A plant that Linnaeus forgot: taxonomic revision of Rhodalsine (Caryophyllaceae)

Abstract: The mainly Mediterranean genus Rhodalsine (Caryophyllaceae) is revised and a single species, R. geniculata, is recognized, distributed from the Canary Islands in the west to Somalia in the east. The history of the taxon, which was known already during the 17th century but entirely overlooked by Linnaeus, is outlined. Variation and taxonomy are discussed and illustrations and a distribution map are provided. Many names are placed in synonymy and most of the names are typified, including six lectotypes designated here. Citation: Kool A. & Thulin M. 2017: A plant that Linnaeus forgot: taxonomic revision of Rhodalsine (Caryophyllaceae). — Willdenowia 47: 317–323. doi: https://doi.org/10.3372/wi.47.47313 Version of record first published online on 20 November 2017 ahead of inclusion in December 2017 issue.


Introduction
According to the results presented in several recent phylogenetic studies (for a summary, see Kool & Thulin 2017), there is good evidence for the view that Minuartia subg. Rhodalsine (J. Gay) Graebn. is a member of the tribe Sperguleae and sister to Spergula L. and Spergu laria (Pers.) J. Presl & C. Presl together. This mainly Mediterranean subgenus therefore needs to be removed from the polyphyletic Minuartia L. (Dillenberger & Kadereit 2014) and treated as a genus of its own. No comprehensive overview of the group has been presented since Williams (1898), who treated it as the genus Rhodalsine J. Gay with two species and several varieties. Since then many other names have been associated with the group and its now fairly complex taxonomy and nomenclature is the subject of the present paper.
The first illustrations of members of Rhodalsine were published by Boccone (1674), as "Alsine maritima longius radicata, Herniariae foliis", from Sicily. Two forms were described and illustrated, one main form with leaves of the same shape and size as Herniaria L., and one more narrow-leaved form called "Alsine maritima altera angustis foliis". Material from Sicily is preserved in the Boccone herbarium in P (Bonnet 1883).
Despite Boccone's illustrations and the mention of the plant many times in the pre-Linnaean literature (e.g. Morison 1680: 552), it was completely overlooked by Linnaeus, presumably as he did not have access to any herbarium material and therefore could not study it properly. Forsskål collected a member of Rhodalsine in Egypt 1761 -1762, but misidentified it as Cherleria sedoides L. (Forsskål 1775) and his material is now lost (Christensen 1922;Hepper & Friis 1994). Not until Poiret (1789) de-

318
Kool & Thulin: Taxonomic revision of Rhodalsine scribed Arenaria geniculata Poir. from Algeria did a member of the group first get a valid binomial. Gay (1845) coined the name Rhodalsine and made the combination R. procumbens (Vahl) J. Gay, based on Arenaria procum bens Vahl (Vahl 1791). However, R. procum bens, the type of Rhodalsine, was nomenclaturally superfluous when published, and is therefore illegitimate, as Arenaria geni culata, the name with priority, was cited in synonymy.
Graebner (in Ascherson & Graebner 1918) treated the group as Minuartia subg. Rhodalsine, and since then it has mostly been seen as part of Minuartia. According to McNeill & Bassett (1974), the number of species in the subgenus is four or five. However, during the last decades Rhodal sine has, to some extent, come back into use again at the generic level (for details, see Kool & Thulin 2013).
In the analysis presented by Kool & Thulin (2017), three samples of Minuartia subg. Rhodalsine were included, representing M. geniculata (Poir.) Thell. from the Mediterranean region (Spain), M. platyphylla (J. Gay ex Christ) McNeill from the Canary Islands (Lanzarote) and M. vestita (Baker) McNeill from Somalia. In the resulting phylogeny, the three samples formed a strongly supported clade and within this clade the samples of M. platyphylla and M. vestita were weakly supported as sisters. When here revising the group, as Rhodalsine, we argue for the recognition of a single variable species in this genus.

Material and methods
Herbarium material and/or images from the following herbaria have been studied: BM, BORD, C, FI, H, K, MA, MPU, O, P, P-Desf, P-LA, PRC, S, TO and UPS. The material studied comprises a representative sample from all parts of the area of distribution, roughly 300 collections in total. As far as possible, the names have been typified, and all type material cited has been seen, either during personal visits to herbaria or as digital images. The distribution map is based almost entirely on information from herbarium labels, but in a few cases also on reliable literature sources, such as Fiori (1923) and Phitos (1997). Note -Psammanthe Rchb. (Reichenbach 1841) was published as a replacement name at generic level for the validly published generic subdivision Alsine [unranked] Psammophilae Fenzl (Fenzl in Endlicher 1840). Two species names, Arenaria thymifolia Sm. and A. procumbens Vahl were cited by Fenzl. Mattfeld (1921)    Rchb., a name not used since its publication, would have had priority over Rhodalsine, but is now, following a proposal by Kool & Thulin (2013), a rejected name (Appelquist 2014; Wilson 2016; Turland & al. 2017).
Description -Herbs perennial or sometimes annual, with indumentum of glandular hairs. Leaves opposite, without stipules, but leaves of each pair with shortly connate bases. Flowers in lax cymes. Sepals 5, connate at base into a short hypanthium. Petals 5, pink or white, inserted at apex of hypanthium, shortly clawed at base. Stamens 10, in 2 whorls, those of outer whorl inserted at base of petals, those of inner whorl antisepalous, inserted on a low, ring-shaped rim surrounding base of ovary; pollen trizonocolpate. Ovary shortly stipitate, subglobose, thin-walled; styles 3, filiform. Capsule opening by 3 valves. Seeds many, orbicular-reniform, wingless; radicle of embryo accumbent. Chromosome base number x = 9.
Distribution -A genus of a single variable species distributed in the Canary Islands, around the Mediterranean and in the Horn of Africa region. Note -Vahl (1791) stated concerning the provenance of Arenaria procumbens: "Legi in ruderatis Tuneti. Forsk. in AEgypto". The reference to "Forsk." is based on the report by Forsskål (1775) of "Cherleria sedoides" from Cairo, collected during his stay in Egypt 1761 -1762. Cherleria sedoides was an obvious misidentification by Forsskål. There is no extant Forsskål material of C. sedoides (Christensen 1922;Hepper & Friis 1994), but two sheets annotated by Vahl as A. procumbens are present in C: (1) a sheet from "Barbaria" without collector [C10000449] and (2) a sheet collected by "Tönder Lund" in "Tuneti" [C10000450]. The second sheet, which consists of a well-preserved flowering branch, is here designated as the lectotype. Vahl's tab. 33 illustrates a richly branched specimen that does not match any of these two sheets. It was most probably based on the now lost Forsskål material from Egypt. This illustration would also be possible as a lectotype, but we prefer the specimen. = Arenaria herniariifolia Desf., Fl.  (1826) in the protologue of Arenaria rosea (as "A. rosea α. lancifolia") cited "t.10, f.1A" in Boccone (1674), an illustration that is therefore available for typification. However, Presl's collection in PRC is preferred as the lectotype. Note -Presl (1826) in the protologue of Arenaria rosea var. spathulata (as "A. rosea β. spathulata") cited "t.10, f.1B" in Boccone (1674), an illustration that is therefore available for typification. However, Presl's collection in PRC is preferred as the lectotype.  (1837), in the protologue of Arenaria procumbens var. linearifolia, gave the following information: "In maritimis Cagliari alla Scaffa, alla Maddalena; Porto scuso. Fl. aprili, junio". In the Moris Herbarium at TO, there are two extant specimens of "Arenaria procum bens var. linearifolia", one with the annotation "littorea maritima majo" and the other with "la Maddalena aprili 1826". The latter specimen agrees better with the protologue and is here designated as the lectotype. = Alsine procumbens f. Description -Stems from a usually ± woody taproot, prostrate to ascending or suberect, 10 -70 cm long, often much branched, sparsely to densely pubescent with glandular hairs 0.2 -1(-1.5) mm long, particularly in the upper part. Leaves sessile or almost sessile, linear to obovate or almost orbicular in outline, 4 -35 × 0.5 -9 mm, sparsely to densely pubescent with glandular hairs, with prominent midrib abaxially, apex acute to obtuse. Flowers in terminal and axillary lax cymes; bracts similar to leaves but smaller; pedicel 3 -25 mm long, erect to spreading, slender, glandular pubescent. Sepals lanceolate to narrowly ovate, 2.5 -5 mm long, obscurely 1 -3-veined, particularly inner sepals with ± wide scarious margin, glandular pubescent abaxially, apex obtuse to subacute. Willdenowia 47 -2017 Petals broadly elliptic to ovate, slightly shorter to slightly longer than sepals, apex obtuse to subacute. Stamens shorter than petals; filaments slender, glabrous; anthers yellow, oblong, 0.5 -0.8 mm long. Styles 1 -2 mm long.
Distribution and ecology -Rhodalsine geniculata is known from Portugal, Spain (including the Canary Islands, Melilla and the Balearic Islands), Gibraltar, Italy (including Sardinia, Sicily, Pantellaria and Linosa), Malta, Greece, Cyprus, Morocco, Algeria, Tunisia, Libya, Egypt and Somalia (Fig. 3). It occurs in sandy, stony or rocky places from near sea level up to 2050 m. In most of its range it is found at elevations below 600 -700 m and it is often confined to habitats on or close to sea shores. However, in Morocco the species is also found at elevations up to about 1500 m in the Atlas Mountains, whereas in Somalia the known altitudinal range is 1350 -2050 m.
Taxonomic remarks -Already Boccone (1674) captured much of the variation within Rhodalsine genicu lata in his illustrations of a broad-leaved and a narrowleaved form of "Alsine maritima …" in Sicily. Many of the names in the synonymy above have been coined for forms with different leaf shapes, but also density of the indumentum and length of the hairs, flower colour and seed characters have been used to differentiate taxa. During the last decades there has been a general agreement in floristic treatments in Europe and most of North Africa that only a single variable species, Rho dalsine geniculata (or Minuartia geniculata), can be recognized (e.g. Meikle 1977;Ghafoor 1978;Jalas & Suominen 1983;Favarger & Montserrat 1990;Halliday 1993;Phitos 1997;Boulos 1999).
However, McNeill (1962) stated that Minuartia platy phylla in the Canary Islands "certainly seems to merit specific recognition", and McNeill & Bassett (1974) also considered M. webbii in the Canary Islands to be distinct, apparently mainly by being annual. Three species have since then generally been recognized in the Canary Islands: M. geniculata, M. platyphylla and M. webbii (e.g. Bramwell & Bramwell 2001;Schönfelder & Schönfelder 2012). Bramwell & Bramwell (2001) described M. webbii as "annual to perennial" and M. platyphylla as having ovate to suborbicular leaves in contrast to the "narrower-leaved Mediterranean species M. geniculata". In Morocco a distinction is currently made between a more broad-leaved Rhodalsine (Minuartia) senneniana with longer hairs and a narrow-leaved R. geniculata with shorter hairs (Ouyahya 1999;Fennane & Ibn Tattou 2005). However, the variation seen in the Canary Islands and Morocco is part of a continuous variation across the whole area of distribution.
Minuartia vestita in Somalia was first associated with Rhodalsine by McNeill (1962), who stated it to be a "distinct species", but no particular characteristics were mentioned. Today a much richer material from Somalia is available and, according to our investigations, M. vestita falls entirely within the variation of Rhodalsine geniculata.