Diagnostics, taxonomy, nomenclature and distribution of perennial Sesuvium (Aizoaceae) in Africa

Abstract The taxonomy of perennial Sesuvium species in Africa has been poorly investigated until now. Previously five perennial species of Sesuvium were recognised in Africa (S. congense, S. crithmoides, S. mesembryanthemoides, S. portulacastrum, and S. sesuvioides). Based on the differing number of stamens, S. ayresii is accepted here as being distinct from S. portulacastrum. Field observations in Angola also led the authors to conclude that S. crystallinum and S. mesembryanthemoides are conspecific with S. crithmoides. A new subspecies, Sesuvium portulacastrum subsp. persoonii, is described from West Africa (Cape Verde, Gambia, Guinea-Bissau, Mauritania, Senegal). The molecular phylogeny indicates the position of S. portulacastrum subsp. persoonii within the “American lineage” as a part of the Sesuvium portulacastrum complex which needs further studies. A diagnostic key and taxonomic notes are provided for the six perennial species of Sesuvium found in Africa and recognised by the authors (S. ayresii, S. congense, S. crithmoides, S. portulacastrum subsp. portulacastrum, S. portulacastrum subsp. persoonii, S. verrucosum and the facultatively short-lived S. sesuvioides). The distribution of S. crithmoides, previously considered to be endemic to Angola, is now confirmed for the seashores of Republic of Congo and DR Congo. The American species S. verrucosum is reported for the first time for Africa (the Macaronesian islands: Cape Verde and the Canaries). It is locally naturalised in Gran Canaria, being a potentially invasive species. These findings as well as new records of S. verrucosum from Asia and the Pacific Islands confirm its proneness to transcontinental introduction. Lectotypes of S. brevifolium, S. crithmoides, S. crystallinum and S. mesembryanthemoides are selected. The seed micromorphology and anatomy of the perennial African species is studied. Compared to the seeds of some annual African Sesuvium investigated earlier, those of perennial species are smooth or slightly alveolate. The aril is one-layered and parenchymatous in all species and usually tightly covers the seed. The aril detachments from the seed coat that form a white stripe near the cotyledon area easily distinguish S. verrucosum from other species under study.


Introduction
Sesuvium L. is one of the most widespread genera of Aizoaceae occuring in many subtropical and tropical regions of the world ). The perennial Sesuvium species often form mono-or oligodominant plant communities in coastal areas (e.g. Oliver 1871, Sauer 1982, Nellis 1994. Sesuvium portulacastrum (L.) L. is considered to be the species with the widest distribution pattern on all continents compared to the other representatives of the genus (Bogle 1970, Lonard and Judd 1997, Bohley et al. 2015. During the last decades the number of recognised species changed from eight (Bogle 1970) to twelve (Hartmann 1993) and reached fourteen after the inclusion of three American species of Cypselea Turpin ).
In its current circumscription, Sesuvium includes perennial or annual herbs with prostrate or ascending, often rooting at the nodes, glabrous or vesiculose stems (additionally with stout warts when dry; Sukhorukov et al. 2017); opposite, more or less succulent leaves with short or hardly visible petioles, which bear two semi-amplexicaulous, membranous or hyaline, entire or fimbriate, marginally concrescent stipules; axillary, bracteolate, pedicellate or sessile flowers; five, bi-coloured (green dorsally and pink or white ventrally) perianth lobes; five to numerous pink stamens; ovary consisting of two to five carpels; circumscissile capsule with the central column bearing 5-50 black or reddish, smooth or diversely sculptured seeds completely or partially covered with thin and hyaline aril.
Sesuvium is the type genus of subfamily Sesuvioideae (Lindley 1853, as "Sesuvieae") which is characterised by stipulate or stipule-like leaf margins; bracteolate, perigynous flowers; externally sepaloid and internally petaloid perianth with the segments mostly apiculate on the back towards the apex, circumscissile capsule and seed usually covered by an aril (Hartmann 1993). Sesuvioideae is sister to all other Aizoaceae (Klak et al. 2003). Two major subclades were recognised within this subfamily: Sesuvioideae s.str. and Tribulocarpus S.Moore (Klak et al. 2003, Thulin et al. 2012. A recent study found the monotypic Anisostigma Schinz to be closely related to Tribulocarpus (Klak et al. 2017a), which together are now recognised as the tribe Anisostigmateae Klak (Klak et al. 2017b). The Sesuvioideae therefore now consists of two tribes, the Anisostigmateae (two genera) and the Sesuvieae comprising Sesuvium (including Cypselea), Trianthema L. and Zaleya Burm.f. Sesuvium is divided into two subclades, the American lineage with C 3 photosynthesis (S. portulacastrum, S. verrucosum Raf., S. maritimum (Walter) Britton, Sterns & Poggenb., as well as the species formerly included in Cypselea) and the African lineage comprising the native African species with a C 4 photosynthetic pathway . The species of each lineage are characterised by several types of leaf anatomy and are distinguished by the shape of the epidermal cells and by the mesophyll structure (Bohley et al. 2015). In the previous paper (Sukhorukov et al. 2017), the annual species of Sesuvium in Africa were revised. Instead of one (e.g. Jeffrey 1961, Hartmann 2002 or two ) species, four native species were accepted (S. digynum Welw., S. hydaspicum (Edgew.) Gonç., S. nyasicum (Baker) Gonç. and S. sesuvioides (Fenzl) Verd.) based on morphological and carpological characters. A new taxonomic treatment of the entire genus  suggested the presence of five perennial species of Sesuvium in Africa: S. congense Welw., S. crithmoides Welw., S. mesembryanthemoides Wawra, S. portulacastrum and S. sesuvioides (also as a perennial species). All perennial taxa usually grow on the seashores of tropical Africa. One of them -S. portulacastrum -is considered to be a widespread species on the continent (Hutchinson and Dalziel 1927, Jeffrey 1961, Lonard and Judd 1997, whereas three others -S. congense, S. crithmoides and S. mesembryanthemoides -have been documented for Angola only (Welwitsch 1859, Oliver 1871). However, Bohley et al. (2017) acknowledged that some of their taxonomic conclusions have been tentative and that further more detailed studies would be required to establish species limits within Sesuvium (e.g., S. crithmoides). Hereby, the results of such a study are published.
The authors' own field investigations, revision of relevant herbarium material and further taxonomic studies revealed a greater diversity of the perennial Sesuvium in Africa in contrast to the latest revision of the genus worldwide . Additionally, the fine seed traits (micromorphology and anatomy) of perennial Sesuvium have been studied for the first time and some new samples have been added to the molecular analysis. Based on this, an improved taxonomy and phylogeny have been presented and the distribution of the perennial Sesuvium in Africa has been discussed.

Methods
Field studies and revision of the herbarium material  Salas-Pascual (2016) and Filip Verloove (March-April 2017) in the Canary Islands (Spain) and by Cornelia Peter Bruyns in Angola (December 2016-January 2017). Additionally, the first author (AS) examined herbarium specimens in the herbaria B (on loan in Mainz, Germany), BM, BR, E, G, K, L (incl. U and WAG, but the African material in WAG was on loan), LE, LY, LYJB, M, MHA, MSB, MW, P, WIND; Filip Verloove identified the material in LPA; Cláudia Baider revised the specimens in MAU and Cornelia Klak in BOL, LUBA and PRE. In addition, some material of Sesuvium portulacastrum (leaves and seeds) collected by AS in Grenada (Lesser Antilles, Caribbean Islands) and Israel (as a cultivated plant in the Dead Sea area) was also used for anatomical and molecular studies.
To assess the conservation status of each taxon as per the IUCN Red List, past and present distribution data from herbarium specimens were collated. When the original specimen label did not give the precise location, a geographical point centred in the locality of the collection cited was used. This information was then assessed based on available ecological data or review of threats to allow insights into understanding the current population and distribution trends useful in defining the IUCN Red List Categories and Criteria (IUCN 2017). The extent of occurrence (EOO) and area of occurrence (AOO) were calculated using GeoCAT ver. β, with a cell of 2 × 2 km 2 (Bachman et al. 2011). These assessements were not sent to the respective SSC IUCN groups prior to the publication of this article.

Leaf anatomy
The leaves of Sesuvium portulacastrum subsp. persoonii were collected by AS in August 2015 in Cape Verde (Sal Island, near Santa Maria village) and soaked in a 70% alcohol solution. The sections were made by hand and stained with 0.2% aqueous toluidin blue. For the description of the leaf anatomy, the terminology by Bohley et al. (2015) was followed. The leaf structure was photographed with a Nikon DS-Vi1 camera (Nikon Corporation, Japan) at the Department of Higher Plants, Lomonosov Moscow State University.

Seed morphology and anatomy
Seed micromorphology was observed using a scanning electron microscope (SEM) JSM-6380 (JEOL Ltd., Japan) at 15 kV after sputtercoating with gold-palladium in the laboratory of Electron Microscopy at Lomonosov Moscow State University. No dehydration of the seeds was required prior to SEM observation due to the absence of soft tissues (e.g. papillae or trichomes) on their surface.
The cross-sections of the seeds were prepared using a rotary microtome Microm HM 355S (Thermo Fisher Scientific, USA) and photographed with a Nikon DS-Vi1 camera (Nikon Corporation, Japan) at the Department of Higher Plants, Lomonosov Moscow State University. Before sectioning, the seeds were soaked in water:alcohol:glycerin (1:1:1) solution, dehydrated in ethanol dilution series and embedded in the Technovit 7100 resin (Heraeus Kulzer, Germany).
The list of specimens used for SEM (perennial species) and anatomical investigations (both annual and perennial taxa) is given below. For seed morphology of the annual Sesuvium taxa, see Sukhorukov et al. (2017). Marais: Ilot Marianne, 18 Jan 1975

DNA extraction and PCR
Total DNA was extracted from silica gel-dried or fresh material of S. portulacastrum (collected in Israel and Grenada), S. portulacastrum subsp. persoonii (Cape Verde) and S. nyasicum (Namibia). The DNA from fresh material was extracted according to Krinitsina et al. (2015) and that from dry leaves was extracted using DiamondDNA Plant kit (DiamondDNA, Russia) with further purification using AMPure Beads (Beckman Coulter, USA) (for details see Krinitsina et al. 2015). The nuclear ITS region (internal transcribed spacer 1, 5.8S ribosomal RNA gene and internal transcribed spacer 2) and three plastid regions (rps16 gene intron, atpB-rbcL intergenic spacer, trnL-trnF intergenic spacer) were sequenced. PCR amplifications were carried out in a Thermal Cycler T100 (Bio-Rad, USA) using primers and cycler programmes listed in Table 1. The reaction mix (25 μl) contained 10 ng of DNA, 1 μM of each primer, 200 μM of each dNTP and 0.5 U hot start TagF polymerase (AmpliSens, InterLabService, Russia). PCR products were checked on 1.2% agarose gels and purified using AMPure Beads (Beckman Coulter, USA) according to the manufacturer's protocol. AMPure Beads suspension was mixed with a solution containing PCR-product at the ratio 1.2:1 (for ITS and atpB-rbcL primer pairs) or 1.4:1 (for all other primer pairs). The sequencing was performed at Genome centre, Engelhardt Institute of Molecular Biology (Moscow, Russia) on Applied Biosystems 3730 DNA Analyser using ABI PRISM® BigDye™ Terminator v.3.1 Cycle Sequencing Kit.

Sequence alignment and phylogenetic reconstruction
The raw forward and reverse sequences were checked and combined in BioEdit sequence alignment editor v. 7.0.5.3 (Hall 1999). The sequences were aligned using Muscle algorithm and MEGA6.0 software package (www.megasoftware.net; see Ta mura et al. 2013). Two data sets were assembled: (1) consisting of three chloroplast markers and (2) the nuclear (ITS) gene region. These data sets were first analysed separately and then in combination using the Maximum Likelihood (ML) method in MEGA 6.0 (Tamura et al. 2013) and Bayesian Inference (BI) in BEAST (Bouckaert et al. 2014). A bootstrapping of 1,000 replicates for ML analysis was processed to estimate the confidence probabilities on each branch of the phylogenetic trees constructed. An initial tree (ML) for the heuristic search was obtained by applying the Neighbour-Joining method to a matrix of pairwise distances estimated using the Maximum Composite Likelihood approach. All positions containing gaps were treated as missing data. Bayesian analyses were run for 20,000,000 generations with four MCMC chains in two independent runs. The first 2,000,000 samples from each run were discarded as burn-in. Convergence was assessed by comparing standard deviation of split frequencies between different runs (MCMC Trace Analysis Tool (Tracer) version v1.6.0; Rambaut et al. 2014). For ML and BI analyses, optimal models of molecular evolution for combined matrices were identified using jModelTest2 (Darriba et al. 2012) (optimal model is GTR + G). Voucher information and GenBank accession numbers are listed in Table 2.

Leaf anatomy
The leaf anatomy of a new subspecies S. portulacastrum subsp. persoonii ( Fig. 1) was investigated.
Description: Leaves terete, of lenticular shape in cross-sections, very succulent, leaf thickness ~4.2 mm; epidermis of the adult leaves mamillate (with slightly convex outer cell walls); hypodermis absent; mesophyll with palisade and water storage cells; palisade cells forming chlorophyll-containing tissue arranged in 3-7 layers below the epidermis (~0.6-0.7 mm from each leaf side), with abundant druses; the cells of innermost palisade layer and adjacent cells of water storage tissue with abundant starch grains (looking like dark stripes: Fig. 1); water storage cells arranged in numerous layers; one or rarely two main vascular bundles in the centre of the leaf are present, with numerous lateral vascular bundles.
The anatomical structure of the leaves of S. portulacastrum subsp. persoonii is similar to that of S. portulacastrum (type subspecies) described by Bohley et al. (2015). The difference between the "Tribulocarpus type" (e.g. Sesuvium maritimum, S. verrucosum,   some individuals of S. portulacastrum with papillate leaves) and the "Sesuvium portulacastrum type" (glabrous forms of S. portulacastrum and S. maritimum) appears to lie only in the presence or absence of papillae (bladder cells) on the leaf epidermis (Bohley et al. 2015). Therefore, the authors propose to unite these two types of leaf anatomy into the "Sesuvium portulacastrum type".

Flower, fruit and seed characters
The position of the ovary in Sesuvium is considered superior (e.g. Jeffrey 1961, Adamson 1962, Bogle 1970, Gonçalves 1995 or semi-inferior . Sometimes the flowers are described as perigynous (Hartmann 2002, Hassan et al. 2005b, but this term does not describe the insertion of the ovary as compared to other floral parts. In fact, the connate part of the tepals forms a true hypanthium, with concrescence of the lower parts of the filaments with the inner surface of the flower cup. The insertion of the stamens seems to be near the top of the hypanthium. However, the ovary itself is situated above the point where the other floral parts are inserted and it should therefore be considered superior as in other Sesuvioideae (Bogle 1970). The perigynous flowers and superior position of the ovary are very characteristic traits for the Sesuvioideae as the basal-most lineage within the Aizoaceae. The fruit in Sesuvium is a circumscissile capsule. The capsule is usually shorter than or rarely almost equal to the tepals, especially in some annual species. The reproductive diaspore type is a seed. The mode of seed dispersal in Sesuvium has not yet been investigated, but it was suggested that the seeds might be dispersed by water (Marais 1990, Tomlinson 2016. Taking into account the coastal habitats where almost all perennial species of the genus are found, this assumption seems to be reasonable. All plant parts of perennial Sesuvium in coastal areas are grazed by cattle (Burkill 1985) and thus endozoochory may also be an important mode for dispersal. The thick seed coat protects the embryo against long-lasting water impact or digestion, as in many other species of Caryophyllales with similar dispersal facilities requiring embryo protection (e.g. Netolitzky 1926, Sukhorukov 2008, Sukhorukov and Zhang 2013. The seeds of all perennial Sesuvium under study are roundish, 0.9-1.1 mm in diameter and slightly flattened (Figs 2 and 3). The aril is one-layered, whitish, ca. 1-2 μm thick in cross-section and consists of thin-walled cells. It tightly adheres to the seed coat. However, S. verrucosum is distinguished by the small detachments of the aril from the seed coat forming a distinctive fold in the cotyledon area (Fig. 3E). The aril usually covers the seed completely (Figs 2, 3A-D), but in some seeds of S. portulacastrum, it is only partially present. An aril covering up to half the seed surface is not common in S. portulacastrum (or any other Sesuvium) mentioned by Hassan et al. (2005a). The presence of a tiny aril apparently does not provide any protective function and its role in seed dispersal or germination is unclear.
The seed coat of perennial Sesuvium is smooth or slightly wavy, often with small, radially elongated striae. Hardly noticeable pits were found only in S. verrucosum (Fig. 3F). In cross-section, the testal layer is much thicker than the 1-3 endotegmal layers. In almost all species, the testa thickness ranges from (25-)30 to 50 μm (Fig. 4), but the testa of a S. portulacastrum specimen from Grenada studied for comparison was found to measure between 70 and 80 μm. The outer periclinal wall of the testa cells is clearly thicker than the inner periclinal wall and the protoplast is usually clearly visible. The walls and protoplast of the testa cells are completely filled with tannins, especially the external areas of the outer cell walls, which appear dark brown. The "stalactites" in the outer cell walls are not prominent in comparison to other representatives of the core Caryophyllales (Takhtajan 1991, Sukhorukov and Zhang 2013. The thickness of the tegmen layers is 2-8 μm (each layer has an average thickness of 3 μm). The embryo is annular and the perisperm is copious.
There are no significant differences in seed structure between perennial and annual Sesuvium species growing in Africa. However, the seed-coat testa of some annual African Sesuvium (S. hydaspicum and especially S. nyasicum) has wrinkle-or ridge-like  outgrowths (Sukhorukov et al. 2017). In all other species, the seeds are smooth, except for the annual North American Sesuvium trianthemoides Correll with rugose seed ornamentation (Correll 1966). This species is known only from the type locality and the character of the seed ornamentation could be of taxonomic importance to distinguish it from other related species . These investigations show that the easily visible detachment of the aril from the seed coat, appearing as a patch near the coty- ledon area, clearly distinguishes S. verrucosum from other taxa encountered in Africa. This character is added to the diagnostic key as a taxonomically important trait. Apart from S. verrucosum, this peculiarity is also observed in the North American annual S. maritimum and South American S. parviflorum DC., a forgotten name of a species that is often identified as S. portulacastrum or S. sessile Pers. (Sukhorukov et al., in prep.). S. verrucosum and S. maritimum appear closely related to each other according to the molecular data (Bohley et al. 2015). Other American species previously considered within the genus Cypselea and recently transferred to Sesuvium based on the molecular phylogeny Kadereit -have much smaller, reddish seeds with a thin seed coat (Sukhorukov, pers. observ.). The seeds of these three species (~0.2 mm across) are amongst the smallest in the large "Globular Inclusion" clade (core Caryophyllales: Cuénoud et al. 2002) along with tiny seeds of some Molluginaceae (Sukhorukov et al., in press).
Many African taxa with an annual or perennial life history (S. congense, S. crithmoides, S. crystallinum, S. digynum, S. sesuvioides) possess an indistinctly striate seed surface (Figs 2 and 3; see also Sukhorukov et al. (2017)). Smooth seeds of Sesuvium sesuvioides or indistinctly wrinkled seeds of S. digynum have relatively thin (20-30 μm) testa. However, thickness varies considerably in S. hydaspicum (from 20 to 50 μm) and especially in S. nyasicum (from 20 to 100 μm) due to the presence of protruding "wrinkles" originating from the testa. The testa is thinner between the wrinkles and much thicker in wrinkled areas.

Molecular phylogeny
Several new samples were added to the molecular phylogeny including S. nyasicum, S. portulacastrum and the new subspecies S. portulacastrum subsp. persoonii. In both ITS and chloroplast trees (Figs 5 and 6), as well as in the combined tree (Fig. 7), Sesuvium is divided into two clades referred to as the "African" and the "American" lineages (Bohley et al. 2015). Although the relationships within the "African lineage" are still not resolved, this clade contains the species native to Africa. In contrast, the "American lineage" consists of the species originating in America, including samples of S. portulacastrum collected in Asia and Africa. In all trees, Sesuvium portulacastrum is not monophyletic. The African Sesuvium portulacastrum subsp. persoonii is nested within the "American lineage" as a part of the Sesuvium portulacastrum complex, either as a sister lineage to S. portulacastrum (the sample from Grenada) in the chloroplast tree (Fig. 5) or amongst the Central American samples of Sesuvium portulacastrum complex (Fig. 6). Due to its well-defined distribution range, this new taxon with clearly petiolate, shorter and thicker leaves is considered here as a subspecies of S. portulacastrum. However, the taxonomic status of S. portulacastrum subsp. persoonii needs further studies for the following reasons: (1) the lack of material from the Indian subcontinent, especially S. repens Willd. and S. portulacastrum (s.l.) from the Americas, Africa (e.g. S. ayresii) and Southeast Asia, precludes recognition   of the exact relationships of all taxa within the large "American lineage" and (2) Sesuvium portulacastrum is still considered a highly variable species distributed worldwide . However, some "strange" forms of this species in Asia (especially in the large biogeographical region of Malesia) are present in the European herbaria in a very limited quantity and were not included in the molecular analysis. The preliminary morphological studies (Sukhorukov et al., in prep.) suggest that at least two species need to be reinstated to species rank (S. parviflorum DC. and S. microphyllum Willd.) and some new taxa from South and Central America are yet to be described.

Taxonomy of perennial Sesuvium in Africa
One American species (S. verrucosum) and one new subspecies (S. portulacastrum subsp. persoonii) are added to the taxonomic list of Sesuvium in Africa. The authors also propose to merge S. crystallinum with S. crithmoides. According to the latest investigations in Angola, S. sesuvioides previously considered as an annual species (e.g. Gonçalves 1970, Sukhorukov et al. 2017 can be a facultatively short-lived perennial herb. In total, six perennial species in Africa (S. ayresii, S. congense, S. crithmoides, S. portulacastrum, S. verrucosum and the facultatively short-lived S. sesuvioides) and one subspecies of S. portulacastrum mentioned above have been accepted.
Diagnostic key to perennial Sesuvium in Africa  Description. The description of S. ayresii was provided by Marais (1978). The most indicative characters of this species are small (up to 25-28 mm long, but usually smaller) terete or semi-terete leaves and (sub)sessile flowers (see Marais 1978, Hartmann 2002. Additionally, Marais (1978) reported a smaller number of stamens (12-20) that have never been observed in S. portulacastrum (stamens more than 30). The smaller seed size (~1 mm) of S. ayresii compared with S. portulacastrum (Marais 1978) seems to be an insignificant diagnostic trait. Leaf shape and leaf size are very variable, sometimes within a given individual.
Ecology. Sesuvium ayresii usually grows on coral rocks, basalts or calcarenites (Marais 1978), but it also can be encountered on sandy seashores, like many other species of the genus. The records of Sesuvium from the calcarenite islets of Les Bénitiers (Johnston 1894) and Rochers des Oiseaux (Johnston 1895) probably belong to S. ayresii. Sesuvium ayresii is reported as the only member of the genus in the Mascarenes (Marais 1990).
Additional specimens examined (Fig. 14).  (IUCN 2017). This assessment is based on the species' EOO of 24,241 km 2 and AOO of 68 km 2 ; together with other factors including the species' habitat being restricted to seashores affected by salt spray, fragmentation of the populations and a high probability of losing sites in the near future due to habitat transformation (construction of hotels, improvement of seashores by removal of vegetation, dumping of refuse in the coastal belt), especially on mainland Mauritius. Only a few of the populations are located in areas with some degree of protection such as Nature Reserves or National Parks (one on Rodrigues; nine on Mauritius), most of them being on small islets. Some records are over 50 years old and need to be updated to determine any decline in its geographic distribution. Competition with invasive alien plants seems not to be a serious problem for this species, although sea-level rise is reducing the area of suitable habitat. Nomenclatural notes. A specimen in LISU has been wrongly stated to be the holotype by Gonçalves (1965) and then by Bohley et al. (2017). Indeed, the sheets of S. congense with the same label and collection number are present in several herbaria, as are many other specimens of Welwitsch's material from Angola (Albuquerque et al. 2009). No specimens and herbarium were cited in the protologue (Welwitsch in Oliver 1871) except the location "Lower Guinea, Congo [Angola as a part of Kongo Kingdom], Ambriz". The lectotype selected here is in accordance with Art. 9.9 of ICN (McNeill et al. 2012). The synonymisation of S. congense with S. portulacastrum (Adamson 1962) is incorrect.

Sesuvium congense
The epithet "congense" probably refers to the "Kingdom of Kongo", a West African kingdom that united the territories of northern Angola (incl. Bengo and Zaire provinces) and the western part of DR Congo, as well as portions of Republic of Congo and Gabon.
Description. The morphological description of the species is provided in Oliver (1871), Gonçalves (1970) and Bohley et al. (2017). This species is sometimes confused with branched S. sesuvioides (especially when the upper parts of the branches are collected) with similar smooth seeds. In contrast to S. congense or related S. crithmoides, S. sesuvioides is glabrous, with turbinate or balustriform flowers (without a rounded perianth cup). General distribution (Fig. 9). Coastal sandy areas in Angola, from Bengo to Namibe provinces, recorded at altitudes between 74 and 252 m a.s.l. (Gonçalves 1965).
Conservation status. Sesuvium congense has an estimated EOO of 54,340 km 2 (which would place the species in LC) and AOO of 36 km 2 (which would place it in EN). However, it is unknown if the species persists in some of these localities. The size of its populations and their threats are little known, but the populations on the seashore and near rivers are probably impacted by development and agriculture. Therefore, the species, at this point in time, should be considered Data Deficient (DD) according to the IUCN Red List Criteria (IUCN 2017).  Note. Welwitsch collected this new species in 1854 and 1858 from several neighbouring locations in Luanda Province. All examined sheets were labelled with the same collector's number (2386) and the location of the lectotype specimen is close to that mentioned in the protologue (Barra do Dande settlement, ca. 30 km N of Luanda). Surprisingly, none of the authentic specimens contained the name of Barra do Dande (Welwitsch 1859) and the species itself was not mentioned in a subsequent treatment of the genus (Welwitsch in Oliver 1871).  -Sesuvium mesembryanthemoides Welw., Ann. Conselho Ultramar. ser. 1: 557 (1859), nomen nudum Note. Welwitsch (1859) mentioned the name Sesuvium mesembryanthemoides (nomen nudum) for the first time, but did not describe the plant morphologically ("Uma nítida espécie de Sesuvium" [a distinct species of Sesuvium]). He probably supposed that it was conspecific with S. crithmoides, which was described in the same article (Welwitsch 1859 Note. Interestingly, Wawra collected the same species in Angola independently from Welwitsch and used the same epithet "mesembryanthemoides" for his new Sesuvium. Unfortunately, the original sheets of S. mesembryanthemoides Wawra cited in the protologue ("in littore maris prope Benguelam, Wawra 210": Wawra and Peyritsch 1860) were destroyed in B, W or WU Johannes Walter, pers. comm.). Wawra and Peyritsch (1860) reported the presence of four to six bracteoles in the flowers of S. mesembryanthemoides and it therefore evidently differs from S. congense (with similar narrow leaves), which has flowers with two bracteoles only. Bohley et al. (2017) have designated the lectotype of S. mesembryanthemoides Wawra in the herbarium LISU (with isolectotypes in BM, BR, C, COI, K, LE) based on Welwitsch's specimens ("Mossamedes [Namibe], seashore, 1 Jul 1859, Welwitsch 2389"). However, the material collected by Welwitsch in Namibe province of Angola is not mentioned in the protologue of S. mesembryanthemoides Wawra and does not belong to the original material. Therefore, this lectotypification cannot be accepted. A lectotype using a Wawra's specimen (syntype) seen in LE has been selected. Two locations ("Mossamedes" and "Benguela") were indicated in the protologue. The lectotype of Sesuvium crystallinum is selected here from the specimens collected by Welwitsch with the number 2389 which were located in different herbaria including LISU ("holotype" in Bohley et al. (2017); not correctable to "lectotype" under Art. 7.10).
Taxonomic and nomenclatural notes. The type material of S. crithmoides comprises the plant fragments with narrow (linear or lanceolate) leaves reaching 8 cm in length. The leaf length and shape is a single character used for its delimitation from the closely related S. crystallinum (Gonçalves 1970) and S. mesembryanthemoides Wawra . Both species are considered to have shorter (up to 5 cm) and broader leaves. However, the authentic material and protologue of S. mesembryanthemoides clearly state that this plant was described as a remarkable species with subtriquetrous-terete (narrow) leaves (Wawra in Wawra and Peyritsch 1860). Therefore, the use of S. mesembryanthemoides as a priority name against S. crystallinum (Hartmann 2002, Figueiredo and Smith 2008) with broader and shorter leaves cannot be accepted. In all characters, including leaf length and shape, S. crithmoides and S. mesembryanthemoides are clearly conspecific. Lectotype (Wijnands 1983). Hermann (1698), Icon. 212 [112, a typographic error], as "Portulaca corassavica …".
Two subspecies of S. portulacastrum growing in Africa have been accepted.  The lectotype is chosen due to inclusion of two elements in the protologue (Schumacher 1827), a specimen cited and a drawing (Table 216, Fig. 1).

Sesuvium portulacastrum subsp. portulacastrum
Taxonomic notes. The autonymous subspecies is of American origin and is known in many parts of tropical Africa and other continents, especially in regions with a hot and humid climate. According to the lectotypification undertaken by Wijnands (1983), the "true" S. portulacastrum is present in Central America (including the Caribbean Islands). The following characters distinguish this subspecies: rampant ramification, glabrous stems and adult leaves with mamillate epidermis, petioles up to 3 mm, oblong-spatulate leaves of 20-60 × 5-10(12) mm and 1.5-4 mm thick, conspicuous (7-12 mm) pedicels, flowers 10-15 mm in diameter and slightly elongated seeds. This description makes sense, because the species is non-uniform in its morphological characters (e.g. leaf length, presence of papillae on stems and leaves, seed ornamentation) and is corroborated by the molecular phylogeny . Although S. portulacastrum is considered to have numerous synonyms , at least some of them need further studies due to the presence of morphological differences, e.g. S. microphyllum Willd. (Caribbean Islands), S. sessile Pers. (South America?) or populations growing in Southeast Asia. In addition, Sesuvium is represented in Central America by at least six taxa (Sukhorukov et al., in prep.) and two of them have to be described as new species.
From humid coastal parts of West Africa, only one perennial species was described, S. brevifolium Schumach. & Thonn. (Schumacher 1827). This species has spatulate or oblong leaves with very short petioles, the characters being typical of Sesuvium portulacastrum. For this reason, S. brevifolium is merged with S. portulacastrum subsp. portulacastrum, this being in agreement with other accounts (Hooker 1849, Welwitsch in Oliver 1871. The autonymous subspecies of S. portulacastrum is distributed along the sea shores of many parts of tropical and subtropical Africa (Exell 1944, Jeffrey 1961, Gonçalves 1979, Gilbert 1993, Friedmann 1994, Sosef et al. 2006, Lisowski 2009, Acebes-Ginovés et al. 2010 and it seems to be present in almost all regions of Africa except South Africa. The causes of such invasion to seashore communities in Africa or in other regions of the Old World are not clear. It can be partially explained by the cultivation of S. portulacastrum in some areas for ornamental purposes, but mostly by transportation of its seeds in the sand ballast of ships sailing between America and other parts of the world in the 15 th -17 th centuries. The examination of the herbarium specimens indicates that S. portulacastrum was sometimes collected in the same places as native Sesuvium (S. congense or S. crithmoides), e.g. on seashores of Kongo-Central province (DR Congo) and Angola.
Additional General distribution. The subspecies seems to be widely distributed on the seashores of the tropics, but some populations from tropical America and SE Asia are distinct in their morphological characters. The distribution of Sesuvium portulacastrum subsp. portulacastrum in Africa is presented in Fig. 14. Diagnosis. Differs from the autonymous subspecies by the absence of rampant ramification, clearly petiolate leaves (petioles 5-10 mm long) that are usually less than three times as long as wide (all blades including those of upper leaves ovoid or oblong, 20-40 × 10-15 mm) and 3-9 mm thick.
Ecology. Sandy beaches near the sea and seasonally flooded, saline plains on the landward side of the coastal dune belt.
Flowering and fruiting. All year round, but most abundantly from September to May (at least in the Cape Verde Islands).
Taxonomic and nomenclatural notes. Franz Wilhelm Sieber labelled his Sesuvium collections from Senegal as S. pedunculatum Pers. The use of this name for the African material is very confusing but explained here.
The name was published by Persoon (1806), who provided a very short diagnosis mentioning pedicellate flowers (not petiolate leaves!) and noted that the species originates from India. It is assumed that Persoon probably did not see the plant in the wild. A specimen was found in the De Candolle herbarium (G-DC) that contains three fragments of different origin: two fragments of S. portulacastrum from the Caribbean and one fragment of Sieber's collection from Senegal (1825) named S. pedunculatum. However, the material kept at G-DC is not a type of S. pedunculatum, but only one of the duplicates sent by Sieber to different herbaria. In Leiden (L), where the largest collection of Persoon's types is deposited, one sheet with two different plant fragments and without any information about their locality (L1693369) was found with the label "Sesuvium pedunculatum Lam." (!) (Fig. 16). Lamarck's authorship of this species is clearly wrong (see Lamarck 1817: 141). The plant fragment on the left side of the herbarium sheet shows typical characteristics of the leaf shape found in S. portulacastrum subsp. persoonii, but it is named by Ch. H. Persoon as S. portulacastrum. The right fragment on the sheet belongs to the autonymous subspecies of S. portulacastrum. According to Persoon's identification, his new species (S. pedunculatum Pers.) is indeed a synonym of the typical S. portulacastrum that has been recorded in India at least since the 17 th century, probably as an alien species (BM, K and L). Sesuvium pedunculatum was treated as a variety under S. portulacastrum (as S. portulacastrum var. pedunculatum) by Cambessedes (in Saint-Hilaire 1829), who described this variety from temperate South America (!) as "les fleurs sont un peu plus grandes, et portées sur des pédoncules longs de deux à trois lignes" [the flowers are slightly larger, with the pedicels two to three lines long]. Furthermore, the synonymisation of S. pedunculatum and S. portulacastrum is confirmed by reference of Persoon (Persoon 1806) to the very clear drawing in Lamarck (1793) showing the shoot, flowers and fruits of typical S. portulacastrum. This image in Lamarck (1793)  doubtful, because this plant was not mentioned by other travellers. However, Hooker (1849) reported Sesuvium as a common plant on seashores of the adjacent Senegal. F.W. Sieber was the first to collect the specimens of S. portulacastrum subsp. persoonii (collections from Senegal in early 19 th century, identified as S. pedunculatum). Other specimens, named as S. portulacastrum and collected in mid-19 th century in Cape Verde (São Nicolau Island) and Senegal (without exact location), are stored in the herbaria E and G, respectively. Sesuvium portulacastrum subsp. persoonii (under the names S. pedunculatum or S. portulacastrum) had not been reported amongst the most common plants in the checklists for West African plants until the early 20 th century (e.g. Engler 1910). Chevalier (1920) cited Sesuvium portulacastrum subsp. persoonii (sub S. portulacastrum) for West Africa (Mauritania and Senegal), with subsequent records for Santiago and Sal Islands (Cape Verde), where it grows spontaneously on the seashores and in saline depressions (Chevalier 1935 (Gilli 1976, Gonçalves 1995, Arechavaleta et al. 2005, all as S. portulacastrum).
All populations of perennial Sesuvium seen by the first author (AS) in Cape Verde belong to S. portulacastrum subsp. persoonii. It is common at least in the southern part of Sal Island on the sandy beaches and seasonally flooded saline depressions by the seashores near Santa Maria and in pristine landscapes in Boa Vista (e.g., Santa Monica beach in the southern part of the island). In Sal Island, S. portulacastrum subsp. persoonii is often a characteristic species of such habitats together with other dominant plants of coastal communities, such as Arthrocaulon franzii (Sukhor. Conservation status. Sesuvium portulacastrum subsp. persoonii is common on sandy inland plains on Sal and Boa Vista islands (Cape Verde). Herbarium labels refer to it as a very characteristic plant of seashore communities in Senegal. Currently the construction of new buildings close to the coast is drastically damaging the natural landscapes, especially on Cape Verde Archipelago (Romeiras et al. 2016, Sukhorukov andNilova 2016) and may negatively affect the number of populations. However, at present, as there is doubt about the origin of this new subspecies (if it is native to the region), it should not be assessed for the IUCN Red List until more data is available.
Lectotype (Sukhorukov & al. 2017 Description. The differences between S. sesuvioides and related annual African taxa were provided in Sukhorukov et al. (2017). Here, it is noted that S. sesuvioides is a facultatively perennial herb and, for that reason, it is also included in the list of perennial species (as in Bohley et al. 2017). General distribution. The distribution of S. sesuvioides was mapped in Sukhorukov et al. (2017), but the presence of this species was erroneously indicated in the eastern part of South Africa, due to a misapplication of the name "Kleinfontein". The record from Kleinfontein (24 Oct 1922, Dinter 4151, BM!) indeed belongs to the small village located south of Maltahöhe (Hardap Region, Namibia) and not to the village in Gauteng province (South Africa) mentioned by Sukhorukov et al. (2017). The authors came to this conclusion after tracing the journeys of Kurt Moritz Dinter, who only visited Namibia (it was known at the time as "South-West Africa": Glen and Germishuizen 2010). Likewise, the lectotype specimen was not collected at Garpia river near Swellendam, Western Cape (as indicated in Sukhorukov et al. (2017)), but on the banks of the Orange River (or Gariep River, spelled by Drège as "Garip"), where S. sesuvioides is frequently found. Therefore the records of S. sesuvioides from Gauteng and the Western Cape provinces (Sukhorukov et al. 2017) are erroneous. In South Africa, the distribution pattern of S. sesuvioides is restricted to the Richtersveld and the lower Orange River valley (Northern Cape province). Records in Namibia and Angola are from the Namib desert (Sukhorukov et al. 2017, see also Fig. 19).
Sesuvium sesuvioides has a large geographical distribution with an estimated EOO of 501,893 km 2 , but its AOO is only 60 km 2 (which would place it in EN). Many localities, especially in Namibia, are in desert areas and are presumably under little threat. Some populations collected in the past are likely to be in protected areas today. However, the current size of the populations is unknown. Therefore, the species should be considered as Data Deficient (DD) according to the IUCN Red List Criteria (IUCN 2017).

Conclusion
The taxonomic diversity of perennial Sesuvium in Africa is greater than previously thought. Some species have a broad distribution pattern in tropical Africa. Sesuvium verrucosum is here considered as a naturalised alien species at least in the Canaries. The micromorphology and anatomy of the seeds in perennial African Sesuvium are similar, in contrast to that in annual species of the genus. However, the seeds of American Sesuvium verrucosum (as well as S. maritimum and S. parviflorum) demonstrate a peculiarity in seed morphology (detachment of the aril from the seed coat in the area of the cotyledons).
Bruyns was, in part, funded by the NRF incentive grant (grant no. 103697 to CK) and the University of Cape Town Research Committee. We are indebted to F.B.V. Florens for taking the images of Sesuvium ayresii and Alexandra Konstantinova for assistance in the observation of Sesuvium portulacastrum subsp. persoonii in Cape Verde and for taking close-up images of its flowers in January 2016. Our thanks are also due to Keith Chamberlain and Geoffrey Harper who proofread the text, Johannes Walter (NHM, Vienna, Austria) who provided the information about the presence of Wawra's collections in the herbaria W and WU, Anand Kumar for the images of Sesuvium portulacastrum from India and to Roxali Bijmoer for the image of the original material from Persoon's collections in Leiden (L). We thank the staff of the herbaria visited and especially those of PRE (South Africa), who provided scanned images of Sesuvium collections from Angola.