Confirming the identity of two enigmatic “spiny solanums” (Solanum subgenus Leptostemonum, Solanaceae) collected by Jean-Baptiste Leschenault in Java

Abstract Taxonomic revision of the tropical Asian species of Solanum revealed two names, Solanum poka Dunal and Solanum graciliflorum Dunal, whose identities were uncertain and whose application has always been tentative. Material collected in Java at the beginning of the 19th century by Jean-Baptiste Leschenault de la Tour and used to describe these taxa has not been found, despite extensive searches in European herbaria. We here stabilise use of these names by comparing herbarium specimens and drawings of original material made by the artist Toussaint François Node-Véran. Detailed descriptions with synonymy, preliminary conservation assessments and specimen citations are provided for both species. Lectotypes are designated for all names (including synonyms) and epitypes designated for Solanum poka and Solanum graciliflorum to stabilise usage.


Introduction
In 1800, shortly after he became First Consul of the Republic of France, Napoléon Bonaparte approved an expedition along the "coasts of New Holland" (Australia). Th e expedition, led by Nicolas Baudin, has been cited as one of the most ambitious and the most enriching for collections of natural history of the great scientifi c expeditions of the early 19 th century (Cornell 1965, Fornasiero et al. 2010. Naturalists brought back from these distant and previously unexplored lands many new plant species, both as herbarium specimens and as living plants or seeds that were grown out mostly in the plant beds and greenhouses of the Muséum National d'Histoire Naturelle of Paris and in Josephine Bonaparte's gardens at Malmaison (Jangoux 2004, Fornasiero et al. 2010. Th e Baudin expedition lasted four years (1800-1804) and its explicit purpose was "observation and research relating to Geography and Natural History". Th e crew included 24 scientists and artists, among them were three botanists and fi ve gardeners that had been carefully selected by Antoine-Laurent de Jussieu, then director of the Muséum National d'Histoire Naturelle (Proust de la Gironière 2002, Jangoux 2004). By the time the Géographe and the Naturaliste reached Port Jackson (New South Wales) in June 1802 for a fi ve month stopover, most of the botanical team had either died or left the expedition; only one botanist, Jean-Baptiste Leschenault de la Tour, and one gardener, Antoine Guichenot, remained (Desmet and Jangoux 2010). After collecting in Australia and continuing with the expedition, in 1803 Leschenault was left behind in Timor to recover from illness (Proust de la Gironière 2002, Desmet and Jangoux 2010). After his recovery, he left Timor for Java, but found himself unable to return to France, probably due to instability in Europe at the time. Leschenault was off ered the protection of Nicolous Engelhard, the Dutch Governor of the northeastern coast of Java, and given the mandate to collect natural history specimens there (Van Steenis-Kruseman and Van Steenis 1950, Desmet andJangoux 2010). For two years (1804-1806) Leschenault visited the islands of Java and Madoera where he claimed to have collected ca. 900 plant species (Leschenault 1807), all of which were presumably sent back to the herbarium of the Muséum National d'Histoire Naturelle in Paris (P). Several duplicates of Leschenault's collections in other groups have been found in G, K and L (Van Steenis-Kruseman and Van Steenis 1950), but no catalogue of his collections exists and an accurate estimate of the extant number of collections has yet to be compiled.
In the course of preparing a monographic revision of the spiny solanums from tropical Asia (see Aubriot et al. 2016 for discussion of the Old World clade of subgenus Leptostemonum Bitter), we were unable to fi nd the type material for two spiny solanums from Java. Solanum gracilifl orum Dunal and S. poka Dunal were fi rst described by Michel-Félix Dunal in 1814 as part of the supplement of Lamarck's Encyclopédie Méthodique edited by Jean Poiret. He cited no herbarium material or collector but cited a drawing ("Dun. Suppl. Sol. tab."; Dunal 1814) from his then unpublished synopsis of Solanum (published later as Dunal 1816). In later treatments of these species Dunal (1816Dunal ( , 1852 stated that the collections he had seen were made by Leschenault during his stay in Java (1803)(1804)(1805)(1806). Th orough searches of the herbarium at P where Leschenault's collections are housed, as well as other herbaria (see Materials and Methods) where duplicates could possibly have been sent, have not revealed any original material upon which the drawings cited in the protologue were based. Toussaint François Node-Véran was the offi cial botanical artist of the Jardin des Plantes in Montpellier in the early part of the 19 th century (appointed in 1813 and stayed there until his death in 1852; Denizot et al. 1994) and worked closely with Dunal in preparing the illustrations for the intended major treatment of the taxonomy of Solanum (Knapp 2007). Several hundred pen and ink drawings of Solanum were made by Node-Véran during the preparation of Dunal's complete treatment of the genus that was never published in its entirety, but only as Solanorum Synopsis (Dunal 1816). Political instability in France during the years of the Napoleonic Wars of the early 19 th century and Dunal's not being appointed director of the Jardin des Plantes in Montpellier could be contributing factors in his failure to publish the complete illustrated volume (Dulieu 1994, Knapp 2007. Several of the species drawn by Node-Véran were drawn directly from herbarium specimens [e.g., S. arboreum Dunal, Lycopersicon hirsutum Dunal (=S. habrochaites S. Knapp & D.M.Spooner); see Knapp andSpooner 1999, Knapp 2007] that are currently in the herbarium at P. We expect he similarly used herbarium material from P (explicitly cited as herbarium material in Dunal 1816) as the basis for the illustrations of S. gracilifl orum and S. poka cited in the 1814 protologues (Dunal 1814). It is possible that specimens were lost during the turbulent times in Europe in the early 19 th century (see Knapp 2007).
Given that no plant specimens corresponding to the protologues have been found, despite extensive searches, we consider the unpublished Node-Véran drawings the most appropriate and only extant possibilities for lectotypifying both S. gracilifl orum and S. poka. Th ese two names have long been treated as confusing, or ignored; they have rarely been used (see below in each species treatment), and few herbarium specimens we have seen have been annotated with either name. Most specimens of the taxa we here recognise as S. gracilifl orum and S. poka have been annotated incorrectly as widespread weedy taxa (e.g., S. torvum Sw.) or with names we here consider synonyms (e.g., S. athroanthum Dunal); this refl ects the limited taxonomic work previously done on tropical Asian Solanum, whose taxonomy has not been revised in detail since Dunal's (1852) treatment for Candolle's Prodromus. Our purpose here is to secure the application of these names by designating lectotypes for S. gracilifl orum and S. poka, as well as providing complete morphological descriptions for these two species. We also designated interpretative types (epitypes), because details of trichome morphology are extremely important in spiny solanum taxonomy, and these are not visible on the illustrations.

Solanum gracilifl orum
Phenology. Th e few known collections were fl owering and fruiting between May and August.
Distribution and ecology. (Fig. 2) Known from the islands of Java, Bali, Sulawesi and Ambon (Indonesia); growing in forest understory; elevation not recorded on any herbarium material we have seen. Th e records (as S. athroanthum) from the island of Luzon in the Philippines (Merrill 1912, Merrill 1923) are based on misidentifi cations of specimens of S. trilobatum L.
Preliminary conservation status. Data Defi cient (DD); known only from seven collections, several of which are of uncertain localities. Solanum gracilifl orum has not been re-collected since the fi rst half of the 20 th century, indicating it is certainly of conservation concern. Recollection of this species and exploration of the type locality are priorities. Discussion. Solanum gracilifl orum is a poorly known species represented by very few collections that presents a combination of morphological features that makes it readily recognisable among tropical Asian spiny solanums. It is superfi cially similar to S. cyanocarphium Blume, a sympatric species that is distributed across the Sunda Shelf region, and to S. retrorsum Elmer, that occurs mainly in the Philippines. Solanum gra- cilifl orum can be distinguished from both of them by its much sparser indumentum, stout, deltate stem prickles (rather than slender and awl-shaped), and tiny, delicate fl owers (hence the species epithet) that are clustered in dense, many-fl owered infl orescences. Molecular data show that S. cyanocarphium and S. gracilifl orum are not closely related; S. gracilifl orum is nested within the Sahul-Pacifi c clade while S. cyanocarphium is an unresolved species of uncertain affi nities (see Aubriot et al. 2016).
Solanum gracilifl orum is the type of section Gracilifl orum (Dunal) Seithe, a section partly based on the informal grouping made in Dunal's (1852) treatment of Solanum in Candolle's Prodromus. In Seithe's (1962) circumscription, section Gracilifl orum included 14 species with stellate trichomes and acicular prickles coming from various region of the world (e.g., S. bahamense L. from the Caribbean archipelago, S. nienkui Merr. & Chun from Southeast Asia, S. paniculatum L. from South America, S. stelligerum Sm. from Australia). Symon (1981Symon ( , 1985 extended the circumscription of the section with the addition of 27 additional species (10 from Australia and 17 from New Guinea), expressing at the same time serious doubts about its coherence. Symon's concerns echoed those expressed in Whalen's systematic treatment of the spiny solanums (Whalen 1984). In this fi rst-ever attempt to include spiny solanums into a morphologically based phylogenetic framework, Whalen did not regard section Gracilifl orum as a natural group and placed members of the section as defi ned by Seithe (1962) into several of his informal groups (e.g., S. bahamense in the 'Solanum bahamense group', S. paniculatum in the 'Solanum torvum group', S. stelligerum in the 'Solanum ferocissimum group'). With limited sampling and knowledge of Old World taxa, Whalen did not clarify the identity of S. gracilifl orum, the type species of the section, and included it in his 'Unusual species group' as a possible synonym of the widespread tropical Asian species S. violaceum Ortega. He considered S. athroanthum to be diff erent from S. gracilifl orum, and placed the former into his 'Solanum dunalianum group' [= Solanum section Dunaliana (Bitter) Symon pro parte], a group of 20 species distributed across the Malayan archipelago, Australia and the South Pacifi c that were characterised by lack of broad-based prickles on mature growth, entire leaves with glabrate abaxial surfaces, infl orescences with tightly spaced hermaphroditic fl owers, and juicy red berries (Whalen 1984). More recently McClelland (2012) proposed a narrower circumscription of sect. Dunaliana, reducing it to six species and excluding S. gracilifl orum (as S. athroanthum) on the basis of its deeply lobed leaves with prickles on the principal veins and its slightly unequal anthers (versus entire to shallowly lobed non-prickly leaves and always equal anthers for all species he recognized as belonging to sect. Dunaliana). Instead he suggested a close relationship between S. gracilifl orum and S. nienkui, a Southeast Asian species that also displays anisandry. Recent molecular phylogenetic analysis of tropical Asian spiny solanums incorporating representatives of sections Dunaliana and Gracilifl orum (including S. dunalianum Gaudich. and S. gracilifl orum) showed S. gracilifl orum to be sister to the Philippine endemic S. lianoides Elmer (Aubriot et al. 2016). Both species are prickly vines, but S. lianoides diff ers from S. gracilifl orum by its denser leaf indumentum, entire leaves and larger fl owers. Both species are closely related to S. dunalianum (Aubriot et al. 2016), a result consistent with Whalen's (1984)  In the protologue Dunal referred to an illustration made by Node-Véran, 'Dun. Suppl. 7. Sol. Mss. tab. 4.', an orthographic error for ' Dun. Suppl. Sol. Mss. tab. 47.' according to the sequence of fi gure numbers and to the caption on the illustration in Montpellier. We were unable to fi nd any herbarium material matching the illustration in either P or MPU, although Dunal later (Dunal 1816(Dunal , 1852 cited Leschenault as the collector of the material he had seen. We designate the unpublished illustration of Node-Véran as the lectotype because it is the only extant original material we have identifi ed to date. We have also designated here an epitype specimen that best matches Node-Véran's illustration, and that corresponds to a collection made in the same geographical area as the lost type specimen (i.e. the island of Java in Indonesia) in order to secure the application of the name (Art. 9.8, McNeill et al. 2012). Dunal (1852) based his description of S. athroanthum on Zollinger 2907 in "hb. DC.". Th ere are two specimens of Zollinger 2907 in G-DC; we select the more complete of these as the lectotype. Th e locality data for Zollinger's collections are often not written on all duplicates; for Zollinger 2907 locality data are only found on P00368940. Description. Shrubs to 3 m, armed. Young stems terete, black to dark brownish, moderately stellate-pubescent, usually densely prickly distally, sometimes unarmed, the stellate trichomes porrect, sessile or variously stalked, the stalks to 0.2 mm long, the rays (4-)5-8, 0.1-0.25 mm long, the midpoints reduced to globular glands; prickles to 3.5 mm long, to 2.5 mm wide at base, straight, awl-shaped to deltate, conical, pale yellow, glabrescent; bark of older stems brownish gray, sparsely stellate-pubescent. Sympodial units difoliate, the leaves geminate. Leaves simple, the blades 11-24 cm long, 4-13 cm wide, ca. 1.5-3 times longer than wide, elliptic to broadly ovate, chartaceous, slightly discolorous; adaxial surface moderately stellate-pubescent with porrect, sessile and less often variously stalked trichomes, the stalks to 0.1 mm long, the rays 4-8, 0.1-0.4 mm long, the midpoints to 0.25 mm long; abaxial surface moderately stellate-pubescent with trichomes like those of the adaxial surface, but more often stalked; prickles 0-6 per leaf side, to 6 mm long, to 1.5 mm wide at base, straight or slightly curved at the tip, awl-shaped, conical, pale yellow, glabrous; major veins 6-8 pairs drying yellow; base shortly attenuate to truncate; margins entire or shallowly to deeply lobed, the lobes 1-5 on each side, 0.5-5 cm long, rounded to apically acute, the sinuses extending up to 2/3 of the distance to the midvein, deltate; apex acute; petiole 1.5-4 cm long, 1/10-1/5 of the leaf blade length, densely stellate-pubescent with porrect, sessile trichomes like those of the blades, with 0-5 prickles like those of the stems. Infl orescences apparently lateral or leaf opposed, 2-5 cm long, unbranched to up to 2 times branched, with ca. 5-20 fl owers, moderately to densely stellate-pubescent, unarmed; peduncle 0.5-1.5 cm long, with 0-1 prickles; pedicels 0.5-1.2 cm long, erect, articulated at the base, densely stellate-pubescent, unarmed; pedicel scars spaced 2-4 mm apart. Flowers 5-merous, apparently all perfect. Calyx 4-7 mm long, campanulate, moderately stellate-pubescent, densely stellate-pubescent on the midvein, unarmed, the lobes 3-5 mm long, the lower part deltate and abruptly constricting to an elongate acumen, the acumen 3/4 the total lobe length, the abaxial surface more or less strongly keeled along the midvein. Corolla 1-2 cm in diameter, white, lobed for ca. 1/2-2/3 of the way to the base, the lobes 5-8 mm long, 2-3.5 mm wide, deltate, spreading at anthesis, densely stellate-pubescent abaxially on parts exposed in bud. Stamens equal; fi lament tube < 0.5 mm long; free portion of the fi laments 0.75-1.5 mm long; anthers 5-6.5 mm long, ca. 0.75 mm wide, connivent, tapering, poricidal at the tips, the pores not lengthening to slits with age. Ovary conical, minutely glandular-puberulent; style 0.6-1 cm long, slender, curved at the apex, with few scattered hairs at the tip; stigma capitate, minutely papillate, stellate-pubescent. Fruit a globose berry, 8-18 per infrutescence, 0.8-1.5 cm in diameter, the pericarp smooth, bluish green when young turning to dark greyish yellow, glabrous; fruiting pedicels 1.2-2.5 cm long, ca. 1-1.5 mm in diameter at the base, ca. 2-3 mm in diameter at the apex, woody, erect, unarmed; fruiting calyx lobes not expanding. Seeds 100-200 per berry, ca. 1.75-2 mm long, 1.5-1.75 mm wide, fl attened reniform, pale yellowish, the surface minutely pitted, the testal cells sinuate in outline.

Solanum poka
Phenology. Flowering and fruiting throughout the year. Distribution and ecology. (Fig. 2) Widely distributed in the Malay Archipelago, from western Sumatra to the Maluku Islands and across Sulawesi, northwards to the Talaud islands; growing in open woodland, forest edges, degraded vegetation, usually on limestone or volcanic rocks; 0-1600 m elevation.
Preliminary conservation status. Least Concern (LC); EOO > 100,000 km 2 and AOO > 10,000 m 2 (see Moat 2007 for explanation of measurements). Although the EOO and AOO measurement indicate a status of least concern, the few collections coupled with the profound transformation in lowland Indonesian habitats where S. poka is found (Margono et al. 2014) suggest that the species is a priority for recollection and reassessment. Discussion. Solanum poka was long ignored after its fi rst publication (Dunal 1814). It has not been included in classical fl oristic treatments of Java (Hasskarl 1848, Backer 1965, van Steenis et al. 2006 or Sumatra (Miquel 1862). It was mentioned by Miquel (1856) and Koorders (1912), but both authors merely repeated Dunal's original description, without referring to any specimens. In Koorders's (1918) botanical report on the fl ora of northeastern Sulawesi he lists several widespread and common species (e.g., S. lycopersicum L., S. melongena L., S. torvum Sw., S. tuberosum L.) as well as two shrubby Solanum species for which he did not provide names ("Solanum spec. A" and "Solanum spec. B"). Two previously undetermined Koorders collections of S. poka from northeast Sulawesi (Minahasa Regency) in April 1895 (Koorders 18035B and Koorders 18037B, both L) correspond to S. poka. It is possible that these two collections correspond to one (or both) of Koorders' (1918) unnamed species, but since he provided no descriptions or specifi c localities this is diffi cult, if not impossible, to ascertain.
Based on morphology, S. poka belongs to the Torva clade (sensu Stern et al. 2011), with its straight prickles, many fl owered infl orescences and corollas with abundant interpetalar tissue (see Fig. 1b). Th is hypothesis is corroborated by the molecular data (Aubriot et al. 2016). Solanum poka is sister to a clade composed of four native Old World species (Solanum dammerianum Lauterb. & K.Schum, S. peikuoense S.S.Ying, S. pseudosaponaceum Blume, S. torvoideum Merr. & L.M.Perry) with which it forms a strongly supported group, the 'Old World torvoids' sensu Aubriot et al. (2016). Morphologically, S. poka most closely resembles S. pseudosaponaceum, a widespread species from Taiwan and southern China to Indonesia, but diff ers in having denser indumentum on the adaxial leaf surface, more numerous straight prickles on the upper stems, fewer, larger fl owers with elongate strongly keeled calyx lobes, and much larger fruits. Flowers of S. pseudosaponaceum are lilac or purplish-white while those of S. poka are always described on labels as white.
In the protologue Dunal referred to an illustration made by Node-Véran, 'Dun. Suppl. 7. Sol. Mss. tab. 55', but cited no herbarium material. Similarly to the situation of S. gracilifl orum, we were unable to fi nd any herbarium material matching the illustration in either P or MPU, although Dunal later (Dunal 1816(Dunal , 1852 cited Leschenault as the collector of the material he had seen. We designate the unpublished illustration of Node-Véran as the lectotype because it is the only extant original material we have identifi ed to date. We designate here an epitype specimen from Java, the cited type locality, (Horsfi eld s.n., BM000886306) that best matches Node-Véran's illustration, particularly with respect to the diagnostic characters for S. poka; leaf shape, prickle shape and calyx lobe morphology.
We have only seen three specimens of S. poka from Java, the cited type locality, all collected by Th omas Horsfi eld, an American physician who collected on Java contemporaneously with Leschenault in the early part of the 19 th century (McNair 1942, Van Steenis-Kruseman andVan Steenis 1950). Solanum poka is, however, rather broadly distributed across the Malay Archiplago, with the distribution centred on Sulawesi and the surroundings islands (Malaku Islands, Talaud Islands) (Fig. 2). Th orough examination of the extensive holdings in Indonesia (particularly those of the Bogor Botanical Garden Herbarium, BO) and, given the historically extensive natural habitat loss recorded for Java (Margono et al. 2014), additional collecting are both needed to better understand the distribution of S. poka. Dunal (1852) cited the herbarium name 'S. quercifolium Banks' taken from a specimen in BM collected by Joseph Banks in Java as part of his treatment of S. poka. Examination of this sheet (BM000886238) shows it belongs to S. pseudosaponaceum.