Spergularia hanoverensis (Caryophyllaceae): Validation and Recircumscription of a Misinterpreted Species from South Africa

The name “Spergularia hanoverensis Simon” has been misapplied to an endemic taxon confined to inland semidesert ecosystems in central-western South Africa. It is commonly accepted as a small annual species occurring in saline habitats in a wide elevation range, but its identity still remains obscure. In the context of taxonomic and phylogenetic research on the African species of Spergularia, we found that the name was never validly published. After revision of herbarium material housed in South African herbaria, a voucher collected from Hanover was found at PRE bearing some labels handwritten by E. Simon that suggest it might be an intended type for the name. Additional herbarium material and wild populations from the Karoo region were identified that matched the samples in that voucher, and taxonomic research was conducted to clarify their identity. Among other characters, those Karoo plants show a woody dense compact habit, woody perennial at base; stems prostrate to ascendent; leaves entirely glabrous, somewhat glaucous; large white-hyaline conspicuous stipules; inflorescence glanduliferous, many-flowered subdichasial cyme, with minute bracts; flowers small, with white petals approximately equalling sepals in length, stamens 7–8, and styles free from base; capsule small, with seeds dimorphic, unwinged to broadly winged, with testa always densely tuberculate. Molecular analyses of plastid (trnL-trnF region) and nuclear ribosomal (5.8S-ITS2 region) DNA sequence data support morphological differentiation of the Karoo plants, for which the name S. hanoverensis is here effectively published. A full morphological description and data on ecology, habitat, distribution, and taxonomic and phylogenetic relationships of S. hanoverensis are compared to other members of the “South African group”, namely S. glandulosa, S. namaquensis, and S. quartzicola, from which the new species considerably differs. The adaptative significance of dimorphic seeds of S. hanoverensis is briefly commented on in the context of the species habitat preference. An identification key is presented for the South African related taxa.


Introduction
Spergularia (Pers.) J.Presl & C.Presl, nom. cons., comprises ca. 60 species of both annual and perennial plants with worldwide distribution, but mainly occurring in the Mediterranean Basin and temperate South America [1]. Some members of the genus are narrow endemics, while others are subcosmopolitan synanthropic plants [2]. In most cases, they occur on soils rich in mineral salts, such as chlorides, sulphates, and nitrates, in both natural and human-disturbed habitats.
Recently, molecular phylogenies recovered a consistent clade, including Spergularia, Spergula L., and Rhodalsine Gay. Some authors [3,4] considered this group a distinct tribe to In the present contribution, Spergularia hanoverensis is validly described for a plant endemic to the Nama-Karoo (NK) biome [16] in western South Africa, according to the original concept "in schedis" of E.E. Simon. A brief story on that name is presented, and data on its morphology, ecology, distribution, and phylogenetic relationships are reported that support recognition of S. hanoverensis at specific rank in the "South African taxa" group.

Morphological and Habitat Studies
Detailed morphological studies were undertaken on both living plants from wild populations and herbarium specimens sourced from the herbaria ABH, BOL, GRA, HBG, K, M, NBG, P, and PRE (acronyms according to Thiers [17]), using an OLYMPUS SZX7 binocular microscope. A personal collection of one of the authors (L.M.) that is currently deposited at ABH, including numerous vouchers of Spergularia, was also studied. Thirty living individuals from two wild populations of S. hanoverensis in Northern Cape province (i.e., Karreekop farm, between Brandvlei and Williston, and Zoekop Farm, SW of Middelpos, on the road to Ganaga Pass) were sampled and analysed in situ. Further, thirteen herbarium vouchers of S. hanoverensis were studied (see Section 3.2), some of them including several duplicate sheets, and with several complete individuals each. The obtained morphological data were focussed on sufficiently illustrating the intraspecific variation of the newly described species, concerning other related perennial South African taxa of the genus, as summarised by Alonso et al. [8,9]. Digital images of Spergularia from iNaturalist (https://www.inaturalist.org/observations/?place_id=any&taxon_id=58170; accessed on 2 May 2023) were also checked, and some were found to meet the distribution of the studied species.
Scanning electron microscope (SEM) micrographs of seeds were taken with a JEOL JSM-IT500HR operating at 15 kV. No special treatment of the material was required prior to observation. At least 5 mature seeds from different individuals of S. hanoverensis from different sites (Table 1) were observed in detail when available. Samples were glued directly onto metallic stubs and then coated with 10 nm platinum in a QUORUM Q150T ES Plus sputter coater. The ImageJ software [18] was used for measurements on SEM micrographs. Authors of the taxa cited in the text follow IPNI [19]. Nomenclatural issues accord with Turland et al. [20]; orthography of geographical names agrees with Leistner and Morris [21]; and the grid-number system is in accordance with the National Geospatial Information [22]. Bioclimate, bioregion, and vegetation classification agree with Mucina and Rutherford [16].

Molecular Analyses
Herbarium vouchers and silica-gel-dried material were used for total DNA extraction employing a modified 2 × cetyltrimethylammonium bromide (CTAB) protocol [23]. For Spergularia hanoverensis, sampling from herbarium material was not permitted, and hence, only silica-gel-dried material from two wild populations (one sample per population) was utilised. Addition of further samples from those same populations did not modify the phylogenetic trees. Total DNA was purified using MOBIO minicolumns and kept in 0.1 × TE buffer (10 mM Tris-HCl, 1 mM ethylenediaminetetraacetic acid (EDTA), pH 8.0). The trnL-trnF region (hereafter trnL-F) of chloroplast DNA (cpDNA) was amplified using specific primers trnL-BOC and trnL-BOF-R as described in Oxelman et al. [24], whereas the whole internal transcribed spacer-ITS-region (ITS1 spacer, 5.8S gene, ITS2 spacer) of nuclear ribosomal DNA (nrDNA) was amplified using the ITS1 (forward) and ITS4 (reverse) primers [25] and then adjusted to match the length of the 5.8S-ITS2 (hereafter ITS2) sequences retrieved from GenBank for alignment. Amplifications were performed on a reaction volume of 25 µL containing 22 µL of ABGene 1.1 × Master Mix, 2.5 mM MgCl 2 (Thermo Scientific, Waltham, MA, USA), 0.5 µL of 0.4% bovine serum albumin (BSA), 0.5 µL of dimethyl sulfoxide (DMSO), 0.5 µL of each primer (10 pmol/µL), and 1 µL of template DNA on a 9700 GeneAmpl thermocycler (Applied Biosystems). The PCR programme for trnL-F was as follows: 2 min at 97 • C, followed by 35 cycles of 97 • C for 20 s, 55 • C for 50 s, 72 • C for 1.5 min, and a final extension at 72 • C for 8 min. The PCR programme for ITS2 was as follows: 2 min at 95 • C, followed by 30 cycles of 95 • C for 1 min, 53 • C for 1 min, 72 • C for 2 min, and a final extension at 72 • C for 5 min.
Sequencer 4.1 (Gene Codes Corp., Ann Arbor, MI, USA) was used to assemble complementary strands and verify software base-calling. Sequence alignment was performed using MUSCLE [26] conducted in MEGA X v.10.2.6 [27] with minor manual adjustments to obtain the final aligned matrix. Forty-five samples belonging to twenty-four species of Spergularia were used for phylogenetic reconstructions, using Rhodalsine geniculata (Poir.) F.N.Williams, Rh. platyphylla Gay in Christ, Spergula arvensis L., Spergula morisonii Boreau, and Spergula pentandra L. as outgroups. Two datasets were built: one for the trnL-F region of cpDNA (matrix with 33 sequences and 990 positions) and another for the ITS2 region of nuclear nrDNA (matrix with 25 sequences and 282 positions). Sequences of each region were retrieved from different plant sources depending on GenBank availability [8,9], except for four of them gathered in South Africa and Spain, which were generated specifically for the present study ( Table 2). For that reason, most accessions in the trnL-F and ITS2 datasets do not come from a unique plant source. Accessions from GenBank filed as "Spergularia fallax Lowe" are shown in our trees as S. flaccida (Madden) I.M.Turner, the name having priority for that species. Phylogenetic analyses of both regions were obtained using maximum parsimony (MP), maximum likelihood (ML), and neighbour joining (NJ) methods. MP analysis was conducted in both PAUP (using heuristic search options with the tree searching strategy based on nearest neighbour interchange, NNI) and MEGA (using heuristic search options with the tree searching strategy based on subtree-pruning-regrafting-SPR-with search level 1; [28]) for result comparison, with 10,000 replicates. ML [29] and NJ [30] analyses were also performed in MEGA, as well as the selection of the best model of DNA substitutions for each method using the Akaike information criterion (AIC; [31]); models with the lowest BIC (Bayesian information criterion) scores were considered to best describe the substitution pattern for the ML and NJ analyses. Phylogenetic reconstructions for ML and evolutionary distances for NJ for the ITS2 matrix were estimated using the K2 model (2-parameter method of Kimura [32]) with the rate variation model allowing for some sites to be evolutionarily invariable (+I, 30.26% sites). In contrast, for the trnL-F matrix, the T92 model (3-parameter method of Tamura [33]) was applied, with a discrete gamma distribution (G = 0.677) modelling the rate variation among sites. In every case, all sites in the matrixes were considered. For comparison purposes, remotion of all ambiguous positions for each sequence pair (pairwise deletion option) was also performed, and no significant differences (only affecting BS values in a few branches) were observed in the obtained phylogenies. For all those methods, support was assessed by the bootstrap [34] with 10,000 replicates but holding only 10 trees per replicate. Clades showing bootstrap percentage (BP) values of 50-74% were considered weakly supported, 75-89% moderately supported, and 90-100% strongly supported.
Furthermore, Bayesian inference (BI) analyses were conducted with MrBayes 3.2 [35], in which the Markov chain Monte Carlo (MCMC) algorithm was run for 10 million generations and sampled every 1000 generations. The general time reversible (GTR) + proportion of invariant sites (I) + gamma distribution (G) model was used in the analyses (set nst = 6 rates = invgamma), according to the results obtained with jModelTest 2.1.10 [36] under AIC. The first 25% of generations (burninfrac = 0.25) were excluded, and the remaining trees were used to compile a posterior probability (PP) distribution using a 50% majority-rule consensus.
More detailed information on plant material sources, GenBank accessions, DNA extraction and sequencing, and data analyses are provided in Alonso et al. [8,9].

On the Trail of Spergularia hanoverensis: A Brief Story of the Name and Its Application
According to Nkonki [15], the name S. hanoverensis should be applied to annual plants occurring in broad elevation and geographical ranges through Northern and Western Cape. However, as previously mentioned, herbarium material labelled S. hanoverensis usually shows morphological traits not matching that broadly assumed concept (see below).
The recovery of this almost forgotten name most probably followed Nkonki's revision of the Spergularia vouchers conserved in the National Herbarium at Pretoria. Among them, the specimen PRE12453 (Figure 1), which was part of E.E. Galpin's personal collection carrying the no. 5967, bears six plant fragments and three labels with relevant information. Firstly, the Galpin's herbarium label with the original collection data handwritten in ink: "Spergularia/Hanover C.C./Coll: T.R. Sim, Jan.
[uary] 1902". The collector, Dr Thomas R. Sim (1858-1838), was an English botanist specialising in forestry, who worked from September 1894 to September 1902 in the Forestry Service of the Cape Colony as superintendent of plantations, stationed at Fort Cunynghame, just north of Stutterheim (Eastern Cape region), and then he moved to Natal. By that time, he collected abundant plant material in the region that he distributed to contemporary botanists [37], including surely Galpin's specimen, as deduced from vouchers housed at several South African and European herbaria. Secondly, a label handwritten in pencil and signed by E. Simon reads, "Species mihi adhuc ignota/An Sp. pallida G.Don.??/Ulterius/nomen dabo" (A species still unknown to me. Perhaps Sp. pallida G.Don.??/Later I will name it). Finally, a third label with unidentified handwriting in ink reads, "Scrap removed for/Dr. E. Simon,/Vice President of the/Soc. Bot. du Centre-Ouest/France,/(sent through the/S. A. Museum C.T. 5.1.39)". This latter label is glued exactly in the place left by the fragment forwarded to Dr. Simon via Cape Town. Undoubtedly, PRE12453 is the material from which the name S. hanoverensis was invented, and hence, it might be considered as the intended type. This fact is crucial for correctly interpreting and further applying the name in its original concept, as shown below. In this regard, it is worth mentioning that some confusion exists around the true authorship of S. hanoverensis. In IPNI [19] and POWO [13], that species name is connected to "C.Simon", which is the standard abbreviation of Charles Simon . On the contrary, most of the South African literature as well as TROPICOS [38] attribute it to In this regard, it is worth mentioning that some confusion exists around the true authorship of S. hanoverensis. In IPNI [19] and POWO [13], that species name is connected to "C.Simon", which is the standard abbreviation of Charles Simon . On the contrary, most of the South African literature as well as TROPICOS [38] attribute it to "Si-mon", the standardisation of Eugène Simon (1848-1924), also known as "E.Simon primus", which is partly congruent with the text annotated on the pencil label on voucher PRE12453. However, a detail in that text allows a different but more accurate interpretation, since the fragment missing in that voucher was indeed sent in 1939 to E. Simon, wrganizarwas then the Vice President of the Societé Botanique du Centre-Ouest (France). According to Guédès [39], Eugène Ernest Simon (1871-1967), whose standard form is "E.Simon" (sometimes also cited as "E.Simon secundus", cf. [40]), was the botanist who held the vice presidency of the regional French botanical society since 1908, and who was specifically working on an unfinished monograph on the genus Spergularia. Therefore, he was the real proposer of the name S. hanoverensis. Irrespectively, the taxon name was never validly published and remained as "nomen nudum", according to Art. 38 Ex. 1 of the ICN [20].
However, that name is currently available in several checklists and web pages featuring the African and/or South African floras, although no more relevant information has been added besides Nkonki's [15]. In particular, S. hanoverensis was assessed as LC (least concern) by Cholo [41], according to the IUCN red list categories [42], mainly on the basis of the already published information that included mapping of four sites far apart based on vouchers at PRE. Nkonki's morphological and distributional data [15] are also shown in APD [43] and in GBIF [44].
The extant published information on S. hanoverensis is still very scarce and confusing, particularly after reviewing herbarium material of Spergularia at PRE. The fragments affixed to voucher PRE12453 ( Figure 1) show some outstanding characteristics allowing safe identification: (i) A subshrubby perennial plant, mostly glabrous, with a compact, often many-branched woody underground base; (ii) Stems often prostrate-ascendent; (iii) Leaves somewhat glaucous, with large whitish stipules, very apparent and showy, long-lasting, giving a Paronychia-like aspect; (iv) Inflorescence glanduliferous, broadly subdichasial, with inconspicuous bracts; (v) Flowers numerous, small, with white petals; (vi) Capsule small, many seeded; (vii) Seeds small, triangular, blackish-brown, matte, densely papillate, unwinged or with a discolorous, vestigial to well-developed wing.
At first glance, they resemble other South African taxa of Spergularia, often misidentified as S. media. This might justify the scarcity of references to S. hanoverensis in the literature prior to Nkonki [15], and also the neglection of that name in more recent accounts [12]. Other specimens at PRE first identified as S. hanoverensis indeed correspond to S. bocconei (collection M.B.Bayer 6006, PRE762662) or S. namaquensis (collection H. M.Steyn 23, PRE583500).
Despite the contrasting interpretations of S. hanoverensis, the Karoo plants matching Simon's concept of that species do not fit the current application of the name in recent checklists and databases of the African flora. Further, the unique combination of characters found in Simon's taxon is missing in the remaining members of the "South African taxa" group of Spergularia, as defined by Alonso et al. [8,9]. The newly obtained morphological and molecular evidence support acceptance of the taxon at specific rank. Description: Subshrub, mostly glabrous excepting the glanduliferous inflorescence, with a compact, often many-branched woody underground base. Stems up to 30 cm high, but commonly smaller, usually prostrate, subcespitose, slightly nodose, with ascending branches. Leaves 4-10(-17) × 0.5-0.7 mm, narrowly linear, semicylindrical and subcanaliculate, green to slightly glaucous-green, caducous when withering, ending in a whitish to yellowish mucro up to 1 mm long; stipules 4-6 × 1-2 mm, whitish-scarious, glabrous, triangular-acuminate, showy and apparent, often reaching at least half the leaf length (namely in the axillary leaf fascicles, where stipules are longer than leaves themselves), those on the young shoots fused up to the basal third and finally only at the base, longlasting after leaf abscission, and giving a Paronychia-like aspect. Inflorescence a subdichasial cyme, broadly branched, many flowered (5-7 flowers per branch), and densely covered with glanduliferous short hairs; bracts 1-2.5 mm long, slightly longer than stipules, inconspicuous, much shorter than leaves. Flowers pentamerous, on erect-patent to patent slender pedicels 2.5-5 mm long at anthesis. Sepals 2-3.3 × 1-1.5 mm, slightly accrescent in fruit, oblong to elliptic, obtuse to subacute, with a central green band 0.5-0.7 mm broad, and scarious margins 0.3-0.4 mm broad (wider in the inner sepals), patent to slightly deflexed at anthesis. Petals 1.8-2.8 × 1-1.2 mm, about equalling to slightly shorter than sepals, white, elliptic, entire. Stamens 7-8, slightly shorter than petals, filament up to 3 mm long, filiform, slightly widened at the base, whitish, anther 0.3-0.4 mm long, yellow, dorsifixed. Ovary ca. 1-2 mm, subglobose, yellowish; styles 3, ca. 1 mm long, free from the base, yellowish, with short apical stigmata. Capsule 3-4 × 2.5-3.5 mm, broadly ovate, slightly longer than sepals, glabrous, shining, yellowish-green outside but reddish inside, opening by three slightly recurved valves, on patent to reflexed pedicels 4-10 mm long, up to 3 times longer than sepals. Seeds numerous, dimorphic; some of them unwinged, 0.5-0.75 × 0.3-0.6 mm, ovate-cuneate to subtriangular in outline, flattened, blackish-brown and matte, the others similar but larger, 0.7-1.1(-1.4) × 0.6-1.2(-1.4) mm, with disk 0.6-0.9 × 0.4-0.8 mm and a discolorous (whitish to greyish), vestigial to entirely developed, eroded wing (both types usually present in a single capsule, with intermediate stages); testa ornamented in all cases with minute irregular tubercles and densely covered all over with stalked globose and also minutely tuberculated papillae.

Taxonomic Treatment
Etymology: The specific epithet (hanoverensis, −e) refers to Hanover, a small town in the Karoo region of the Northern Cape province in central South Africa, where the plant was collected and is native to. Provided that E.E. Simon was the first to recognise this taxon as new "in schedis", we preserve the original name he chose later.
Phenology: Flowering in late October-early January (occasionally in July-August), fruiting in November-February (occasionally in August-September).
Habitat and distribution: Spergularia hanoverensis is an edaphic specialist species usually found on seasonal stream banks, in riverbeds and depressions, often on saline calcareousclayish or sandy soils but sometimes among rocks or stony saline substrates (Figure 2). The elevation of the localities ranges between 700 and 1400 m. The known distribution of the species extends through most of the southern part of the Karoo region in central and western South Africa, ranging from Ceres and Calitzdorp in the Western Cape to Hanover in the Northern Cape province (Figure 3). That territory is mostly included in the Nama-Karoo (NK) biome and reaches the southern Succulent-Karoo biomes (mostly the SKk, SKt, and SKv bioregions) sensu Mucina and Rutherford [16], where it occurs in the so-called "Bushmanland vloere" (code AZi 5). In those areas, the climate is subdesert arid and continental (only scarcely ameliorated by the ocean influences), with average temperatures ranging from −5 • C in winter to 43 • C in summer and frosts being usual at high elevations. The average annual precipitation varies between 100 and 500 mm, though rather differently distributed, with the rainfall occurring mostly during late summer (December to April) with a peak in March [16]. Hanover in the Northern Cape province (Figure 3). That territory is mostly included in the Nama-Karoo (NK) biome and reaches the southern Succulent-Karoo biomes (mostly the SKk, SKt, and SKv bioregions) sensu Mucina and Rutherford [16], where it occurs in the so-called "Bushmanland vloere" (code AZi 5). In those areas, the climate is subdesert arid and continental (only scarcely ameliorated by the ocean influences), with average temperatures ranging from −5 °C in winter to 43 °C in summer and frosts being usual at high elevations. The average annual precipitation varies between 100 and 500 mm, though rather differently distributed, with the rainfall occurring mostly during late summer (December to April) with a peak in March [16]. Notes: Wild populations of S. hanoverensis include numerous individuals covering a large territory in South Africa, and no special threats are known so far that might lead to any inferred decline in either the number of populations or the number of individuals. Therefore, its conservation status is suggested here as least concern (LC) according to IUCN [42]. Nonetheless, extensive fieldwork is still needed to locate new populations in suitable habitats among the known populations, which will allow completing the distribution area of S. hanoverensis or/and detecting eventual variation of adaptative characteristics in distinct environments. The identification key in Appendix A can help to do this.  Notes: Wild populations of S. hanoverensis include numerous individuals covering a large territory in South Africa, and no special threats are known so far that might lead to any inferred decline in either the number of populations or the number of individuals. Therefore, its conservation status is suggested here as least concern (LC) according to IUCN [42]. Nonetheless, extensive fieldwork is still needed to locate new populations in suitable habitats among the known populations, which will allow completing the distribution area of S. hanoverensis or/and detecting eventual variation of adaptative characteristics in distinct environments. The identification key in Appendix A can help to do this. Other

Taxonomic and Phylogenetic Relationships of Spergularia hanoverensis
In the context of an ongoing phylogenetic survey on the South African taxa of Spergularia based on trnL-F cpDNA and ITS2 nDNA sequences, we recently reported [8,9] the first preliminary phylogenetic trees that included South African taxa of that genus. Our results were congruent with the plastid phylogeny of Sperguleae (sensu [3]) obtained by Kool and Thulin [7], and also supported that trnL-F and ITS2 regions offer information useful enough for credible phylogenetic reconstructions of Spergularia.
Adding samples of S. hanoverensis to our molecular matrix yields trees almost identical in general topology to those obtained in our previous contributions [8,9]. Further, the new species falls nested in the "South African taxa" subclade of Clade A (sensu [8,9]) in both the plastid and the nuclear phylogenies (Figures 4 and 5), as recovered in our BI consensus trees (in which PP values are placed above branches and BP values below branches, respectively, from the ML and MP analyses). However, the internal relationships among the four members of that southern lineage are not equally resolved, often with low PP and BP values.
The aligned trnL-F database was 990 bp, 147 of which (14.85%) were potentially parsimony informative. Analyses of this dataset using NJ, MP, ML, and BI methods yielded trees with similar topologies and similar bootstrap and branch length values. The obtained trnL-F BI phylogenetic tree (Figure 4)  Kool and Thulin [7], and also supported that trnL-F and ITS2 regions offer information useful enough for credible phylogenetic reconstructions of Spergularia.
Adding samples of S. hanoverensis to our molecular matrix yields trees almost identical in general topology to those obtained in our previous contributions [8,9]. Further, the new species falls nested in the "South African taxa" subclade of Clade A (sensu [8,9]) in both the plastid and the nuclear phylogenies (Figures 4 and 5), as recovered in our BI consensus trees (in which PP values are placed above branches and BP values below branches, respectively, from the ML and MP analyses). However, the internal relationships among the four members of that southern lineage are not equally resolved, often with low PP and BP values.  BP), a subshrub from saline coastal areas of southwestern and southern South Africa. They both are well-supported sister (0.99 PP, 88/94 BP) to the subclade including the Namaqualand woody subshrubs S. quartzicola and S. namaquensis, which constitute a strongly supported group (1.00 PP, 96/97 BP) not well resolved internally. All these are strongly supported sister (1.00 PP, 93/95 BP) to a third subclade formed by the reddish-flowered S. rubra and S. rupicola (1.00 PP, 88/92 BP) plus the Central European S. echinosperma (Čelak.) Asch. & Graebn. All those constitute the strongly supported (0.98 PP, 90/97 BP) Clade A. Those molecular relationships of Spergularia hanoverensis to the remaining members of the "South African taxa" group correlate to some morphological traits obtained from our morphological studies. Table 3 shows the most important characters of S. hanoverensis compared to other related perennial relatives of the "South African taxa" group, illustrating the intraspecific variation of the newly described species. Spergularia hanoverensis shares some floral resemblance with S. glandulosa, such as the small white flowers with petals equalling sepals in length, and the styles entirely free from the base, features that Those molecular relationships of Spergularia hanoverensis to the remaining members of the "South African taxa" group correlate to some morphological traits obtained from our morphological studies. Table 3 shows the most important characters of S. hanoverensis compared to other related perennial relatives of the "South African taxa" group, illustrating the intraspecific variation of the newly described species. Spergularia hanoverensis shares some floral resemblance with S. glandulosa, such as the small white flowers with petals equalling sepals in length, and the styles entirely free from the base, features that might justify some phylogenetic closeness as recovered in our ITS2 tree ( Figure 5). However, S. hanoverensis shows a compact habit; leaves glabrous and long acuminate (ca. 1 mm), bearing at base longlasting, broad stipules; and inflorescences multiflowered, subdichasial, with minute bracts. These characters all allow easy separation at first glance. In fact, S. hanoverensis exhibits a combination of characters missing in any known members of the South African group. However, some of them occur in any of the three other taxa. In particular, the compact and usually many-branched woody underground base and the multiflowered subdichasial inflorescences with minute bracts are shared with S. namaquensis, and the entirely glabrous and broadly stipulate leaves are also present in S. quartzicola. Nonetheless, both latter taxa show larger flowers with 10 stamens and columnar, long-fused styles, instead of the smaller flowers with 7-8 stamens and free styles of S. hanoverensis. The main morphological vegetative features exhibited by S. hanoverensis inform about some adaptative characteristics of the species to high environmental stress (i.e., high salinity and persistent dry soil at highelevation open habitats), such as the uniformly subshrubby compact habit, along with the slightly glaucous colour of leaves and the long-lasting conspicuous white-hyaline stipules (bringing a silvery Paronychia-like appearance at anthesis) that favour sunlight reflection.  Finally, the capsule and seed features offer the most remarkable diagnostic differences among all those South African species. Spergularia hanoverensis produces the smallest capsules (ca. 3-4 × 2.5-3.5 mm) and also smaller seeds (0.4-0.5 mm long). These are markedly dimorphic, some of them being unwinged, ovate-cuneate to subtriangular in outline, flattened, blackish-brown and matte, whereas the others are similar but with a discolorous (whitish to greyish), vestigial to entirely developed, eroded wing, and testa surface ornamented with minute irregular tubercles and densely covered all over with stalked globose and also minutely tuberculated papillae. Both seed types are sometimes present in a single capsule (Figure 6a-c,f,g), with intermediate stages among individuals ( Figure 6b) and among populations. This fact makes S. hanoverensis the only known member of the "South African taxa" group with such dimorphic seeds.  Seed dimorphism is rather frequent in Spergularia [40,[45][46][47][48][49][50], and for a long time, it has been considered a good diagnostic character for species delimitation [51]. As occurring in other dimorphic taxa such as the widespread annual pinkish-flowered S. marina (L.) Besser (incl. S. salina J. & C. Presl) [52,53], heteromorphism also includes variable seed morphology as well as considerable variation in seed size in S. hanoverensis, the winged seeds being larger, and their disk is also larger than the unwinged seeds. However, seeds of S. hanoverensis, although dimorphic in size and gross morphology (winged and unwinged), are homogeneous in micromorphology, with traits quite distinct from S. marina [54]. According to our observations from herbarium material, no correlation appears between seed features and their geographical provenance. Hence, individuals from different sites apparently exhibit seeds with similar size and gross morphology variation patterns. In other words, most of the variation in seed size occurs within rather than between populations, as also reported for S. marina [53].
Species with seed dimorphism combine different morphological syndromes for a more efficient dispersal in contrasting habitats, the wind being the vector for winged seeds and the water for unwinged ones, as suggested by Telenius and Torstensson [55]. Acquisition of dimorphism in S. hanoverensis might have brought a positive evolutionary effect for dispersal in the inland karroid saline ecosystems it inhabits in South Africa (i.e., seasonally wet depressions, riverbeds and ravines with seasonal flow, etc.), by combining anemochory (winged seeds) and hydrochory (unwinged seeds). Its phylogenetically close relatives from Namaqualand (S. namaquensis and S. quartzicola) also occur in subdesert karroid ecosystems, mostly in dry sandy or quartz substrates. Therefore, the production of unwinged seeds would have no favourable adaptative value in sites not necessarily connected to marsh habitats.

Conclusions
Although Spergularia hanoverensis has been widely interpreted in a sense that includes several biological entities, the name is validated and circumscribed here as it was first outlined "in schedis" by E.E. Simon. Our molecular and morphological data are distinctive enough to accept it as a proper species, which falls together with taxa in the "South African taxa" group of Spergularia. It is a halophytic specialist, mostly endemic to the inland saltmarsh habitats of the Nama-Karoo biome (central-western South Africa). It shows some adaptative characteristics to stressful environments with high salinity and dry soils at high-elevation sites.
New data from wild populations are needed to complete the distribution area of S. hanoverensis and to test the accuracy of our observations on herbarium vouchers regarding within-or between-population variation in seed size and gross morphology, as well as eventual variation of adaptative characteristics of the species to different environments. The dichotomous identification key in Appendix A will surely help to that purpose.