic characteristic given by Bailey [27] (p. 270), i.e., the colour of the flowers ("Spikes white or greenish white"), is very difficult to verify in specimens. In fact, as we know, colours of amaranths change after the exsiccation. I here designate a specimen preserved at RO ( Figure 4) which was identified as "Amaranthus caudatus var. albiflorus" by Alfredo Cacciato, who was an expert of the genus Amaranthus in Italy in the 1970s. I studied most of Cacciato's exsiccata during the last 15 years and I am sure that he referred to plants having white flowers (see e.g., [2,10]). (3) Amaranthus caudatus var. gibbosus: I tried to find a specimen collected in America (the native area of A. caudatus) whose morphology matches Bailey's concept. Unfortunately, no specimen was found at either the main American herbaria (e.g., NY, PH, and US) or in some important European ones (e.g., BM, K, and P). Therefore, I was forced to choose from my own recent collection in Serbia (Eastern Europe) ( Figure 5).
According to the treatment proposed in the present paper, (1) Amaranthus atropurpureus: since the colour of the leaves often change after exsiccation of amaranths and the diagnostic character of this variety is "Foliage blood-red [27,28], the designation of a neotype was not simple since colours of amaranths usu ally change during the drying process. So, a coloured illustration (e.g. no. 227 pub lished by Step & Bois [30]) would be desirable. Fortunately, I found just one specimen at GH (GH01928945) bearing the terminal part of a plant of A. caudatus with two leaves, of which one is clearly red-coloured. This plant was collected in America GH01928945 is here designated as the neotype of Amaranthus atropurpureus. (2) Amaranthus caudatus var. albiflorus: the diagnostic characteristic given by Bailey [27 (p. 270), i.e., the colour of the flowers ("Spikes white or greenish white"), is very dif ficult to verify in specimens. In fact, as we know, colours of amaranths change afte the exsiccation. I here designate a specimen preserved at RO (Figure 4) (6)

Amaranthus dussii
Amaranthus dussi was honoured by Sprenger [36] (p. 178) to French Father Dussi who lived in Martinique and often sent plants to C. Sprenger; a description was given on the basis of plants growing in the Botanical Garden of Naples (Southern Italy) from seeds collected in Martinique (Lesser Antilles).
Carl Ludwig Sprenger was a German botanist (30 November 1846-13 December 1917) who lived in Naples from 1877 to 1907 where he was partner in the horticultural house of Damman & Co. of San Giovanni Testuccio (a district of the eastern area of Naples city). Sprenger collected many seeds and prepared hundred specimens which, however, were destroyed after the eruption of Vesuvius on 4 April 1906 [37] (p. 268). Original material for Amaranthus dussi is not, therefore, in extant and, according to Art. 9.8 of ICN, a neotypification is required. On the basis of the original description [36] (p. 178), A. dussi displays synflorescence with "fiori riuniti in lunghe e grosse spighe conglomerate prime erette e poi elegantemente riflesse e pendule" (="flowers arranged in long and big spikes ammassed before erect, then stylishly reflexed and pendulous"). This trait is typical of just one Amaranthus species, i.e., A. caudatus [1,2,23]. I here propose a specimen preserved at NAP (barcode NAP0000610), collected in Naples Province, as the neotype of the name Amaranthus dussi (Figure 6). Based on Sprenger's description [36] (p. 178), and the treatment here proposed, A. dussi is to be considered as a synonym of A. caudatus. Spegazzini [38] (p. 163) provided a detailed diagnosis and description, as well a provenance ("Hab. Cultivado en la región árida y montañosa de la provincia de Salta la población indigena"). Bayón [6] (p. 276) indicated the holotype for this name ("T cultivado en La Plata, s.f., C. L. Spegazzini s.n. (holotipo, LPS-12665 en LP-16325!)"). H ever, first, no holotype was cited in the protologue and a lectotypification is neces according to the Arts. 9.1, 9.3, and 9.4 of ICN (see also the considerations by McNeill [ So, Bayón's holotype's indication has to be considered as a lectotype. However, accor to Art. 7.11 of ICN, "designation of a type is achieved only if… on or after 1 January 2 if the typification statement includes the phrase "designated here" (hic designatus) o equivalent". Since this phrase was not reported by Bayón [6] (p. 276), his typificati not valid. I here designate the specimen LP12665 (cited by Bayón [6] (p. 276)) as the l type of the name Amaranthus edulis (Figure 7).
The lectotype at LP is identifiable as Amaranthus caudatus s.lat. on the basis o shape of the tepals which are spatulate with obtuse apexes and, according to the treatm proposed in the present paper, as A. mantegazzianus Passer., having erect termina
Spegazzini [38] (p. 163) provided a detailed diagnosis and description, as well as the provenance ("Hab. Cultivado en la región árida y montañosa de la provincia de Salta por la población indigena"). Bayón [6] (p. 276) indicated the holotype for this name ("TIPO: cultivado en La Plata, s.f., C. L. Spegazzini s.n. (holotipo, LPS-12665 en LP-16325!)"). However, first, no holotype was cited in the protologue and a lectotypification is necessary according to the Arts. 9.1, 9.3, and 9.4 of ICN (see also the considerations by McNeill [39]). So, Bayón's holotype's indication has to be considered as a lectotype. However, according to Art. 7.11 of ICN, "designation of a type is achieved only if . . . on or after 1 January 2001, if the typification statement includes the phrase "designated here" (hic designatus) or an equivalent". Since this phrase was not reported by Bayón [6] (p. 276), his typification is not valid. I here designate the specimen LP12665 (cited by Bayón [6] (p. 276)) as the lectotype of the name Amaranthus edulis (Figure 7).  (Figure 8). CORD specimen is identifiable as A. caudatus s.lat. on the basis of the shape of the te spatulate with obtuse apexes, and according to the treatment proposed in the presen per, as A. mantegazzianus Pass., having erect terminal florescence. The diagnostic ch teristics given in the protologue (seed colour, length, and structure of the bracts) ha taxonomic value (see, e.g., [1,2,6,22,23]), and this variety name is synonymized wi mantegazzianus. The lectotype at LP is identifiable as Amaranthus caudatus s.lat. on the basis of the shape of the tepals which are spatulate with obtuse apexes and, according to the treatment proposed in the present paper, as A. mantegazzianus Passer., having erect terminal florescence.

Amaranthus edulis var. spadiceus
The var. spadiceus was proposed by Hunziker [40] T. H.) . . . "). The above cited collection was found at CORD, where Hunziker's collection is preserved, and it is the holotype (Figure 8). This CORD specimen is identifiable as A. caudatus s.lat. on the basis of the shape of the tepals, spatulate with obtuse apexes, and according to the treatment proposed in the present paper, as A. mantegazzianus Pass., having erect terminal florescence. The diagnostic characteristics given in the protologue (seed colour, length, and structure of the bracts) has no taxonomic value (see, e.g., [1,2,6,22,23]), and this variety name is synonymized with A. mantegazzianus.
Hunziker [40] (p. 330) designated a neotype for Passerini's name on a specimen lected in Salta Province (CORD00002607; Figure 9); isoneotypes (at K, SI, and US) also reported. Since Hunziker [40] (p. 330) did not cite the herbarium Parma, where serini's collection is preserved, I tried to check this herbarium, but unfortunately no inal material was traced (R. Brusi pers. Comm.). As a consequence, Hunziker's cho correct, and it is to be accepted.
Hunziker [40] (p. 330) designated a neotype for Passerini's name on a specimen collected in Salta Province (CORD00002607; Figure 9); isoneotypes (at K, SI, and US) were also reported. Since Hunziker [40] (p. 330) did not cite the herbarium Parma, where Passerini's collection is preserved, I tried to check this herbarium, but unfortunately no original material was traced (R. Brusi pers. Comm.). As a consequence, Hunziker's choice is correct, and it is to be accepted.
[21] (p. 110 in Table 5), is to be considered as a nomen nudum, and, therefore, in according to Arts. 38.1 and 38.2 of ICN.

Taxonomic Notes
Amaranthus caudatus was validly published in the first edition of Species Plant [42] (p. 990) and correctly typified on a Linnaean specimen (Herb. Linn. 1117.26) by To send [43] (p. 10). This species is currently accepted by the scientific community, a morphologically differs from the other monoecious Amaranthus taxa by the following ual characteristics: terminal, lax, pendulous (especially the terminal one), erect, or ding, and very long (up to 80 cm) often red or purple synflorescences; five spatulatevate tepals, equal or subequal to the bracts; and dehiscent fruit.

Conclusions
On the basis of morphological, cytological, and molecular data, the taxa caud

Illegitimate and Invalid Names
The names Amaranthus pendulinus and A. pendulus were reported by Moquin-Tandon [24] (p. 256) in Condolle's Prodromus as synonyms of A. caudatus var. albiflorus. These two names were not validly published according to Art. 36.1a of ICN.
Bailey [31] (p. 252) listed the name "Amaranthus abyssinica" as synonym of A. caudatus. According to Art. 36.1a of ICN, Bailey's name is not validly published.

Taxonomic Notes
Amaranthus caudatus was validly published in the first edition of Species Plantarum [42] (p. 990) and correctly typified on a Linnaean specimen (Herb. Linn. 1117.26) by Townsend [43] (p. 10). This species is currently accepted by the scientific community, and it morphologically differs from the other monoecious Amaranthus taxa by the following sexual characteristics: terminal, lax, pendulous (especially the terminal one), erect, or nodding, and very long (up to 80 cm) often red or purple synflorescences; five spatulate-obovate tepals, equal or subequal to the bracts; and dehiscent fruit.

Conclusions
On the basis of morphological, cytological, and molecular data, the taxa caudatus, mantegazzianus, and gibbosus deserve to be treated as separate species, as proposed below. A new combination would be necessary for Bailey's var. gibbosus. However, note that an Amaranthus gibbosus was already and validly published by Bailey [27] (pp. 55-56) (diagnosis: "pigweed and beet-roots"), and a new combination of the var. gibbosus by Native distribution area. The origin of Amaranthus caudatus remains uncertain at the current state of knowledge. According to several authors (e.g., [1,2,23,59]), this species most likely originated in South America (Argentina, Equador, Perù, and Bolivia) by domestication and crossing with the wild A. quitensis Kunth.
Current distribution area. According to the current available data, Amaranthus caudatus would occur currently as alien species in Asia [60], Australia [59,61], Europe [62], and Africa [63]. However, it cannot be possible, at present, to confirm the occurrence of this species at the national level for the following reasons: Neotype (designated by Hunziker [64]  Native distribution area. Unknown, but likely South America (Argentina). Current distribution area. The holotypes of Amaranthus edulis var. pseudopaniculatus (both f. pseudopaniculatus and f. oblongipetalus) came from Tanzania, whereas the specimens below listed were from Ethiopia (they are the types of A. caudatus var. alopecurus Moq., which was considered by Iamonico [12] as a synonym of A. caudatus s.s.)). I here consider A. mantegazzianum as a probably alien species (casual) for Africa. Further investigations will be necessary to give a distribution of Amaranthus mantegazzianus out of its native range.