Euphorbia milii Des Moulins, 1826

Haevermans, Thomas & Hetterscheid, Wilbert L. A., 2021, Novelties in Malagasy Euphorbia (Euphorbiaceae), Phytotaxa 488 (1), pp. 1-63 : 4-60

Euphorbia milii Des Moulins nomenclatural and taxonomic reassessment.

The precise identity of the famous crown of thorns, Euphorbia milii Des Moulins , has remained a mystery since its original publication. The protologue and accompanying drawing (the lectotype, see Fig. 1B) leave ample room for ambiguity as to the identity of the material behind the name and thus the application of the name itself. The name therefore has been applied to various shrubby spiny euphorbias with reddish cyathophylls, and several taxa of infraspecific ranks have been created to accommodate the extensive diversity of the group in Madagascar. Several similarly spiny species (e.g., Rauh 1970a) were proposed and related to a “ Euphorbia milii group”, without defining whether this concept equals Euphorbia section Goniostema Baill. ex Boiss. as a whole or just the most closely related taxa, at a time when no molecular phylogenies were available to test the hypotheses. This instability in taxonomic ranks used for the new taxa, as well as the apparent randomness with which they were applied, emphasizes the ambiguity in the correct application of the name E. milii and thus the identity of the plant material identified by it. Here we propose a solution to the problem by analysing all available pieces of evidence and we reinforce our interpretation by assigning an epitype to the lectotype of the name E. milii .

It seems that most, if not all, authors on the taxonomy of E. milii and its infraspecific taxa have overlooked the fact that Des Moulins described truncate leaves for his species (along with more oval/elliptic ones on the same plant) and clearly shows them in his drawing ( Des Moulins 1826: Pl. 1, see Fig. 1B, here), until now the type of the species’ name (lectotype designated in Newton 1994: 86). Even when this is acknowledged, the protologue and type drawing leave ambiguity, but with our observations, the possible number of candidates for the correct application of the name was reduced to four in our opinion, of which finally one turned out to be the most similar one to Des Moulins’ intention.

First off, there is the invalidly published designation ‘ imperatae’, standing for low, bushy plants with either exclusively rotund leaves or with a mix of rotund and short obtriangular leaves (see, e.g., Laius 2019: fig. 28). Rauh (1967a) discussed this taxon extensively based on observations in the direct surroundings of Manantenina (Mahatalaky), where Leandri’s described plant also came from ( Leandri 1946, 1952, Ursch & Leandri 1954). The specimen on which Leandri based his concept is Decary 3863 (MNHN-P-P00221940!, see Fig. 15). The very small, fleshy leaves and cyathophylls and the strongly plagiotropic growth (see Fig. 16) led us to reject this variety as a suitable choice for true E. milii , as it is not in accordance with Des Moulins’ original description. We consider this taxon to be a distinct entity deserving recognition at the specific rank (see under E. imperatae ).

The second candidate, based on the possession of obdeltoid, truncate and broadly elliptic/oval leaves, is E. bevilaniensis Croizat (1934: 96) . It is diagnosed to be a slender plant of about 5 feet (ca. 1.50 m) high. However, we also reject this species as a candidate for the correct application of the name E. milii because of its typical leaf type not being drawn or described by Des Moulins (1826).

A third candidate we considered is the invalidly published E. beharensis var. truncata Rauh (1999b: 11) , but this taxon has no oval/elliptic leaves in addition to truncate ones and very tiny flowers and very small leaves that are not compatible with the concept of E. milii as originally published by Des Moulins.

In herbarium material from P we found a fourth candidate: a set of herbarium specimens with the required robustness and mix of elliptical and truncate leaves, all growing in the vicinity of Fort-Dauphin.Among these specimens, one specimen best represented E. milii in our opinion: the specimen Decary 10139 (MNHN-P-P00221947!) from Toalagnaro (Fort-Dauphin), Pointe Itaperina (see Fig. 5).

From the above it is clear that we have concentrated our search for “true” E. milii in the southeast of Madagascar in the Anosy Province, where Toalagnaro is located. One argument follows from the observation that no truncateleaved Euphorbia species from the milii -alliance have been found elsewhere on Madagascar, but there is also a very strong circumstantial piece of evidence pointing to southeast Madagascar for the area of origin of the true E. milii . The plant used by Des Moulins to describe his E. milii was brought to Bordeaux Botanical Garden by Baron Pierre Bernard Milius, then Governor of Réunion Island, to whom Des Moulins dedicated the taxon’s specific epithet. Records of the activities of Mr. the Baron Milius for the years 1818–1821 show that the officers mandated by him travelled along the eastern coast of Madagascar ( Valette 1962), from Ile Sainte Marie south to Fort-Dauphin ( Valette 1971), making it very probable that they collected E. milii from this area. In search of possibly surviving material from this time to help apply the name E. milii , we had the opportunity to study original material from Des Moulins’ herbarium in the BORD herbarium (Bordeaux, France). The specimen “BORD_DESM_081_Euphorbes exotiques” found there (reproduced in Fig. 4C ), bore on the label mentions by Des Moulins’ own handwriting added at various times. The oldest one says that this represented his new species E. milii , and that it was pressed from the Botanical Garden’s collection from a plant brought back by Mr. the Baron Milius. The specimen branch and leaves (without the flowers) also matched perfectly the drawing in Des Moulins (1826) selected as lectotype by Newton (1994: 86). This specimen can be considered part of the original material because it obviously was used to draw the figure in the protologue selected as lectotype, even if the specimens mostly bears posterior writing. As the lectotype has been already designated from the syntypes, we chose to designate this specimen as the epitype of E. milii (see taxonomic treatment).

The identity of two other related species and described around the same time which were used interchangeably as scientific names for the crown of thorns can now be clarified, namely, Euphorbia splendens Bojer ex Hooker (1829: t. 2902) from the Imerina highlands, and E. bojeri Hooker (1836 : t. 3527), from a cultivated plant of undocumented origin (see reproductions of the lectotypes in Fig. 2). Previously treated as a variety of E. milii (Ursch & Leandri 1954) , we propose that E. splendens is a taxon of specific rank, originating from the northern part of the Imerina Plateau (“Emirne” of Bojer 1829), while the illustration and protologue of E. bojeri clearly show that it is conspecific with E. milii , showing the same mixture of both truncate and elliptic-oval leaves on the same plant. The results of these observations and considerations are summarized in the taxonomic treatment.

The epithet milii has been used frequently in horticultural literature and trade for a widely variable swarm of spiny cultivars generally designated as crown of thorns. The epithet milii is applied nearly always irrespective of any known details of the botanical background of the parental plants or merely based on badly informed guesses ( Jankalski 2000). The present re-interpretation of a number of names of species in E. sect. Goniostema does not seriously disrupt the nomenclature and classification of cultivars derived from species in this section. Firstly, there is the non-imperative use of species epithets in full cultivar names, as cultivar epithets need to be unique in their UPOV (Union pour la Protection des Obtentions Végétales) or ISHS (International Society for Horticultural Science) denomination class (for explanatory notes, see Brickell et al. 2016: 13, Art. 6), which is usually a genus, as in this case Euphorbia . Thus, a full name of a cultivar is principally a genus name plus a cultivar epithet, not necessarily accompanied by a species epithet. This serves to distance the nomenclature of domesticated plants from that of plants found in the wild and governed by ICN and serves the stability of nomenclature of domesticated plants, which is a different realm than that of plants in nature (see Hetterscheid & Brandenburg 1995a, b). Sensible classification and derived nomenclature of domesticated plants can be achieved by using the Group (cultivar-group) concept as explained in Brickell et al. (2016: 10, Art. 3), but incompletely applied to the spiny Euphorbia -cultivars by Jankalski (2000). Although Jankalski uses the cultivar-group as a classification tool, he still uses a latinised Linnean binomial: E. × lomi Rauh (1979: 257), as basis, whereas the name Euphorbia would have sufficed (e.g., Euphorbia Poysean Group instead of Euphorbia × lomi Poysean Group). This modus operandi circumvents instability of names, which is unavoidable when only nomenclature according to ICN is applied. Also, the name E. × lomi is defined as covering only hybrids of E. lophogona Lamarck (1786: 417) and E. milii , whereas it is well-known that in many of these cultivars, other species than these two are involved. Again, an argument against using restrictive Linnean binomials for domesticated plants.

Taxonomic treatment

Euphorbia milii Des Moulins (1826: 27) (Figs. 1AB, 2A, 3, 4, 5, 6).— Euphorbia breonii Noisette (1833: 189) , nom. illeg. superfl. [superfluous renaming of Euphorbia milii Des Moulins , see Newton (1994)]. Lectotype (designated by Newton 1994: 86):—[ MADAGASCAR.] Illustration (t. 1) in Des Moulins (1826) [here reproduced in Fig. 1B]. Epitype (designated here):— MADAGASCAR. “Jardin Botanique de Bordeaux. Pieds donnés par M. le Baron Milius, de Magadascar”, s.dat., Des Moulins s.n. (BORD_DESM_081_Euphorbes exotiques BORD!) [reproduced here in Fig. 4C ].

— Euphorbia bojeri Hooker (1836 : t. 3527).— Tumalis bojeri (Hook.) Rafinesque (1838: 114) . — Sterigmanthe bojeri (Hook.) Klotzsch & Garcke (1859: 252, nom. nud.; 1860: 100).— Euphorbia splendens var. bojeri (Hook.) Costantin & Gallaud (1905: 353 , in clav.).— Euphorbia splendens subsp. bojeri (Hook.) Denis (1921: 84) . Type:— MADAGASCAR. Illustration in Curtis’s Botanical Magazine 63: t. 3527 (1836) (lectotype!, here designated) [reproduced here in Fig. 2A].

Additional specimens examined:— MADAGASCAR. Toliary : Anosy , Taolagnaro , Mandromodromotra, “Pointe Itaperina (District de Fort-Dauphin)”, [24°55′0″S 47°7′60″E], s.dat., Decary 10139 (MNHN-P-P00221947! reproduced here as Fig. 5; L!). “Cap Evatra (District de Fort-Dauphin)”, [24°58′60″S 47°6′0″E], 25 October 1932, Decary 10874 (MNHN-P-P00217866!). Ranopiso, “Mont Ambohibato”, [25°4′0″S 46°37′0″E], 16 October 1901, Grandidier s.n. (MNHN-P-P02284082!). Soanierana, “Environs de Fort-Dauphin, forêt de Manantantely”, 180 m elev., [24°58′60″S 46°55′0″E], 22 September 1928, Humbert 5806 (MNHN-P-P00217905!). Ampasy, “Inselberg: Ambondrondria, FKT: Mangaika, Village: Enato, Com.Ampasy-Nampoana, Rég:Anosy”, 449 m elev., 24°53′38″S 46°58′58″E, 30 November 2010, Ramandimbisoa 135 (MO!, P!, TAN) .

Note:— Bréon (1820: 55) mentions (without scientific names) two euphorbias from Madagascar cultivated in Réunion Island at the Jardins Botanique et de Naturalisation in Saint-Denis: a “ euphorbe à crêtes ” [ E. lophogona from Fort-Dauphin–Sainte-Luce area in Madagascar], and a “euphorbe élégante” which may be E. milii , the latter likely being the plant published as E. breonii by Noisette (1833) as an avowed renaming of E. milii . However, Noisette (1833) says the plant he describes as E. breonii is a plant found by M. Bréon in Madagascar in 1822 and brought back to France by Mr. Neuman in 1824 and cultivated at the Jardin des Plantes in Paris (Muséum national d’histoire naturelle). The color plate in Noisette (1833), reproduced here in Fig. 1A, clearly represents E. milii as now understood ( Newton 1994). As a result of our choice, E. milii can shortly be diagnosed, based on measurements made on the epitype, to be a bushy, erect, spiny, much branching plant, up to ca. 1.5 m high. The spines are hard, straight and at a straight angle with the branches, up to 15 mm long. The leaves are both oblong and obtriangular on the same plant, mucronate, darker [green] on the adaxial side, paler on the abaxial side (as described by Des Moulins), 10–18 × 10–30 mm, up to 3 times the length of the spines as seen on the lectotype. The cyathophylls, ca. 8 × 10 mm (wider than long), also show the difference in color described by Des Moulins (1826).

Euphorbia splendens Bojer ex Hooker (1829: t. 2902) (Figs. 2B, 7, 8).— Euphorbia milii var. splendens (Bojer ex Hook.) Ursch & Leandri (1954: 148) .— Lacanthis splendens (Bojer ex Hook.) Rafinesque (1837: 94) .— Sterigmanthe splendens (Bojer ex Hook.) Klotzsch & Garcke (1859: 252, nom. nud.; 1860: 100 [definitive association with the generic name Sterigmanthe ]). Type:— MADAGASCAR. Illustration in Curtis’s Botanical Magazine 56: t. 2902 (1829) (lectotype!, here designated) [reproduced here in Fig. 2B]. Epitype (designated in Haevermans et al. 2009: 291):— MADAGASCAR. Antananarivo: Vakinankaratra, Faratsiho, Ambatoasana Valabetokana, “Hab. ad versum agrorum in prov. Emerina”, [19°22′24″S 46°42′22″E], s.dat., Bojer s.n. (K000253185!, isoepitypes: MNHN-P-P00078105!, G-DC!)

— Euphorbia splendens Bojer ex Hook. forma lutea Leandri (1952: 112) , nom. invalid. [no Latin diagnosis].

Additional specimens examined:— MADAGASCAR. Antananarivo: “ Nanisana ”, [18°52′60″S 47°34′0″E], April 1906, Aleizette 596 (MNHN-P-P00217892!). “Nanisana, near Tananarive”, [18°53′11″S 47°32′52″E], s.dat., Aleizette s.n. (L-L.2215637!). “ Environs de Tananarive , en terrains rocheux à Soanierana”, s.dat., Alleizette s.n. (L-L.2216978!). “Madagascar - Madagascar/Faritanin Antsiranana ”, [19°22′24″S 46°42′22″E], s.dat., Berger 41 (G-DC!). Analamanga, Ankazobe , “PK 44, route de Majunga”, 1500 m elev., [18°37′0″S 47°15′0″E], 16 June 1974, Cremers 3224 (MNHN- P-P02284090!). “Vicinity of junction of Route #4 to Majunga and road to Ivato, northwest of Tananarive ; roadside”, 1400 m elev., [18°49′60″S 47°27′0″E], 18 January 1975, Croat 28669 (MO!). “Tananarive (Ankatso)”, [18°55′0″S 47°31′60″E], 2 February 1924, Decary 2510 (MNHN-P-P00221989!). 2 February 1924, Decary 2515 (MNHN-P- P00221990!). 23 September 1928, Decary 6798 (MNHN-P-P00221996!). Bongolava, Fenoarivo-Centre, “Tampoketsa, au Nord-Est de Fenoarivo ”, [18°15′0″S 46°40′0″E], 16 March 1930, Decary 7595 (MNHN-P-P00217864!). “Environs d’Ankazobe”, [18°19′0″S 47°6′0″E], 23 March 1930, Decary 7700 (MNHN-P-P00221999!). “Ambohimalaza, près d’Ankazobe”, [18°26′60″S 46°34′60″E], 25 March 1930, Decary 7727 (MNHN-P-P00222000!). “Ilafy”, [18°51′0″S 47°34′0″E], 4 February 1917, Decary s.n. (MNHN-P-P00221988!). s.loc., s.dat., Desvaux s.n. (MNHN-P-P00217918!). “Centre: Manerinerina sur le Tampoketsa , entre Ikopa et Betsiboka ”, 1600 m elev., [18°1′0″S 47°8′60″E], December 1923, Perrier de la Bâthie 16761 (MNHN-P-P00217931!, L-L.2216979!). “ C. Cultivé aux environs de Tananarive”, Perrier de la Bâthie 18578 (MNHN-P-P00217934!). “Centre: Rocailles et basaltes dénudées du Ketsa (Amberimay)”, December 1910, Perrier de la Bâthie 8058 (MNHN-P E], April 1917 -P00217928!). s.loc., s.dat., Verbeke 542 (BR!). “Tananarive”, [18°55′0″S 47°31′60″E], January 1916, Waterlot s.n. (MNHN-P-P00217899!, MNHN-P-P00217900!) .

Note:— From our observations of herbarium material, living plants and the first author’s own herbarium collections made in Madagascar, we conclude that E. splendens only occurs naturally (and is cultivated as well) on the northwestern Imerina Plateau (see plants in Fig. 7, and distribution in Fig. 8 ). We propose E. splendens to include spiny shrubby plants with red, yellow and several intermediate cyathophyll colours and elliptical or obovate, non-succulent leaves with a varying degree of hairiness on the abaxial leaf surface or glabrous. We consider this variation intrinsic to E. splendens at species level, as fieldwork conducted by the first author in the area demonstrates, and therefore we see no reason to recognise infraspecific taxa to name this variation. Two names validly published by Drake have been mentioned as being synonyms of E. splendens : Euphorbia melanacantha Drake (1903: 45) and Euphorbia platyacantha Drake (1903: 45) . This synonymy, mentioned in Denis (1921: 82), has been followed since that date in the various treatments published thereafter. Euphorbia melanacantha Drake is poorly known but nevertheless validly published, and a (very damaged and uninterpretable) fragment of the type specimen survived from Tsimalaha, which is far beyond the known distribution of E. splendens as we now understand it. Concerning E. platyacantha , no original material survived to help interpret Drake’s description of a plant reputedly from the Horombe plateau, along the Ranohira to Ihosy road. Several distinct taxa exist in this area and the description of E. platyacantha and E. melanacantha are too imprecise and compatible with several of the spiny taxa growing there. Without being able to ascertain the correct application of these names, we consider these two names doubtful though definitely not part of the synonymy of E. splendens .

Review of taxa previously subordinate to E. milii and E. splendens

Under both E. milii and E. splendens , several varietal names have been published, fluctuating with the interpretation of each of these names. With our concepts of E. milii and E. splendens presented here, we re-examined all these varietal names and materials, both in herbaria and in several living collections, and material seen and collected by the first author in Madagascar during several expeditions. We discuss their taxonomic status and names, leading amongst others to the recognition of a number of new species and several status changes from variety to species level and accompanying nomenclature. Invalid names are mentioned and left as is.

Euphorbia betrokana Castillon & Castillon (2020: 12) (Figs. 9A, 10). Replaced synonym:— Euphorbia milii var. longifolia Rauh (1967c: 182) . Type:— MADAGASCAR. Toliary: Anosy, Betroka, Ianabinda, “In collibus gneissiacis 20 km in directione meridionali a Betroka distantibus”, [23°25′40″S 46°0′21″E], December 1959, Rauh M1464 (holotype: HEID!, in spiritu).

Description:—Habit bushy, branching from the base, to ca. 1 m high, branches 80–100 cm long, almost terete, 1.5–2 cm thick, rebranching infrequently, at first erect, then bending horizontally, carrying numerous brachyblasts 0.5– 3 cm long, bark grey. Stipular spines simple, 1–1.5 cm long, base slightly thickened, brown when young, later grey to greyish violet. Leaves crowded at the tips of branches and on the brachyblasts; those on the main branches lanceolate, 10–20 × 0.8–1.5 cm, very short petiolate, top short apiculate, adaxial surface glabrous, green to greyish green, abaxial surface pale green, glabrous, main vein strongly raised, those on the brachyblasts 0.5–1 × 3–4 cm. Inflorescences subterminal, single or few, common peduncle 1–6 cm long, 3–4 mm thick, sticky, reddish, carrying 4–16(–32) cyathia. Cyathophylls 7 × 8–9 mm, spreading, top rounded, short apiculate, adaxial surface yellow, abaxial surface pale yellow. Cyathium 3 × 2.5 mm, slightly urceolate. Cyathial glands 5, transversely boat-shaped, yellow or orangish yellow. Pistillate flower: Ovary glabrous, style 2 mm long, the lower 2/3 rd fused. Interglandular bracts wedged-shaped, upper margin densely dentate-fimbriate. Fruits and seeds not known.

Etymology:—The new name is referring to the type locality, Betroka, around which the plant occurs.

Additional specimens examined:— MADAGASCAR. Toliary: Anosy, Betroka , Nanirona, “Anosy-Betroka. Colline rocheuse à 12 km au sud de Betroka, sur le bord de la RN 13”, 1011 m elev., 23°21′54″S 46°2′17″E, 22 November 2009, Aubriot et al. 19 (MICH!, MNHN-P-P00685098!). Aubriot et al. 20 (MNHN-P-P00685099!) .

Note:— Plants in cultivation usually develop slightly larger leaves and fewer brachyblasts. Rauh (1967c) published the name E. milii var. longifolia for plants seen 20 km N of Betroka (fide Rauh notebook entries HEIDRAUHFDB002_132 HEIDRAUHFDD043_ 138 in The Werner Rauh Heritage Project at http://wrhp.cos.uniheidelberg.de/scriptorium/index.php). We agree with Castillon and Castillon (2020) that this taxon is of specific rank and not fitting the morphological variation of E. milii as we now understand it, but the name E. longifolia is already occupied by several illegitimate homonyms and one rejected name (cf. Whitehouse 2006: 1046). The “ type plant” still grows in Heidelberg Botanical Garden under the nr. 140143, with duplicate grown from cuttings in the collection of the Nationaal Bomenmuseum Gimborn, the Netherlands (National Tree Museum Gimborn) (illustrated here in Fig. 10). For a more extensive description see Rauh (1967c): “ Euphorbia milii var. longifolia ”.

Euphorbia bevilaniensis Croizat (1934: 96) (Figs. 6, 12).— Euphorbia splendens var. bevilaniensis (Croizat) Leandri (1946: 159 , in clav. adnot. 3).— Euphorbia milii var. bevilaniensis (Croizat) Ursch & Leandri (1954: 150) . Lectotype (designated here):— MADAGASCAR. Toliary: Fort-Dauphin, Bevilany, “growing on gneiss in the forest, collected at Bevilany between the regions of Anosy and Androy” ( Fig. A , p. 97 in Croizat 1934 in The National Horticultural Magazine 20). Epitype (designated here):— MADAGASCAR. Toliary: Anosy, Taolagnaro, “Bevilany, limite de l’Anosy et de l’Androy. Forêt gneissique”, [25°0′0″S 46°36′0″E], 14 November 1932, Decary 10956 (MNHN-P- P00077911!) [reproduced here in Fig. 12].

Additional specimens examined: — MADAGASCAR. Toliary: Anosy, Taolagnaro, “Prov. Tuléar: Along route n° 10, 16 km W of Manambaro ”, 110 m elev., [25°3′0″S 46°40′0″E], 21 February 1975, Croat 31943 (MO!, MNHN-P- P00221946!) . “Parcelle 3 Ankazofotsy, Fort-Dauphin, Andohahela RNI”, 175 m elev., 25°0′60″S 46°37′60″E, 22 June 1994, Eboroke 827 (MO!, MNHN-P-P00275227!, TAN) . “Toliary, Toalagnaro (Ft. Dauphin), on RN 13, marker #51 from Toalagnaro, in transitional zone between eastern rain forest and spiny forest, exposed hillsides”, 100 m elev., 25°1′9″S 46°38′10″E, 4 August 1995, Edmondson 95-64 (MO!, MNHN-P-P00221943!, TAN) . “Toliary, Toalagnaro (Fort-Dauphin), W of town along highway on southern boundary of Reserve Integral N°11, Andohahela, Parcelle3, W of Col de Ranopiso ”, 130 m elev., 25°1′0″S 46°38′60″E, 27 January 1990, McPherson 14920 (MO!, MNHN-P- P00221944!, MNHN-P-P00708210!, TAN) . “Reserve d’Andohahela, parcelle 3; transitional forest with Neodypsis decaryi ”, 100 m elev., 25°1′60″S 46°40′0″E, 17 February 1990, Phillipson 3535 (MO!, MNHN-P-P00221945!, TAN)

Note:— Haevermans et al. (2009) chose Decary 10956 (MNHN-P-P00077911!, see Fig. 12), to be the holotype of the name E. bevilaniensis but this was in error, as the drawing with Croizat’s protologue should have been designated as the lectotype. Croizat (1934) based the protologue on material brought to him in alcohol by R. Decary (Decary s.n., in spiritu), collected from Bevilany and consisting of the distal part of a twig with leaves and flowers but we could not trace it in any herbarium, nor was a collection number of Decary mentioned. Although Decary 10956 (MNHN-P- P00077911!) has no formal status under the nomenclature code as it was not seen by Croizat, the specimen is of some historical importance as it is the only collection by Decary in P matching Croizat’s protologue. Croizat’s diagnosis is based on Decary’s notes, with added observations by himself.Also, the leaves as drawn by Croizat in the lectotype (the illustration in the protologue) with some of them having a double constriction (leaves that from top to bottom show a pronounced, wedge-shaped upper part, then a sudden narrowing to a more or less narrowly obtriangular part and then a sudden narrowing to the short leaf stalk), perfectly match leaves on the Decary specimen. We therefore designate this specimen as epitype of E. bevilaniensis .

Euphorbia hislopii Brown (1913: 304) ( Figs. 8 , 13, 14).— Euphorbia splendens Bojer ex Hook. var. hislopii (N.E.Br.) Leandri (1946: 159) .— Euphorbia milii Des Moul. var. hislopii (N.E.Br.) Ursch & Leandri (1954: 150) . Type:— [ MADAGASCAR]. “Specimen cult. at K from cutting brought from Durban Bot. Gard. in 1911. Prob. native of Madagascar”, s.dat., Hislop s.n. (holotype: K000185057!) [reproduced here in Fig. 14].

— Euphorbia milii Des Moul. var. breonii sensu Ursch & Leandri (1954: 148 , Fig. 38).— Euphorbia splendens Bojer ex Hook. var. breonii sensu Leandri [(1946: 159), and (1952: 114)], non E. breonii Noisette (1833: 189) [a superfluous renaming of E. milii Des Moulins ], syn. nov. .— Euphorbia splendens Bojer ex Hook. var. breonii Leandri ex Castillon & Castillon (2020: 18) , syn. nov. Type:— MADAGASCAR. “Cultivée dans le jardin botanique de Mr François à Tananarive de pieds provenant des environs de Fianarantsoa ”, s.dat., Perrier de la Bâthie 18577 (Holotype: MNHN-P-P00221939!).

— Euphorbia milii Des Moul. var. vulcani “Leandri” in Ursch & Leandri (1954: 150), nom. invalid. [no Latin diagnosis or description].— Euphorbia splendens Bojer ex Hook. var. vulcani Leandri (1946: 159) , nom. invalid. [no Latin diagnosis or description], syn. nov. .— Euphorbia splendens Bojer ex Hook. var. vulcanii Leandri ex Castillon & Castillon (2020: 18) [as “vulcani”], syn. nov. Type:— MADAGASCAR. Antananarivo: Vakinankaratra, Betafo, Betafo, “Betafo: sur les laves”, [19°49′60″S 46°51′0″E], 24 November 1938, Decary 13797 (holotype: MNHN-P-P00078123!).

Additional specimens examined:— MADAGASCAR. Antananarivo: “ Env. de Tananarive , Nanisana”, [18°52′60″S 47°34′0″E], September 1906, Aleizette 1215m (MNHN-P-P00217921!) . Fianarantsoa: Amoron’i mania, Ambositra, Tsarasaotra , “Fokontany Tranobongo, Inselberg en bon état d’Ambohitrinimasina. Echantillons préservés en alcool”, 1361 m elev., [20°25′26″S 47°10′18″E], 12 March 2010, Andriantiana 657 (MO!, TAN) . Ambatofinandrahana , 1700 m elev., [20°30′29″S 46°30′59″E], September 1956, Bosser 9899 ( TAN!) . Itremo , “Ouest d’Ambatofinandrahana”, [20°34′60″S 46°37′60″E], 19November 1939, Decary 15117 (MNHN-P-P00221933!) . Antananarivo:“ Tananarive, dans un jardin”, [18°55′0″S 47°31′60″E], 2 January 1924, Decary 2376 (MNHN-P-P00221991!, MNHN-P-P02284076!). “Soavinandriana”, [19°10′0″S 46°43′60″E], 27 April 1917, Decary s.n. (MNHN-P-P00221937!). “Rocaille du Jardin Botanique”, 26 March 1944, Jardin Botanique 6262 (MNHN-P-P00217914!), 26 March 1944, Jardin Botanique 6263 (MNHN-P-P00217915!). “Parc Botanique, échantillon fait par E. Ursch au Jardin Botanique de Tananarive”, s.dat., Leandri 2388 (MNHN-P-P00078043!). “C: De Tananarive; couramment planté en Imerina”, July 1910, Perrier de la Bâthie 9687 bis (MNHN-P-P00217879!) . Fianarantsoa: Amoron’i mania, Ambatofinandrahana, Itremo , “C: Environs d’Itremo (W Betsileo)”, 1500 m elev., [20°34′60″S 46°37′60″E], February 1919, Perrier de la Bâthie 12456 (MNHN- P-P00217881!). “Amoron’i Mania, Faliarivo Anjoman’Ankona, Andakatanikely, Vinany, Ampandrasa, Inselberg Vohibatazana. Sur le sommet de l’Inselberg”, 1609 m elev., 20°38′5″S 47°2′30″E, 13 December 2015, Rabarimanarivo 565 (MO!, TAN) . Ambositra, Tsarasaotra , “Fokontany: Tsarasaotra, Inselberg: Ambohitrinimasina”, 1490 m elev., 20°25′35″S 47°10′12″E, 8 December 2010, Ravololomanana 158 (MO!, TAN) .

Note:— The epithets “ hislopii ”, “ breonii ” and “ vulcanii ”, presented in several combinations used in Euphorbia , have been sources of considerable confusion. The name E. hislopii seems to be quite straightforward, since it is based on material at the time in cultivation and then not known from the wild. It was later reclassified as a variety of either E. milii or E. splendens (see above). Rauh (1995: 229–230) mentioned that he found it “between Antsirabe and Fianarantsoa ”, but the pictures he presented (Figs. 648, 650) are from plants cultivated as hedges in Antananarivo. Several specimens have been collected in the area mentioned by Rauh (see Additional specimens examined). These plants match perfectly with the protologue and holotype. Rauh (1995: 230) suggested that E. milii var. hislopii best be synonymized under E. milii var. breonii .

The designation breonii is wrought with misinterpretations, starting with Leandri (1946), Leandri (1952), Ursch & Leandri (1954) and culminating with Rebmann (2016). The name E. breonii Noisette is a self-admittedly superfluous replacement name for E. milii Des Moulins ( Newton 1994) . Noisette’s description and plate indeed describe E. milii as understood by us, and as explained by Noisette. Leandri (1946) presented the name E. splendens var. breonii Leandri , and again in 1952. In the latter publication he presented a full description, which Newton (1994) correctly identified as a plant very different from Noisette’s intention (= E. milii ). In Ursch & Leandri (1954) the designation was changed to E. milii var. breonii and a drawing was provided, proving further that a totally different plant than Noisette’s was indicated. From the description and plate of respectively Leandri (1952) and Ursch & Leandri (1954: pl. 38) we conclude that Leandri’s taxon is identical to E. hislopii . Leandri (1952) mentioned Perrier 18577 as the type, cultivated in the Parc Botanique et Zoologique de Tsimbazaza (#16) reportedly from Fianarantsoa (Betsileo). This specimen appears to be in P (MNHN-P-P00221939!), the label of which mentions the plant having been found around Fianarantsoa and cultivated in the Parc Botanique et Zoologique de Tsimbazaza, but the mention of a number 16 is missing from it. The specimen has large foliaceous bracts on the branches of the inflorescence, but this may be an artefact of cultivation. Rauh (1998: 230) mentioned plants of “ E. milii var. breonii ” found on granite rocks “near Antsirabe, not far from Fianarantsoa ”, in fact Euphorbia hislopii .

The third confusion was generated by Rebmann (2016) who took up the designation breonii to describe a new species from the Fianarantsoa region, south of Andoharanomaintso on an inselberg. Rebmann claimed to have found Ursch & Leandri’s (1954) plant and substantiated it by referring to Leandri’s description of it in 1952. He took up the epithet breonii again and published the new subspecies E. milii ssp. breonii Rebmann and chose a neotype for it. By failing to choose a holotype but rather a neotype for his de facto new name, Rebmann automatically invalidated his name. The other surprise is that his plant does not at all resemble Leandri’s (1946: 1952) and Ursch & Leandri’s (1954) plant. The latter has leaves many times larger than Rebmann’s, and Rebmann’s plant has very peculiar swollen spine-bases, absent from the Leandri plant which we consider to belong to E. hislopii . A comparison of Rebmann’s pictures and description with the drawing of the Ursch & Leandri plant leaves little doubt about Rebmann’s inaccurate interpretation of Leandri’s and Ursch & Leandri’s plant. Considering all the above cited confusion associated with the epithet breonii , we decided to once and for all dismiss all names containing the designation “ breonii ”. Rebmann’s concept is possibly conspecific with a taxon we describe in this paper ( Euphorbia pachyspina Haev. & Hett. , see below). We cannot assess in detail Rebmann’s taxon, as the type deposited at BR is not available for examination.

For a plant, in our opinion identical with E. hislopii, Leandri (1946) proposed the designation E. splendens var. vulcani (nom. inval., no Latin diagnosis, in key) and described it invalidly in 1952 (nom. inval., without a Latin diagnosis), typified by Decary 13797 (MNHN-P-P00078123!) from Betafo. Ursch & Leandri (1954) changed the name to E. milii var. vulcanii but left that designation also invalidly published, as no Latin diagnosis was provided. Castillon & Castillon (2020: 17) published validly the name as a variety of Euphorbia splendens [and as “ vulcani ”, to be corrected in “ vulcanii ”]. We consider this taxon to be E. hislopii and defer both the invalidly published designations, and the variety by Castillon & Castillon (2020) to the synonymy of it. All these taxa now gathered under E. hislopii consist of similar plants (Fig. 13), attested by cultivated specimens from Antananarivo and surrounding localities of the Imerina region, or found on rocks in the area (maybe of cultivated origin too and escaped). The wild distribution of the taxon is documented from an area extending further South, South-West, to the surroundings of Antsirabe and Ambatofinandrahana (see distribution in Fig. 8 ).

Euphorbia imperatae (Leandri ex Castillon & Castillon) Haev. & Hett. , comb. & stat. nov. (Figs. 6, 15 & 16). Basionym :— Euphorbia milii Des Moulins var. imperatae Leandri ex Castillon & Castillon (2020: 10) . Type :— MADAGASCAR. Toliary : Anosy , Taolagnaro, Mahatalaky, “Manantenina”, [24°43′60″S 47°7′60″E], 10 June 1925, Decary 3863 (holotype: MNHN-P-P00221940!) [reproduced here in Fig. 15].

— Euphorbia splendens var. imperatae Leandri (1946: 159) , nom. invalid. (no Latin diagnosis or description, in a provisional key).— Euphorbia splendens var. imperatae Leandri (1952 : name 109, description 114), nom. inval. (no Latin diagnosis). — Euphorbia milii var. imperatae (Leandri) Ursch & Leandri (1954: 150) , comb. invalid. (reference to an invalid “basionym”).— Euphorbia milii var. imperatae Ursch & Leandri (1954: 150) , nom. inval. (no Latin diagnosis or description).— Euphorbia milii var. imperatae Leandri ex Rauh, (1967a: 29) , nom. inval. (publ. inval., no Latin diagnosis or description).

— Euphoria milii forma lutea Rauh (1967a: 29) [valid combination of forma lutea with the name milii , see note below]. Type:— MADAGASCAR. Toliary: Anosy, Taolagnaro, “ca. 50 km nördlich Fort-Dauphin”, 27 November 1960, Rauh M1357 (holotype: HEID!)

Diagnosis:— Similar to E. milii Des Moulins , but differs by its decumbent, plagiotropic branches, its small rotund succulent leaves and shiny rounded, labiate convex cyathophylls.

Description:—Shrubby, spiny, herbaceous plant, much branched, low growing, 30–50 cm high, branches strongly plagiotropic with age, arching, bark at first reddish then grey, 0.5 cm in diam., smooth, with numerous brachyblasts and stipular spines. Leaves gathered at the tip of the branches and brachyblasts, deciduous, very short petiolate. Petiole 1 mm long. Lamina: terminal leaves oval or elliptic, 7–10 × 10–15 mm, leathery, top acute, apiculate, adaxial surface dark green, abaxial surface paler, brachyblast leaves similar but smaller, more orbicular, to 10 × 10 mm, top often slightly scalloped, apiculate. Stipular spines, simple, 1–1.5 cm long, thin, grey with a reddish-brown top. Inflorescences subterminal, dichasial, 2–4 cm long, once or twice dichotomous, carrying 2 to 4 cyathia, branches sticky, green or reddish brown. Cyathophylls erect, ca. 10 × 7 mm, upper half spreading, transversely oval, deep red or yellow shiny. Cyathium ca. 1 mm long. Cyathial glands 5, nearly touching laterally, carmine red. Style trichotomous near the base, then each branch dichotomous near their top. See Rauh (1967a: 27–29) for a description (in German) and see Rauh (1995: 126) for a summary description in English.

Additional specimens examined (paratypes):— MADAGASCAR. Toliary: Anosy, Taolagnaro , Mahatalaky , “Sur rocher à 43 km Nord de Fort-Dauphin, vers Manantenina; fleuri à Tananarive”, 1200 m elev., [24°43′60″S 47°7′60″E], 2 December 1974, Cremers 2912 (MNHN-P-P00078046!) . “ Mahatalaky , sur rochers”, 29 m elev., 24°47′20″S 47°4′39″E, 11 May 2001, Haevermans et al. 64 (P!, TAN!) . Fianarantsoa: Atsimo-Atsinana, Vangaindrano, Sandravinany , “Tolagnaro, Manantenina. Along the road going to Vangaindrano, between the Manantenina and Befasy villages”, 94 m elev., 24°5′37″S 47°22′14″E, 24 October 2014, Randrianasolo 1690 (MICH!, MO!, P!, TAN)

Note:— There is little doubt this taxon is of specific rank as it is well characterized by its plagiotropic branches and scrambling habit on rocks to the north of Fort-Dauphin around the locality of Mahatalaky where it seems to be restricted. The epithet ‘ imperatae’ though never validly published (never accompanied by a Latin diagnosis, see designations list above), has been recently validated by Castillon & Castillon (2020: 10) as a variety, which we elevate to specific rank. Rauh described a yellow form of this species as E. milii var. imperatae f. lutea ( Rauh 1967a: 29) but he used the invalid designation imperatae in the name. As a result, he actually validly published Euphorbia milii f. lutea Rauh. The fact that he attributed it to the invalid “ var. imperatae ” is without nomenclatural effect, “ f. lutea ” being correctly attributed to a specific name and accompanied by a Latin description. He however invalidly published Euphorbia milii var. imperatae Leandri ex Rauh by failing to describe the variety imperatae (see Article 36.3, example 13 of the ICN, Turland et al. 2018). Despite the valid publication of E. milii f. lutea Rauh and the inclusion of its type in our synonymy of the species we chose not to publish a replacement name based on this taxon.

Euphorbia neobosseri Rauh (1992a: 264) (Figs. 9A, 17, here). Replaced synonym:— Euphorbia milii var. bosseri Rauh (1970a: 271) . Type:— MADAGASCAR. Toliary : “forest of Sakaraha, east of Toliara (Tuléar)”, s.dat., Rauh 10856 (holotype: BGH10856 HEID!, in spiritu) .

— Euphorbia sakarahaensis Rauh (1992a: 266) , syn. nov. Type:— MADAGASCAR. Toliary: “ Silva de Sakaraha (Zombitsy), Prov. Toliara, SW Madagascar”, November 1991, Hofstätter s.n. (holotype: BGH72317 HEID!, in spiritu).

Diagnostic description:— Euphorbia neobosseri is very similar to E. berorohae Rauh & Hofstätter and basically only differs from the later in the spreading part of the cyathophylls being narrower and bases not overlapping with the opposite cyathophyll.

Additional specimens examined:— MADAGASCAR. Toliary : Atsimo-Andrefana , Sakaraha, Mahaboboka, “Sakaraha, Commune Mahaboboka, Village d’Ambinanintelo, Forêt d’Analaraty”, 520 m elev., 22°49′15″S 44°17′17″E, 16 November 2010, Randrianarivony 229 (MO!, TAN) .

Note:— The protologue and existing photographs of E. sakarahaensis shows it to be identical to E. neobosseri . The protologue of E. neobosseri is based on one single plant, growing in the Heidelberg Botanical Garden, with very narrow cyathophylls and linear leaves. New observations on plants from the type area show the variation of this species to be larger, extending to broader cyathophylls and slightly broader to lanceolate-spathulate leaves, representing intermediate morphologies with E. sakarahaensis . The latter species also originates from exactly the same area as E. neobosseri . We therefore decided to merge the two, of which the name E. neobosseri has priority as it is a replacement name for a taxon published in 1970. It must be noted, that in several publications after the protologue of E. neobosseri, Rauh pictured plants under that name from the Heidelberg Botanical Garden, that are in fact E. mahafalensis var. itampolensis (Rauh) Castillon & Castillon (2019: 160) . The confusion may have occurred because the living type plant of E. mahafalensis var. itampolensis in Heidelberg has a similar habit as E. neobosseri and has a height of 15 cm only, but that seems to be an artefact of cultivation as E. mahafalensis var. itampolensis is usually much more erect but tends to develop more straggling plants from feeble cuttings (W.L.A. Hetterscheid, personal observation). See Castillon & Castillon (2019: 160) for further discussion.

Euphorbia roseana (Marn.-Lap. ex Demoly) Castillon & Castillon (2020: 11) (Figs. 11A, 18, here). Basionym:— Euphorbia milii var. roseana Marn. -Lap. ex Demoly (2002: 189). Neotype (designated in Demoly 2002: 189):— MADAGASCAR. Toliary: Atsimo-Andrefana, Sakaraha, “Cultivé au Jardin Botanique des Cèdres [Saint-Jean-Cap- Ferrat, France], from a plant imported in 1993 from the forêt de Zombitsy”, [22°52′0″S 44°40′0″E], December 1997, Teissier s.n. (MNHN-P-P00577111!).

Additional specimens examined:— MADAGASCAR. Toliary : Atsimo-Andrefana, Sakaraha, “Zombitse National Park, along Sentier des Orchidées”, 810 m elev., 22°53′13″S 44°41′29″E, 30 November 2012, Gillespie 10823 (MO!, MICH!, TAN) . “ Zombitse forest ”, 834 m elev., 22°53′9″S 44°41′23″E, 9 November 2001, Haevermans et al. 191 (P!, TAN!) . “ Zombitse , dry part”, 750 m elev., 22°53′20″S 44°41′28″E, 18 March 2006, Haevermans et al. 341 (P!, TAN!) , Haevermans et al. 342 (P!, TAN!) , Haevermans et al. 343 (P!, TAN!) . Toliary : Atsimo-Andrefana , Sakaraha, Sakaraha, “Forêt de Zombitsy (Sakaraha), aux confins des bassins du Fiherenana et de l’Onilahy”, 725 m elev., 22°45′0″S 44°41′30″E, 26 March 1955, Humbert 29589 (MNHN-P-P00248892!) .

Note:—The name E. milii var. roseana was used by Marnier-Lapostolle (1962), based on a specimen collected by Montagnac from the Zombitse forest (now Zombitse-Vohibasia National Park) and grown in the Jardin des Plantes Botanical Garden in Paris (MNHN) and in the Botanical Garden Les Cèdres. It is unclear whether a dried specimen was prepared, but as Demoly (2002) correctly observed, there is no existing specimen by Montagnac of this variety. Also, the original living plant in Les Cèdres had since died. This prompted Demoly to neotypify the variety with a collection, Teissier s.n., prepared from a living plant in Les Cèdres collected by Marc Teissier near the original locality of Montagnac’s plant. In a personal communication, Mr. Teissier informed the first author that this recollected material conformed without doubt to the original Montagnac plant, an observation that was reported to him by the former curator of the garden who had studied it. We consider Marnier’s variety to be different in several respects from morphologically close relatives, notably from E. fanjahiraensis Haev. & Hett. (see below), and to be of species rank. Euphorbia roseana is a relatively small, bushy, decumbent, mound-forming, much branched plant from shaded areas in the Zombitsy forest. The cyathophylls are small, with the margins often recurved and not overlapping, ranging from pale yellow to red through various shades of pink (see Figs. 18B-E). Euphorbia fanjahiraensis has similar cyathia with similar color variation (see Rebmann 2012, under “ E. isaloensis ”), but it is much larger and erect and grows throughout the Isalo range in exposed places on gneissic rocks.

Euphorbia tananarivae (Leandri ex J.-P.Castillon & J.-B.Castillon) Haev. & Hett. comb. & stat. nov. ( Figs. 8 & 19). Basionym:— Euphorbia splendens var. tananarivae Leandri ex Castillon & Castillon (2020: 16) .— Euphorbia splendens var. tananarivae Leandri (1946: 159 , in clav.) nom. invalid. (no Latin diagnosis, no type designated). — Euphorbia splendens var. tananarivae Leandri (1952: 110 , 114), nom. inval. (no Latin diagnosis).— Euphorbia milii var. tananarivae (Leandri) Ursch & Leandri (1954: 150) , comb. invalid. (reference to an invalid “basionym”).— Euphorbia milii var. tananarivae Ursch & Leandri (1954: 150) , nom. inval. (no Latin diagnosis or description).— Euphorbia milii var. tananarivae Leandri (1954: 150) , nom. inval. [no Latin diagnosis].Type:— MADAGASCAR. Antananarivo: “Centre: Communément cultivée aux environs de Tananarive”, s.dat., Perrier de la Bâthie 18576 (holotype: MNHN- P-P00221935!).

Diagnosis:—Belonging to E. subgenus Euphorbia section Goniostema , the species is similar to E. hislopii by its shrubby, infrequently branching habit, thick stems and cyathophylls of ca. 1 cm in diam., but distinguished by the following characters: inflorescences with max. 16 cyathia (up to 64 in E. hislopii ), cyathophylls yellow (red in E. hislopii ) stipular spines without enlarged base or only slightly so (enlarged and laterally flattened in E. hislopii ), leaves with hairs at the base on the abaxial surface (leaves with glabrous abaxial leaf surface in E. hislopii ). The species also resembles E. splendens Bojer ex Hook. by its similar leaves, the hairy abaxial leaf base surface (although this is sometimes absent in E. splendens ) and the stipular spines without enlarged base but differs in the much thicker stems of up to 3 cm in diam. (only 1–1.5 cm in diam. in E. splendens ).

Etymology:—The species epithet is derived from Tananarive, the historical name for the capital of Madagascar (Antananarivo) which was used by Leandri for his designations, and also because the species is planted and grows in the area.

Description:—Shrubby, herbaceous plant, branching at the base, hardly rebranching, ca. 1 -2- m in diam., ca. 70 cm high, branches semi-erect to spreading, bark grey, with or without a few brachyblasts in the upper parts, leaves only at the branch tips or on the few brachyblasts. Leaves deciduous, very short petiolate. Petiole 2 mm long. Lamina elliptic-obovate with the basal part gradually tapering to the petiole, not obtuse, those on the branch tips 3–6(–10) × 2– 3(–4) cm, those on the brachyblasts 1–2 × 0.8–1.5 cm, adaxial surface glabrous, mid green, abaxial surface pale green, with tiny, white hairs near the base and on the main vein, top rounded, shortly apiculate. Stipular spines, 1–1.5 cm long, base not or slightly enlarged, sometimes carrying a secondary spine on the basal half. Inflorescence subterminal, one or two per branch, dichasial, 4–6 cm long, 2–4 times dichotomous, carrying 4 to 16 cyathia, branches sticky, reddish brown (paler in cultivation) inflorescence bracts scaly, very small, caducous. Cyathophylls ca. 10 × 10 mm wide, upper part spread, transversally oval, apex rounded, apiculate, glabrous, pale to darker yellow, margin sometimes flushed reddish. Cyathium diameter 3 mm, sex bisexual, cyathium and cyathophyll erect, cyathium size ca. 3 × 4 mm. Cyathial glands greenish to dirty yellow. Interglandular bracts 1 × 1 mm. Staminate flowers pedicel length 2–3 mm, filament size 1–2 mm. Pistillate flower erect, 3 carpels, ovary 1 × 1.5 mm, glabrous, style 1–2 mm long, connate. Fruit and seed unknown.

Distribution and habitat:—Restricted to Madagascar. Antananarivo area, Imerina (see Fig. 8 ). Euphorbia tananarivae has been cultivated in the capital Antananarivo and its immediate surroundings for unknown times.

Additional specimens examined (paratypes): — MADAGASCAR. Antananarivo: “Environs de Tananarive (sur les rochers d’Ankatso)”, [18°55′0″S 47°31′60″E], 15August 1928, Decary 6717 (MNHN-P-P00221994!). “Environs de Tananarive (Ambohipo)”, [18°55′0″S 47°34′0″E], 26 August 1928, Decary 6763 (MNHN-P-P00221995!). “Environs de Tananarive (Ambohidratrimo)”, [18°49′60″S 47°25′60″E], s.dat., Decary 6793 (L-L.2216976!, MO!, MNHN- P-P00221997!). “Imerina”, [19°22′24″S 46°42′22″E], January 1881, Hildebrandt 3832 (MNHN-P-P00217932!). “Rocaille du Jardin Botanique”, 26 March 1944, Jardin Botanique de Tananarive 6259 (MNHN-P-P00217911!). “Rocaille du Jardin Botanique”, 26 March 1944, Jardin Botanique de Tananarive 6260 (MNHN-P-P00217912!). “Rocaille du Jardin Botanique”, 26 March 1944, Jardin Botanique de Tananarive 6261 (MNHN-P-P00217913!). “Province de Tananarive, District d’Andramasina, près d’un village entre Andramasina et Tsiafahy, vers 1450 m ”, 19°8′48″S 47°33′11″E, 4 October 1912, Viguier 1944 (MNHN-P-P05479931!). Ambohitro, 26 August 1928, Decary 6763 (MNHN-P-P00221995!). Living plant in the collection of the National Tree Museum, Netherlands, acc. nr. 2020VGA0936 (a cutting from a plant donated to Mr. J. Keizer, The Netherlands, by Werner Rauh; no records in the Heidelberg Botanical Garden of this plant, no longer alive there) .

Notes:—Though never validly published before Castillon & Castillon (2020), the name (‘var. tananarivae’) of a yellowish Euphorbia had been invalidly published as a variety of E. splendens by Leandri (1946: 159), then invalidly again as a variety of E. milii in Ursch & Leandri (1954: 150), has been known in cultivation in the West for a long time under the name ‘ E. milii var. tananarivae’. These plants are much smaller than the “real” E. tananarivae , much more branched and likely are hybrids and possibly identical to the plant from the Botanical Garden of Antananarivo mentioned by Ursch & Leandri under the garden number 1 (1954: 150, Pl. 39). An herbarium specimen in Paris (MNHN-P-P00078045!) was collected from this plant by Bosser in 1960, and is very similar indeed to the abovementioned cultivated plants, of which a living specimen was studied in the collection of the National Tree Museum, The Netherlands (acc. nr. 2020VGA0964). This constitutes in fact one of the first, if not the first yellow-flowering cultivar of, judging from the leaves, E. splendens .

Euphorbia tenuispina (Rauh & Razaf.) Castillon & Castillon (2020: 11) (Figs. 9A & 20). Basionym:— Euphorbia milii var. tenuispina Rauh & Razaf. ( Rauh 1991: 34, Figs. 20, 21). Type:— MADAGASCAR. Fianarantsoa: Ihorombe, Ihosy, “Mountains around Ihosy”, [22°24′13″S 46°3′30″E], s.dat., Razafindratsira s.n. (holotype: BGH68736 HEID!, in spiritu).

— Euphorbia milii var. tulearensis Ursch & Leandri (1954: 152) , syn. nov. Type:— MADAGASCAR. Toliary: Atsimo-Andrefana, Toliary, “Tuléar, La Table; échantillon fait par E. Ursch au Jardin Botanique de Tananarive”, 1200 m elev., [23°23′60″S 43°46′60″E], s.dat., Leandri 2385 (holotype: MNHN-P-P00078047!).

Diagnostic description:— E. tenuispina has the general habit and soft spines of E. roseana but differs in having more robust stems and growing a bulbous base (at least in younger plants), the branches usually being reddish brown (versus pale grey in E. roseana ), the spines often being complex (with 3–4 additional smaller spines on the base of the main spine) and the cyathophylls being much larger (to nearly 1 cm in diam. versus only 5 mm in E. roseana ).

Additional specimens examined:— MADAGASCAR. Fianarantsoa: “ Massif de l’Isalo ”, 20 May 1940, Decary 15939 (MNHN-P-P04780921!). “ Isalo , sur roc”, 1041 m elev., 22°30′31″S 45°22′4″E, 24 May 2001, Haevermans et al. 93 (P!, TAN!). “ W of Isalo NP, 10 km W of Ranohira. Sandstone outcrop”, 700 m elev., 22°34′60″S 45°19′60″E, 13 August 1993, Lewis 543 (L-WAG.1794345!, MO!, MNHN-P-P00221823!)

Note:—For a full description and further details see Rauh (1991). We add to the description that the base of plants grown from seed develop a caudiciform/tuberous base and swollen root bases (Fig. 20B). While the collector of the type specimens mentions that the plants grows in Ihosy and Isalo, none of the specimens we consulted, besides the type and an observation from Mahasoa, originates outside of the Isalo area (see distribution map in Fig. 9A). The application of the name E. milii var. tulearensis , described from a cultivated plant reportedly coming from Tuléar (Fig. 9B) has always been problematic. Castillon & Castillon (2019: 162) put this taxon in synonymy with E. beuginii Rebmann (2012: 45) , but we do not agree with this decision, not taking into account a key character described in the protologue: E. milii var. tulearensis is described with very soft spines, oblong leaves and rather large, red, spreading cyathophylls. This character combination is found only in E. tenuispina , but not in E. beuginii . Euphorbia tenuispina occurs in Isalo and Ihosy areas, far from the reported type locality of E. milii var. tulearensis . Previous expeditions to Tuléar never yielded any red-flowering specimens, so there remains the possibility that the locality information may be incorrect. It is, of course, not too safe to have trust in a third person providing such information, as occurred for the type of this taxon, so we consider this information as doubtful. We therefore reject the synonymy by Castillon & Castillon (2019: 162), and propose E. milii var. tulearensis as a new synonym for E. tenuispina .

Revised taxonomy of selected Malagasy taxa of Euphorbia L. subg. Goniostema Baill. ex Boiss.

Euphorbia begardii (Cremers) Haev. & Hett. , comb. & stat nov. (Figs. 11B & 22BC). Basionym:— Euphorbia primulifolia Baker var. begardii Cremers (1984b: 386 , Figs. 7–8). Type:— MADAGASCAR. Fianarantsoa: Ihorombe , “Isalo Sud”, 22°31′0″S 45°16′60″E, 16 November 1973, Cremers 2843 (holotype: MNHN-P-P00078085!, isotype: TAN!).

Additional specimens examined:— MADAGASCAR. Toliary: Atsimo-Andrefana, “ Plateau de l’Horombe entre Ihosy et Betroka, à 55 km au sud d’Ihosy, sur le bord de la RN 13”, 1005 m elev., 22°53′19″S 44°8′47″E, 21 November 2009, Aubriot et al. 1 (MO!, MICH!, MNHN-P-P00685080!, TAN!). “ Isalo (près de Ranohira ); (fleur en sept.-nov. en jardin botanique)”, [22°34′0″S 45°25′0″E], November 1962, Bosser 16385 (MNHN-P-P00224929!, MNHN-P-P00224928!). “ Plateau de l’Horombe , Piste de Satrokala , PK 8”, [22°26′60″S 45°36′0″E], 12 February 1970, Bosser 19890 (MNHN- P-P00224933!). “ Isalo ”, [22°31′0″S 45°16′60″E], s.dat., Bosser 73 (MNHN-P-P00224926!). “ Massif de l’Isalo ”, [22°31′0″S 45°16′60″E], 31 October 1940, Decary 16291 (MNHN-P-P00224927!). “ Isalo ”, 833 m elev., 22°39′3″S 45°18′52″E, 9 November 2001, Haevermans et al. 196 (P!, TAN!). s.loc., s.dat., Schlieben 8230 (BR0000017303187!). Toliary: Androy, Bekily, Beraketa, “Environs d’Ampandrandava (entre Bekily et Tsivory), Beraketa”, 550 m elev., [24°10′60″S 45°42′0″E], January 1944, Seyrig 803 (MNHN-P-P00224937!) .

Note:— This geophytic taxon was originally published as a variety of Euphorbia primulifolia Baker (1881: 278) , but their ecologies and morphologies are quite distinct (high altitude laterite soils in the vicinity of Antananarivo in E. primulifolia vs. lower altitude arid sands around Isalo and Horombe plateau in E. begardii ). Euphorbia begardii is further distinguished from E. primulifolia (Fig. 22A) by its non-triangular cyathophylls without lacerated margins, its lamina with purple-greenish underside and with sub-crispate to undulate reddish margins. Preliminary DNA sequences comparison studies further show that they are not closely related ( Evans et al. 2014), hence the elevation of this variety to the rank of species.

Euphorbia berevoensis Lawant & Buddensiek (2008: 6) (Fig. 11A). Type:— MADAGASCAR. Toliary : Atsimo- Andrefana, “between Berevo and the river Tsiribihina”, 19°44′23′′S 44°54′56′′E, 1993, Röösli 43/93 (MNHN-P- P00577728!).

— Euphorbia nicaisei Rebmann (2013: 11 , figures labelled « Euphorbia nicaisei »), syn. nov. Type:— MADAGASCAR. Toliary: Berevo, “sur falaise gréseuse et sable de décomposition (arène)”, 65 m elev., 22 October 2012, Rebmann 23 (holotype: BR [not available]).

Note:— Euphorbia nicaisei Rebmann (2013: 11) was described from a specimen collected in the same locality (the Berevo area) as the previously described E. berevoensis Lawant & Buddensiek (2008: 6) . The paper by Rebmann compares the new species to others related species, namely Euphorbia erythrocucullata Mangelsdorff (2005: 3) , E. kondoi Rauh & Razafindratsira (1989: 113) , and E. pedilanthoides Denis (1921: 76) but he did not compare it to E. berevoensis , which is remarkably similar and grows in the same locality as his new species. We emphasize that Rebmann’s diagnostic characters are not sufficient to separate his species from E. berevoensis , both having the same morphological characters (as shown by the illustrations in Rebmann’s paper), and both occur in the same general locality. Consequently, we consider E. nicaisei Rebmann to be a heterotypic synonym of E. berevoensis Lawant & Buddensiek.

Euphorbia crassicaulis (Rauh) Haev. & Hett. , comb. & stat. nov. (Figs. 23 & 29). Basionym:— Euphorbia francoisii Leandri var. crassicaulis Rauh (1996: 183) .— Euphorbia decaryi var. crassicaulis (Rauh) Castillon & Castillon (2016: 153) . Type:— MADAGASCAR. Toliary: Anosy, Taolagnaro, “Toliary: Ankilimasy near Fort-Dauphin”, [25°4′60″S 46°36′0″E], March 1993, Rauh 73936 (holotype: HEID!, in spiritu).

Diagnostic description:— Euphorbia crassicaulis differs from E. decaryi Guillaumin (in Gagnepain et al. 1934: 120) in having thicker branches (to 2 cm thick vs. 1 cm in decaryi ), stipules to 1 cm (5 mm in decaryi ), the leaves on one plant uniformly oblong (very variable on one plant and between plants in E. decaryi ), leaf margin regularly crispedundulate, (irregularly undulate to flat in E. decaryi ), the leaf base quite abruptly tapering to the petiole (gradually tapering in E. decaryi ).

Note: — The corrected type citation cited above is following Castillon & Castillon (2016: 153) —the original protologue type reads “under dense bushes near the coast, close to the village of Andrahomana (in southern Ranopiso, between Tolanaro and Amboasary), March 1993 ”, which did not match Rauh’s collecting notes. A co-existing red form mutation of this taxon is mentioned by Rauh (1998) under the nomen nudum Euphorbia francoisii var. crassicaulis forma rubrifolia , which is a red variant (Fig. 23) belonging to the “normal” green E. crassicaulis (Rauh) Haev. & Hett. In cultivation numerous hybrids exist between E. decaryi and E. crassicaulis , with character combinations intermediate between the two parental species.

Euphorbia delphinensis Ursch & Leandri (1954: 166) (Figs. 24 & 40A). Type:— MADAGASCAR. Toliary: Anosy, Taolagnaro, s.dat., Ursch s.n. (MNHN-P-P00077966!, herbarium specimen prepared by E. Ursch from a living plant in the Parc Botanique and Zoologique de Tsimbazaza , originally collected by Mr. Duran in Fort-Dauphin and with the number 3 of the garden and deposited in the P herbarium by J. Leandri as Leandri 2407).

— Euphorbia beharensis var. squarrosa Rauh (1999b: 14 , left part of Fig. 8 p . 29). nom. inval., syn. nov. Type :— MADAGASCAR. Toliary : Anosy, “In xerophytic forest at Lac Anony, Tolanara prov.”, s.dat., Rauh 74789 (holotype: HEID!, in spiritu).

Description:—Shrubby, spiny, herbaceous plant, plagiotropic or partly erect, to ca. 1 m high, sparingly branched, branches at right angles relative to the main stems, carrying irregularly arranged brachyblasts, main stems to 1 cm in diam., branches to 0.5 cm in diam., bark dark greyish green. Leaves deciduous but long-lived, semi-succulent, along the greater parts of the stems and branches and clustered on brachyblasts, similar in shape on stems, branches and brachyblasts, or those on the brachyblasts more elliptical or oval, short petiolate, narrowly obovate, 10–20 × 4–7 mm, top shortly acuminate, adaxial surface dull mid-green, abaxial surface paler. Petiole 1 mm long. Stipular spines simple, 5 mm long, reddish when young, turning grey, base slightly thickened. Inflorescence subterminal, several per branch, dichasial, 3–4 cm long, 1–2 times dichotomous, carrying 2–4 cyathia, branches sticky, green, inflorescence bracts minute, scaly, fused with their bases, oval, truncate, reddish green. Cyathophylls broadly oval, 3 × 3 mm, more or less spreading, upper margin rounded, apiculate, adaxial surface pale greenish yellowish with a more or less clear carmine-red flush, especially near the margin. Cyathium cylindric or cup-shaped, ca. 2 × 2 mm. Glands very pale, dirty yellowish-brownish with or without a pale carmine-red outer margin. Interglandular bracts green, top dentate. Style trichotomous, branches entire, dirty pale-brownish.

Note:— Rauh (1999b) published a number of new varieties of E. beharensis (see E. guillemetii notes) of which only E. beharensis var. squarrosa differs considerably from the others. We propose that this variety does not belong to E. guillemetii at all but instead is the same as E. delphinensis . We compared the type and protologue of E. delphinensis with Rauh’s plant (including the living plant from which Rauh prepared the type of his variety and of which a cutting is grown by W.L.A. Hetterscheid in the National Tree Museum Gimborn) and found no difference among them. Rauh (1967b) introduced a plant under the name E. delphinensis but did not reveal its origin. This plant is also known in cultivation today under this name. It is much larger in all its parts than the actual E. delphinensis . We found a specimen in P of this plant, collected in the wild, and we propose a new species for this ( E. werneri Haev. & Hett. , see below). In his protologue, Rauh (1999b: 14) presented a description and pictures of E. beharensis var. squarrosa , but they seem to be a mixture of the two different taxa. We consulted the type of var. squarrosa in Heidelberg and concluded that the right-hand picture of Rauh (1999b: 29, Fig. 8 ) is not that variety but another undescribed taxon. Of this second taxon, a cutting from the living plant in the Heidelberg Botanical Garden, was grown by the second author and it showed an erect habit, substantially longer leaves and much larger and yellow cyathophylls than var. squarrosa . Thus, the exclusion of Rauh’s (1999b: Fig. 8 , right hand picture) from the protologue of E. beharensis var. squarrosa Rauh. We also noted that Rauh (1999b: 15) wrongly cited the type number as 74989 (a pseudo-Rauh number in the WRHP database, https://wrhp.cos.uni-heidelberg.de/) but it should be 74789 according to his notebook and the label of the pickled holotype (HEID). Pseudo-Rauh numbers are numbers given by W. Rauh to plants he received from other people and not collected by himself.

Euphorbia fanjahiraensis Haev. & Hett. , sp. nov. (Figs. 11A, 25 & 26). Type:— MADAGASCAR. Fianarantsoa: Ihorombe, Ihosy, “Plateaux et vallée de l’Isalo: environs de Fanjahira”, 450 m elev., [22°45′0″S 45°15′0″E], November 1924, Humbert 2761 (holotype: MNHN-P-P00078015!, isotypes: MNHN-P-P00078018!, MNHN-P-P00708248!) [holotype reproduced here in Fig. 25].

— E. isalensis Leandri (1946: 163) , nom. inval. [as “ nov. spec. interim ”], syn. nov. Holotype: MNHN:— MADAGASCAR. “ Euphorbia laissée par H. Humbert au Jardin de Tananarive provenant peut-être de l’Isalo”, [22°31′0″S 45°16′60″E], s.dat., Perrier de la Bâthie s.n. (MNHN-P-P00221824!).

Diagnosis:— In general habit and cyathia shape, E. fanjahiraensis resembles E. roseana but is twice as high and not decumbent, has much more robust and longer spines and longer and narrower leaves.

Description:— Bushy, spiny, deciduous plant, erect or decumbent, main stem branching frequently or infrequently, grey, branches spreading or slightly oblique, carrying few or many brachyblasts. Leaves concentrated at the tips of the branches and on the brachyblasts. Leaves on the branch tips lanceolate, oblong-lanceolate or oblong, 3–6 × 1–1.5 cm, short petiolate (petiole 1–2 mm long), apex shortly acuminate, adaxial surface green, with or without a narrow reddish margin, abaxial surface pale green. Leaves on the brachyblasts smaller and more elliptic, to 1.5 × 1 cm. Stipular spines simple, rigid or rarely soft, 1–1.5 cm long, grey. Inflorescence to 9 cm long, 2–4 times dichotomous, carrying 4–32 cyathia, common peduncle to 7 cm long, reddish, sticky. Cyathophylls 8 × 3–4 mm, spreading part narrowly triangular-cordate, apiculate, margins often strongly reflexed, base of opposite cyathophylls not overlapping, or rarely slightly so, adaxial surface pale yellow, or orangeish-yellow, abaxial surface similar but paler. Cyathium campanulate, 4 × 4 mm, glands touching laterally, erect, bilabiate, halfmoon-shaped, dirty yellowish, yellowish-brown, or orange. Interglandular bracts slightly longer than glands, wedge-shaped, with rounded, slightly bilobed top, upper margin dentate. Staminate flowers filaments pink. Pistillate flowers ovary glabrous, style branches fused nearly to 1/3 rd from the top, pale yellow.

Additional specimens examined:— MADAGASCAR. Fianarantsoa: Ihorombe , “Plateaux et vallée de l’Isalo”, 700 m elev., [22°31′0″S 45°16′60″E], November 1924, Humbert 2786 (MNHN-P-P00078016!, MNHN-P-P00078017!, MNHN-P-P00708249!). Toliary: Atsimo-Andrefana, “Sud Ouest: Mt Votaka, près Benenitra (Onilahy)”, [23°26′60″S 45°4′60″E], July 1910, Perrier de la Bâthie 9778 (MNHN-P-P00217878!). Toliary: Atsimo-Andrefana, “Ouest: Environs de Benenitra (Onilahy)”, 600 m elev., [23°26′60″S 45°4′60″E], July 1919, Perrier de la Bâthie 12706 (MNHN-P-P00217883!) .

Note:— We preferred to propose an entirely new name not based on Leandri’s voucher and thus not validate the name E. isalensis Leandri (1946: 163) , due to the existence of a homonym ( E. isaloensis Drake (1899: 307)) , a name we consider dubious as being possibly described based on mixed material, and of which no type material survived, see Costantin & Gallaud (1905). Euphorbia isalensis Leandri was initially published as a “ nov. spec. interim ” (see Haevermans et al. 2009), which was considered a valid publication of the name even though the name was rejected in Ursch & Leandri (1954: 166), see Haevermans et al. (2009). However, an example added to the code since the publication by Haevermans (2009) explicitly indicated that the name publication has to be considered invalid ( Turland et al. 2018: Art. 36.1—Ex. 6). Leandri (1953: 144) uses the name as a synonym of E. isaloensis Drake , considering Drake’s taxon to be conspecific and surprisingly include in the synonymy the unrelated E. mainiana Poisson (1912: 37) . The name was not validated in this publication which was not a taxonomic revision fide Leandri (1953). The final rejection of E. isalensis Leandri occurs in Ursch & Leandri (1954: 166) where both E. isalensis and E. isaloensis are cited in the synonymy of E. mainiana . Besides the surprising interpretation they make of E. mainiana (represented in their paper by a cultivated plant of unknown origin: Jardin Botanique de Tananarive 33 [MNHN-P-P00221882!]), they further reject E. isalensis Leandri as being a later homonym of E. isaloensis Drake , a doubtful species. As a result, considerable confusion exists about the application of E. isalensis Leandri , also involving another taxon from the same general area: E. roseana (= E. milii var. roseana , see Fig. 18 and earlier discussion of E. roseana ) from Zombitsy Forest (Sakaraha). Leandri’s 1946 concept is a small (“ Planta humilis ….”), branching, spiny plant grown at the Parc Botanique et Zoologique de Tsimbazaza in Antananarivo and according to Humbert (quoted by Leandri in the protologue) “probably originating from Isalo”. This quote is found written by Humbert on a specimen in Paris, Perrier s.n. (MNHN-P-P00221824!), which also mentions on the label that it was brought to the Botanical Garden of Antananarivo. Leandri also referred to two collections by Humbert as possibly belonging to E. isalensis, Humbert 2761 from the surroundings of Fanjahira (Toliary province) and Humbert 2768 from the same area. We consider both Humbert specimens mentioned by Leandri as indeed belonging to this concept. For representative photographs of E. fanjahiraensis in the field, see Rebmann (2012: all figs with the name “ Euphorbia isaloensis ”). See Fig. 26 here for photographs of a live plant in cultivation showing the characteristic habit and inflorescence.

Euphorbia guillemetii Ursch & Leandri (1954: 168) (Figs. 11B & 21).— Euphorbia beharensis Leandri var. guillemetii (Ursch & Leandri) Rauh (1999b: 10) , nom. inval. Type: — MADAGASCAR. Antananarivo: Cultivated, “Jardin de Tsimbazaza sous le nº 36, plante récoltée à Ranomainty”, s.dat., Leandri 2398 (holotype: MNHN-P-P00077992!).

— Euphorbia beharensis Leandri (1946: 164) , nom. inval., syn. nov. Voucher (improperly designated as a lectotype by Haevermans et al. 2009: 281):— MADAGASCAR. Toliary: Anosy, Taolagnaro, Ambatoabo, “Vallée de la moyenne Mananara (limite orientale de l’Androy)”, [24°49′0″S 46°37′0″E], 26 November 1931, Decary 9435 (MNHN-P-P00077906!).

— Euphorbia beharensis Leandri var. adpressifolia Rauh (1999b: 13) . nom. inval., syn. nov. Voucher :— MADAGASCAR. Toliary: “circa Beharam, in harena, in territorio var. beharensis, 1996 ”, Rauh 74672 (holotype: HEID!, in spiritu).

— Euphorbia beharensis Leandri var. truncata Rauh (1999b: 11) . nom. inval., syn. nov. Voucher:— MADAGASCAR. Toliary: “silva Didiereacearum apud Behara in harena (Madagascar meridionalis-occidentalis)”, March 1987, Rauh 68567 (holotype: HEID!, in spiritu).— E. beharensis ‘Truncata’ (“cv. Truncata”) in Rauh (1998: 141, figs 495, 496, 500, p. 141, 501–502, p. 142), not established according to ICNCP, art. 12.11 ( Brickell et al. 2016), syn. nov.

Description:—Shrub, spiny, multi-stemmed, tuberous-rooted, to ca. 1 m high, numerous stems emerging from the plant base, to 2 cm in diam. at the base, fully erect or the upper half bending sideways to various degrees, rebranching rarely or frequently, branches 20– 60 mm thick, developing numerous brachyblasts. Leaves clustered at top of stems and on brachyblasts. Leaves semi-succulent to succulent, short petiolate (petiole 1–2 mm long), at a right angle with the stem/branch to strongly reflexed, parallel with and pressed against the stem/branch, glabrous on both sides, adaxial surface green or greyish green or grey, abaxial surface paler, margin straight or strongly undulate, those on the stem/ branch-endings elliptic or obtriangular-truncate, 10–15 × 5–20 mm in diam., tapering to the acute top, or top abruptly acute, leaves on the brachyblasts similar but smaller. Stipular spines 3–20 mm long, variable on individual plants and among plants, simple, closely or distantly arranged, at first pale to dark brown, soon turning grey. Inflorescences 6–50 mm long, 1–3(–4) times dichotomous, carrying 2–8(–16) cyathia. Cyathophylls 4–6 × 3–4 mm, erect, oblique or spreading, spreading part half-moon shaped, apiculate, adaxial surface pale yellow, with or without reddish margin or entirely reddish yellow to reddish, abaxial surface paler. Cyathium cylindric, 2 × 2 mm, glands erect, transversely oval, bilabiate, yellow or reddish. Interglandular bracts slightly longer than glands, upper margin dentate. Pistillate flowers ovary glabrous, style trichotomous in the middle, style branches dichotomous at the top.

Additional specimens examined:— MADAGASCAR. Toliary: Androy, Tsiombe, Tsihombe, “Androy-Tsiombe. Dans les fourrés à 7 km au nord-est de Tsiombe, près de la RN 10”, 73 m elev., 25°18′19″S 45°33′4″E, 1 December 2009, Aubriot et al. 60 (MICH!, MO!, MNHN-P-P00685140!, MNHN-P-P00685141!, MNHN-P-P00696159!, TAN!). Anosy, Taolagnaro, “Belivany (proche Tuléar)”, [24°59′51″S 46°34′49″E], March 1964, Bosser 19355 (MNHN-P- P00275244!). “Madagascar”, Cours 4581 (MNHN-P-P00275248!). “Ambovombe, Ampasimpolaka”, [25°7′42″S 46°23′26″E], 14 June 1924, Decary 3038 (MNHN-P-P00275254!, MNHN-P-P00275255!). “Ambovombe”, [25°7′42″S 46°23′26″E], 3 September 1924, Decary 3041 (MNHN-P-P00275253!). “Behara (province de Fort-Dauphin): sable gneissique”, [24°57′6″S 46°23′12″E], 10 July 1926, Decary 4276 (BM000832487 BM!, MNHN-P-P00077907!, US!). “Ambovombe”, [25°8′43″S 46°4′48″E], 27 February 1931, Decary 8574 (MNHN-P-P00275252!). “Région du lac Anongy [Anony], sud est d’Ambovombe”, [25°7′42″S 46°23′26″E], 2 October 1931, Decary 9265 (MNHN- P-P00275249!). Cultivated, “Station de Betanimena”, Dequaire 27429 (MNHN-P-P00275247!). Behara, “Environs d’Antanimora (Androy), 20-25 km au SSE”, 350 m elev., [24°53′60″S 46°31′0″E], 7 February 1955, Humbert 28875 (MNHN-P-P00220691!). “Côtes et plateaux calcaires à l’estuaire de l’Onilahy (côte sud)”, 76 m elev., [23°31′32″S 43°47′25″E], 20 March 1955, Humbert 29538 (MNHN-P-P00275243!). “Bassin inférieur du Mandrare: environs de Behara (piste de Behara à Ankirikiry)”, 60 m elev., [24°55′60″S 46°22′60″E], September 1928, Humbert 5653 (MNHN-P-P00220686!, MNHN-P-P00220687!, MNHN-P-P00220688!). “Ranomaintsy; échantillon fait par E. Ursch au Jardin Botanique de Tananarive. Jardin Botanique No 36”, [25°0′0″S 46°34′0″E], s.dat., Leandri 2398 (MNHN-P- P00077992!). “Ranomainty, à l’est d’Ambosary, près d’un baobab au sud de la route”, [25°0′12″S 46°32′11″E], s.dat., Montagnac 18 (MNHN-P-P00275250!). “Tsiombe, Sakamasy forest, 8 km from Tsiombe”, 67 m elev., [25°17′52″S 45°32′52″E], Rakotonasolo 1622 (BR0000016000599!, TAN). “Environs de Behara”, [24°55′60″S 46°22′60″E], s.dat., Rauh M1302 (MNHN-P-P00275251!). “ 9 km au S d’Ambatomika, vallée du Mandrare”, 24°38′60″S 46°28′0″E, 5 November 1994, Teissier 181 (MNHN-P-P00078928!, TAN). “E Amboasary rte de Fort-Dauphin, vers bifurcation nouvelle route du lac Anony”, 25°1′60″S 46°22′60″E, 9 November 1994, Teissier 199 (MNHN-P-P00078933!, TAN).

Note:—The name E. beharensis Leandri (1946: 164) was published as a “ nov. spec. interim ”, invalidating the name ( Turland et al. 2018: Art. 36.1—Ex. 6), and was still not validated in Leandri (1953: 144) and in Ursch & Leandri (1954: 170) as the authors did not “explicitly recognize” the taxon as both works are not considered taxonomic revisions by the authors, as explained in their respective introductory texts. Rauh (1999b) did not explicitly validate the name either and listed E. beharensis Leandri with a reference to a diagnosis in Leandri (1953: 144), which was incorrect, as this work only contains a description in French and further referred to the taxon as “ E. beharensis Ursch & Leandri ”. As his new varieties and new combination are all based on an invalidly published names, they are therefore invalid. As a result, the designation E. beharensis Leandri is still invalid as it has never been explicitly validated. There is no need to validate this name as another validly published name is available for this taxon: E. guillemetii Ursch & Leandri. We consider the above mentioned three (out of four) invalidly published varieties of E. beharensis proposed by Rauh (1999b) not worthy of independent recognition at any taxonomic level. The differences among them and more ‘typical’ E. guillemetii seem too small and variable, as intermediate forms between the varieties are frequently seen in collections and on herbarium specimens. Exception must be made for a fourth name proposed by Rauh, var. squarrosa , which differs strongly in habit from any of the other varieties and the species itself. We propose that this invalid variety is synonymous with E. delphinensis Ursch & Leandri (see above).

Euphorbia leandriana Boiteau(1941:4) .(Figs.27,28&29). Neotype (designated here,see Fig.27):— MADAGASCAR. “Sampled from the same plant as the type, cyathophylls of two colors, sometimes yellow, sometimes red-orange”, Jardin de Tsimbazaza , s.dat., Boiteau s.n. (P!) [specimen chosen here: a dried specimen (originally in spiritu) taken by Boiteau from the original plant at the botanical garden of Antananarivo, reproduced in Fig. 21. A picture of this original “type” plant alive in Antananarivo was made by J. Bosser and reproduced here in Fig. 28A].

— Euphorbia horombensis Ursch & Leandri (1954: 154) , syn. nov. Lectotype (designated by Haevermans et al. 2009: 286):— MADAGASCAR. Fianarantsoa: Ihorombe, “Horombé; échantillon fait au Jardin Botanique de Tananarive par E. Ursch”, [22°30′0″S 45°49′60″E], s.dat., Leandri 2409 (MNHN-P-P00078001!).

Additional specimens examined:— MADAGASCAR. Toliary: Anosy, Betroka, “Anosy-Betroka. Plateau de I’Horombe entre Ihosy et Betroka, à 30 km au nord de Betroka, flanc Ouest d’une colline pierreuse, sur le bord de la RN 13”, 1271 m elev., 22°59′59″S 46°8′8″E, 21 November 2009, Aubriot et al. 5 (MICH!, MO!, MNHN-P-P00685084!, TAN!). Fianarantsoa: Ihorombe, Ihosy, Ranohira, “Isalo Sud”, [22°31′0″S 45°16′60″E], 16 November 1973, Cremers 2844 (MNHN-P-P02284085!). Toliary: Anosy, Betroka, Ambalasoa, “Cultivée et fleurie à Tananarive: ramenée d’un rocher au Sud d’Ihosy au PK87, en direction de Fort-Dauphin”, [23°3′0″S 46°6′0″E], 15 November 1974, Cremers 2945 (MNHN-P-P00078002!). “Cultivée et fleurie à Tananarive: sur un affleurement rocheux à 100 km Sud d’Ihosy vers Fort- Dauphin”, [23°0′0″S 46°7′60″E], 2 December 1974, Cremers 3679 (MNHN-P-P00078003!). Ihorombe, Ihosy, “Massif de l’Isalo”, [22°31′0″S 45°16′60″E], 1 November 1940, Decary 16352 (MNHN-P-P02284020!). “Vallée de l’Ihosy”, [22°23′60″S 46°7′60″E], 5 September 1898, Grandidier s.n. (MNHN-P-P00217895!). “Isalo, rocaille au soleil”, 935 m elev., 22°29′41″S 45°22′40″E, 24 May 2001, Haevermans et al. 90 (P!, TAN!), Haevermans et al. 91 (P!, TAN!). Haevermans et al. 92 (P!, TAN!), “Vallée d’ihosy”, 900 m elev., [22°23′60″S 46°7′60″E], November 1933, Humbert 11592 (MNHN-P-P00221932!). Toliary: “Mont Vohipolaka au N de Betroka (Centre Sud), vers le sommet”, 1550 m elev., [23°10′0″S 46°4′60″E], November 1933, Humbert 11726 (MNHN-P-P00217935!). Toliary: Anosy, Betroka, Ambalasoa, 990 m elev., 23°0′0″S 46°6′55″E, 18 March 1992, Phillipson 3930 (MO!, TAN). Fianarantsoa: Ihorombe, Ihosy, “Granite de l’Isalo”, [22°24′26″S 45°19′49″E], 12 September 1922, Poisson 579 (MNHN-P-P00217897!).

Note: —The interpretation of the name E. leandriana Boiteau has long remained unclear because Boiteau did not designate a single type specimen in his protologue for the species, but referred instead to a group of plants grown in the Parc Botanique et Zoologique de Tsimbazaza, collected by Humbert (numbers 22 and 228 in the French description, 222 and 228 in the Latin diagnosis) and without any reference to an illustration or a preserved part. Specimens must have been prepared subsequently though, as he mentioned in the introduction of his paper that the types of all the species he described will be sent to the Muséum national d’histoire naturelle (P) in Paris, when the war would be over. Haevermans et al. (2009) considered this name as invalidly published, and Castillon (2020) correctly claimed that it was not the case, as Boiteau (1941) mentions the preparation of types to be deposited at P. In his paper, Castillon (2020) went to great length to attribute the name E. leandriana to plants he found in 2018 west of “Berohara” [likely Beroroha] and published photographs of it. These photographs showed plants resembling the one (Jardin Botanique, plante n°11) illustrated in Ursch & Leandri (1954: plate 43) depicting what they thought was E. leandriana (Fig. 39B, here reproduced). Circumstantial evidence based on Humbert’s travels also led Castillon (2020) to suggest Humbert 11346 (MNHN-P-P00217916!, reproduced here in Fig. 39A) is E. leandriana and designate this specimen as the lectotype of E. leandriana Boiteau ( Castillon 2020: 50) , even though this specimen cannot be considered as being part of the protologue, thus not a lectotype. The author also put E. tsimbazazae Leandri (1946: 162 , nom. invalid.) in synonymy with E. leandriana , likely because Leandri posteriorly added the mention “ E. leandriana Boiteau ” on a specimen bearing the mention “plant 11”.

In 2017 we found two pieces of evidence showing that another species, E. horombensis Ursch & Leandri (1954) is the same as what Boiteau described as E. leandriana in 1941. First, we found a box in the Paris Herbarium with dried material originally collected in alcohol. It contains a label in the handwriting of Jean Bosser, who obviously copied the text from the original label written by Pierre Boiteau, as J. Bosser also copied the signing at the bottom of the label, reading “P. Boiteau”. This label identifies the sampled material as E. leandriana Boiteau and contains text fragments directly from the protologue of E. leandriana by Boiteau. Probably J. Bosser copied the label because the original had gone bad or became difficult to read. The description of the cyathophyll colors on the label match the protologue as being yellow and red-orange. The thick stems, sturdy spines with laterally compressed bases, oblongoval leaves and the small cyathia immediately reminded us of E. horombensis . The second piece of evidence followed when we consulted J. Bosser’s slide collection, in which we found several slides of original plants growing in the early 1960s in the Parc Botanique et Zoologique de Tsimbazaza. Bosser duplicated the name tags, and one slide says “ E. leandriana ” and is obviously taken from the type plant, then still alive (see Fig. 28A). The picture shows unequivocally a cream form of the species later named E. horombensis by Ursch & Leandri. With both pieces of evidence, and based on both species protologues, we do not hesitate to identify E. leandriana as being identical with the well-known E. horombensis , the latter name here presented as a new synonym of E. leandriana . One could ask why Ursch & Leandri (1954) presented E. leandriana and also their new species E. horombensis in the same publication. Their differentiating character was the size of the inflorescence, which is said to be small (1–1.5 cm) in E. leandriana and larger (10 cm) in E. horombensis . However, both should have inflorescences with up to 20 cyathia according to the accompanying key—this would be very hard to envision in an inflorescence of just 1.5 cm, so we do not trust this observation.

From the way the 1954 Ursch & Leandri paper is structured, it is now clear that there was likely a mixing up by the authors of the identity of the plants used to describe E. leandriana by Boiteau and another plant (cultivated under #11) sent by Boiteau and used by Leandri as the type of E. tsimbazazae Leandri. Euphorbia leandriana appeared in Leandri (1952: 110) in a key at a position where it was supposed to possess cyathophylls with an acumen “as long as the cyathophylls itself”. Leandri (1953: 141) mentioned that E. leandriana is represented by live plants brought back by Humbert under numbers 222-228, repeating Boiteau (1941), and that the plant #11 is E. tsimbazazae . Then Ursch & Leandri (1954: 156) decided that Euphorbia leandriana is represented by the plant #11, despite the fact that nowhere in Boiteau’s work there was a text linking this number 11 to E. leandriana . On the same page Ursch & Leandri mentioned they now doubted the existence of E. tsimbazazae , initially typified on a specimen prepared from “plante #11 sent by Boiteau in 1939”, the type of which is now mentioned as “a plant cultivated in Jardin de Tsimbazaza in 1939”, without number 11, and “possibly also a hybrid”. They also corrected the assertion in Leandri (1952: 110) that E. leandriana had acuminate cyathophylls.

The specimen Ursch s.n. (MNHN-P-P00078021!) which was identified by Haevermans et al. (2009: 292) as the holotype of the name E. tsimbazazae , based on it bearing the number 11 from the Tsimbazaza Botanical Garden, is now rejected by us as holotype of the name E. tsimbazazae (nom. inval.). The original label contains a text by Ursch saying “Pl 32. Euphorbia . Nov. Dec. Parc Botanique.”. From this text, the part “Pl 32” is stricken and a number 11 is put in front of it, in Leandri’s handwriting, proven by the second label on the sheet annotated by Leandri, which also mentions the number 11 in exactly the same handwriting as the nr. 11 on the earlier label. So, for whatever reason, Leandri wanted this specimen to be the nr. 11 from the garden in Tsimbazaza, to which he attached the name E. tsimbazazae in 1946 and 1953. One can only guess why they did not use the alcohol specimen sent by Boiteau to Paris, identified as E. leandriana by Boiteau himself. Then they would have recognized it to be what they eventually published as E. horombensis . In any case, Leandri put a new label on the Ursch specimen, identifying it as E. leandriana and putting a number of his own on it, Leandri 2384.

Because of this and the fact that there now appears to be no existing original Boiteau material for E. tsimbazazae , Castillon’s (2020) synonymy is rejected here. All these facts lead us to consider the invalidly published name E. tsimbazazae as uninterpretable.We think that the plants figured in Castillon (2020) wrongly illustrated as E. leandriana , represent an undescribed species of which material has been seen in cultivation over the last ca. 5 years, and is here described as a new species, named E. psammiticola Haev. & Hett. (see further). We further consider the plate in Ursch & Leandri (1954: Plate 43) copied in Castillon (2020), and here in Fig. 39B, to wrongly represent E. leandriana (see above). It may represent our new species E. psammiticola , but we cannot establish it. Euphorbia leandriana is very variable in inflorescence characters and colors. The first author (T.H.) saw many plants in the wild (then identified as E. horombensis ) from various localities and noticed that for instance those in the Isalo-range can be with inflorescences of varying sizes and present a wide array of colors (Fig. 28). In Haevermans et al. (2009: 286), a lectotype for the name E. horombensis was designated using the only syntype from the protologue that could be located (corresponding to Jardin de Tsimbazaza n°5). A photograph of the live plant known as J.B. n°35 (one of the former syntypes) was recently found in the Bosser slide collection (September 1962, reproduced here as Fig. 28B).

Euphorbia mangokyensis Denis (1921: 80 , Fig. 22) (Figs. 11A & 30). Lectotype (designated by Haevermans et al. 2009: 288):— MADAGASCAR. Toliary: Menabe, “Rocailles boisées sèches des gneiss et cipolins, bassin du Mangoky, rive droite de la Menamaty”, [21°27′30″S 44°31′0″E], September 1917, Perrier de la Bâthie 9660 (lectotype: MNHN- P-P00078039!, isolectotype: MNHN-P-P00078040!). [Lectotype reproduced here in Fig. 30].

— Euphorbia razafindratsirae Lavranos (2002: 6) , syn. nov. Type :— MADAGASCAR. Toliary: Atsimo-Andrefana, “in silva xerophila prope urbem Tongobory”, 275 m elev., [23°31′30″S 44°19′60″E], 1999, Razafindratsira s.n. (holotype: MNHN-P-P00577114!).

Additional specimens examined:— MADAGASCAR. “ Ouest : Bassin du Mangoky , rive droite de la Menamaty”, [21°45′0″S 45°30′0″E], November 1911, Perrier de la Bâthie 9650 (MNHN-P-P00221921!, MNHN-P-P00221922!, K000185196!). Toliary: Atsimo-Andrefana, “Ouest/Centre: Grès de l’Isalo et Mt Vohibasia l’Onilahy et le Mangoky”, [22°36′0″S 44°49′0″E], July 1910, Perrier de la Bâthie 9770 (MNHN-P-P00221923!) .

Note:—The protologue and description of E. razafindratsirae leave no doubt that it is the same species Denis (1921: 80, Fig. 22) described as E. mangokyensis . The unique habit with the subterranean irregularly tuberous main plant body, from the top of which grow snake-like, horizontal, twisting branches and rather soft, disappearing or partly remaining spines, is unique for Euphorbia in Madagascar (see Fig. 30). The closest similarity in Madagascar is with E. hofstaetteri Rauh (1992b: 112) , which has a perfectly rounded caudiciform base with much sturdier, straighter and often more erect growing branches with hard, persistent spines.

Euphorbia pachyspina Haev. & Hett. , sp. nov. (Figs. 9A & 31). Type:— MADAGASCAR. Fianarantsoa: Ihorombe, Ihosy, “ Chaîne de l’Ambinda , à l’Ouest d’Ivohibe”, 1300 m elev., [22°34′0″S 46°22′0″E], 26 September 1926, Decary 5652 (holotype: MNHN-P-P00221993!) [reproduced here in Fig. 31].

Diagnosis:—Similar to Euphorbia splendens Bojer ex Hook. , but differs from it (and all other spiny, red-flowered Euphorbia species from Madagascar) by the swollen, spindle-shaped, spine bases and orange-bronze color of its young stems.

Description: —Shrub 60–80 cm high, branched, with angular branches at least 1.5–2 cm thick, shiny brownishgrey, stipular spines robust and slender, 10–20 mm long, borne on a fleshy bulbous base, slightly pointing towards the apex of the branches, leaves obovate, 20–50 × 15–25 mm. Inflorescences subterminal, large (main peduncle 40–50 mm long), with 4–16 cyathia each. Cyathophylls large, 6 × 10–12 mm, scarlet red. Fruit and seeds unknown.

Note:— Known only from the type. This taxon may be conspecific with Rebmann’s (2016) concept of E. milii subsp. breonii , see notes in E. hislopii , above.

Euphorbia paulianii Ursch & Leandri (1954: 132) . (Figs. 29, 32, 35, 38B & 38D).Type:— MADAGASCAR.“Antsingy [?], échantillon fait par E. Ursch au Jardin Botanique de Tananarive n°9”, [16°25′55″S 45°19′45″E], s.dat., Leandri 2382 (holotype: MNHN-P-P00078064! [partial leaf & inflorescence]; isotype: MNHN-P-P00078065! [inflorescences]) [holotype reproduced here in Fig. 32].

— Euphorbia perrieri var. elongata Denis (1921: 88 , excl. Fig. 24 = Euphorbia perrieri Drake ), syn. nov. Type :— MADAGASCAR. Antsiranana: “Centre: Sambirano: massif du Manongarivo, alt. 800 m ”, [14°0′0″S 48°22′0″E], April 1909, Perrier de la Bâthie 9916 (holotype: MNHN-P-P00078071!) [reproduced here in Fig. 34].

Additional specimens examined:— MADAGASCAR. Antsiranana: Diana , “Presqu’île d’Ampasindava, forêt d’Ambohimiravahavy, à 40 km au sud-ouest d’Ambanja”, 456 m elev., 13°45′32″S 48°5′27″E, 1 March 2011, Aubriot et al. 154 (MNHN-P-P02284524!, TAN) . Mahajanga: Sofia , “Maromandia (Sandrakoto)”, [14°18′0″S 48°27′0″E], 18 September 1922, Decary 1103 (MNHN-P-P00221537!) . Antsiranana: Diana , “Ambongomirahavavy - Bord de la rivière Manongarivo, entre MG11 et MG12/Faritanin Antsiranana ”, 280 m elev., 13°59′12″S 48°18′32″E, 28 September 1996, Gautier 3155 (G!, TAN) . “ Ampasindava , forêt de Bongomihiravavy”, 405 m elev., 13°45′0″S 48°5′24″E, 17 November 2008, Nusbaumer 2851 (G, MO!, TAN) . “ Ouest: Cultivée dans mon jardin— Le pied provenant du Ht Bemarivo ”, [14°4′30″S 49°11′56″E], March 1910, Perrier de la Bâthie 9605 (MNHN-P-P00221536!) . “ Massif Manongarivo. W slopes of peak E of Beraty. Dense humid forest”, 1000 m elev., 14°1′60″S 48°16′60″E, 2 July 1987, Phillipson 2026 (MO!, MNHN-P-P00078072!, TAN) .

Note:—The identities of both E. paulianii and E. perrieri Drake have been unclear since their publication, both in literature and in cultivation. The similarity of their inflorescences has been used as an argument suggesting they are the same species. Another theory, originating with Ursch & Leandri (1954: Fig. 29) and repeated by Rauh (1962), states that E. paulianii has inflorescences with around 300 cyathia, but that is a misinterpretation of the type of E. paulianii , which contains several inflorescences, which all together appear to be one (Figs. 32–33). Further issues relate to the supposed locality of origin of the original plant of E. paulianii , grown in the Parc Botanique et Zoologique de Tzimbazaza (Antananarivo) and from which the type specimen was prepared: the label on the type says “Antsingy”, which could be one among the separate tsingys [dogtooth limestone formations] of Ankarana, Bemaraha or Namoroka. Euphorbia paulianii as we now understand it, is absent from these localities and favors the wetter Manongarivo area, whereas E. perrieri as we understand it, grows in the tsingy of Namoroka. We conclude that the information about the source of the living plant of E. paulianii is in error or at least it is uninformative. The identity of E. paulianii became clear when we found a slide by Jean Bosser of the type plant of E. paulianii (see Fig. 35), which was still growing in the Botanical Garden of Antananarivo in 1960. His slide shows that this is the plant widely distributed in cultivation under the name E. perrieri var. elongata Denis , characterized by a narrowly oblong leaf, simple spines in regular, straight or helically arranged crests, and cyathophylls ca. 4 mm long, surpassing the cyathial cup and with the upper part usually spreading and pale yellowish or yellowish greenish (Fig. 35). The leaf is incompletely represented by the type of E. paulianii (see Fig. 32) because the base and central half of the leaf are missing and both remaining parts are glued almost against each other, suggesting one continuous, broadly elliptic lamina, instead of elongate oblong as in the protologue of E. paulianii and in the protologue and type (see Fig. 34) of the synonymous E. perrieri var. elongata . Rauh (1962) was correct in showing pictures of a living plant in his treatment of E. paulianii , which conforms to the same plant as the one grown in Antananarivo. But in 1995, Rauh mixed up both species considerably, keeping E. paulianii separate from E. perrieri var. elongata and mixing var. elongata with the true E. perrieri ( Rauh 1995: Fig. 923), and E. paulianii with the true E. perrieri ( Rauh 1995: Fig. 926). We checked the type of the name E. perrieri var. elongata (P, reproduced here in Fig. 34) and the protologues of E. paulianii and E. perrieri var. elongata , and concluded that they both represent the same taxon, which we propose to recognize at the species level and for which the name E. paulianii has priority over E. perrieri var. elongata . The association of the epithet elongata with the species name E. perrieri by Denis is discussed below under E. perrieri .

Euphorbia perrieri Drake (1899: 308) . (Figs. 29, 33, 36, 37, 38A & 38C). Type :— MADAGASCAR. Mahajanga: [Betsiboka, Maevatanana], “Rochers humides Firingalava”, [17°36′0″S 46°54′60″E], April 1898, Perrier de la Bâthie 571 (holotype: MNHN-P-P00078070!).

Additional specimens examined:— MADAGASCAR. Mahajanga: Boeny, Soalala, “Namoroka National Park, N

part of park to E of Namoroka village ”, 121 m elev., 16°24′35″S 45°20′15″E, 27 October 2016, Luino 288 (MO!,

TAN). “Rés. N° 8 [Namoroka]”, [16°26′60″S 45°21′0″E], October 1964, Morat 840 (MNHN-P-P00221531!). “Rés. N° 8 [Namoroka]”, [16°26′60″S 45°21′0″E], October 1964, Morat 936 (MNHN-P-P00218000!). “Ouest: Rochers calcaires du Namoroka (Ambongo)”, [16°22′60″S 45°17′60″E], August 1903, Perrier de la Bâthie 571 bis (MNHN-P- P00221534!). “Ouest: Boïna”, Perrier de la Bâthie 9605 bis et 571 bis (MNHN-P-P00221535!). “District de Soalala, Commune Rurale Andranomavo, Parc national de Namoroka, tsingy de Namoroka, situé à l’ouest du village Vilanandro (environ 3 km). Formation sur Tsingy au NE du lac Antsofotra” , 117 m elev., 16°25′42″S 45°22′9″E, 11 September 2012, Rakotovao 6047 (MO!, TAN) .

Note:— The type of E. perrieri (Fig. 33) shows a leaf type much more elliptical and shorter than that of E. paulianii and with a clear petiole, which is missing (or nearly so) in E. paulianii . There is a specimen, Perrier de la Bâthie 571bis (MNHN-P-P00221534!), which we agree with Denis (1921) is E. perrieri . This specimen contains a part of the stem, showing a spinal arrangement different from that of E. paulianii , E. perrieri having a smaller base and a longer thin upper part and additional, smaller spines on the base of the main spine and a much more irregular arrangement in vertical or spiralling lines but often also almost without any ordering. This latter character is very rare in E. paulianii , in which the base of the spines is much larger than the free part and forms very prominent crests along the main stem. Another difference with E. perrieri is the length and shape of the cyathophylls, being only 2 mm long in E. perrieri (vs. 4 mm in E. paulianii ), and as long as the cyathial cup, never spreading and with a dark greyish blackish colour. The note seemingly added later on specimen Perrier 571bis mentions that the species grows on shaded rocks in gorges of Firingalava and among shaded limestone rocks in Namoroka (Ambongo). The latter is confirmed by us through fieldwork in the area, additional material in P, and observations and pictures from Mr. Petr Pavelka.

Denis (1921: 88) described his variety E. perrieri var. elongata with the protologue being accompanied by a drawing ( Denis 1921: Fig. 23). In the P herbarium we found a black-and-white picture of a living plant, which is very clearly the basis of Denis’ drawing but annotated as E. perrieri Drake and carrying the number ‘9605’ (Fig. 36A). In his enumeration of specimens of E. perrieri Drake , just above the protologue of E. perrieri var. elongata, Denis includes the specimen Perrier de la Bâthie 9605, cultivated in Perrier’s garden but supposedly from Haut-Bemarivo, making this locality compatible with the natural distribution of E. paulianii around the Manongarivo area. We examined this specimen (MNHN-P-P00221536!, reproduced here in Fig. 37A), and it appears clearly prepared from the plant in the black-and-white picture, also bearing the number 9605 (Fig. 36A). Another specimen from P, likely prepared from the same plant, is labelled Perrier de la Bâthie 9605bis (MNHN-P-P00221535!, reproduced here in Fig. 37B) from “West: gneiss rocks of Haut Boïna” (with the mention “571bis” on the same sheet, emphasizing the posterior numbering of Perrier’s collections). This is also Euphorbia perrieri , but with a locality much more compatible with the natural distribution of this taxon. This indicates that the supposed Haut Bemarivo origin of Perrier de la Bâthie 9605 is likely wrong and confirms the plant is really E. perrieri Drake. The plant shown on the picture (Fig. 36A) and the Perrier de la Bâthie 9605 specimen prepared from it, is clearly Euphorbia perrieri Drake , in accordance with Denis’ inclusion of this specimen under E. perrieri . This tends to indicate that the figure in Denis (1921: 89, Fig. 23), reproduced here on Fig. 36B, has a wrong caption and represents E. perrieri and not, as per the caption, E. perrieri var. elongata (= Euphorbia paulianii ).

We therefore propose to exclude this drawing from the protologue of E. perrieri var. elongata Denis. At the same time, we conclude that the black-and-white picture is that of a living plant of the hitherto difficult to interpret name E. perrieri Drake and does agree with the protologue. It shows a plant similar to recent observations and dried specimens of plants found in Namoroka. This also shows a clear geographical disjunction between E. paulianii in the north (Sambirano/Manongarivo area) and E. perrieri further south in the (Namoroka massif). The type locality of E. perrieri is “Gorges du Firingalava” which is not very close to Namoroka, where all modern collections of this species come from. This can be due to an error in the recording of the verbatim locality of Perrier de la Bâthie 571 (Fig. 33).

Euphorbia psammiticola Haev. & Hett. , sp. nov. (Figs. 39A, 40B & C). Type :— MADAGASCAR. Toliary : Atsimo- Andrefana, Beroroha, “Vallée du mangoky. Escarpement gréseux au NW de Beroroha ”, 300 m elev., 21°40′60″S 45°10′0″E, October 1933, Humbert 11346 (holotype: MNHN-P-P00217916!) [reproduced here in Fig. 39A].

Diagnosis:— Euphorbia psammiticola resembles E. leandriana in general habit, leaf shape and cyathia but the latter has much thicker branches, broader leaves with glabrous abaxial surface, the inflorescence carrying 4–32 cyathia (vs. 2–4 in E. psammiticola ), the basal part of the cyathophylls of opposite pairs touching or overlapping, the spreading part of the cyathophylls shorter and transversely elliptic with a more or less truncate top (vs. obcordate with an acute top in E. psammiticola ), the cyathial top rising above the level of the spreading part of the cyathophylls, the inside of the cyathium wall glabrous (vs. having 5 vertical rows of acicular hairs in E. psammiticola ).

Description:—Erect, deciduous, spiny, bushy plant, branching frequently from the main stem, branches horizontal or obliquely erect with the leaves concentrated at their tips. Leaves lanceolate, 8–15 × 1–2 cm, short petiolate (petiole 2 mm long), top acute with a 2 mm long needle-like apiculum, adaxial surface green with reddish margin, abaxial surface paler and with short acicular hairs, distantly spread but denser close to the midrib. Stipular spines simple with expanded base, 0.5–1.5 cm long, dark reddish brown when young, pale brown or grey when old ( Fig. 40B ). Inflorescences (1–) 5–10 cm long, common peduncle to 7 cm long, reddish brown, sticky, once or twice (-thrice?) dichotomous, carrying 2–4(–8?) cyathia. Cyathophylls 1 cm long, lower part of opposite cyathophylls not mutually overlapping, spreading part 7 × 8 mm, obcordate with broadly acute apex and 1 mm long apiculum, adaxial surface whitish to pale yellowish, abaxial surface paler. Cyathium elongate, subcylindric, 4 × 2 mm at the base, up to 2.5 mm in diam. at the top, top remaining below the level of the spreading parts of the cyathophylls, inside of the wall with 5 vertical rows with short acicular hairs, each one situated below the middle of a cyathial gland. Cyathial glands transversely elliptic, cup-shaped. Interglandular bracts elongate, reaching beyond the glands, lateral margins fimbriate, upper margin dentate. Staminate flowers filament pinkish, pedicel white ( Fig. 40C ). Pistillate flowers ovary glabrous, style trichotomous from the base, style branches thin, wiry, white or pale pink.

Etymology: —The species epithet is derived from the ecology (growing on sandstone base rock) detailed on the type specimen label.

Note: —Known only from the type. Maybe the invalidly published E. tsimbazazae Leandri is the same taxon (see discussion under E. leandriana Boiteau ). This species corresponds to Euphorbia leandriana sensu Castillon (2020) , non Boiteau (1941).

Euphorbia werneri Haev. & Hett. , sp. nov. ( Figs. 40A , 41 & 42). Type:— MADAGASCAR.Toliary:Anosy, Taolagnaro, “Bassin de réception de la Mananara, affluent du Mandrare. Pentes occidentales des Montagnes entre l’Andohahela et l’Elakelaka”, 600 m elev., [24°40′0″S 46°43′0″E], January/ February 1934, Humbert 13732 (holotype: MNHN-P- P00217909!) [Reproduced here in Fig. 41].

Diagnosis:— Resembles superficially E. imperatae but has thicker, greyish-brown shiny branches, larger and wider dark green oval to obovate leaves and smaller, yellow cyathophylls.

Description (adapted from Rauh 1967b, under E. delphinensis non Ursch & Leandri):—Shrubby, spiny, herbaceous plant, much branched, ca. 50 cm high, branches semi-erect to spreading, reddish-green, greyish in older parts, 1 cm in diam., glabrous, with or without a few brachyblasts. Leaves at the tip of the branches and brachyblasts, deciduous, sessile or very shortly petiolate. Petiole 1 mm long. Lamina obcordate, obtriangular or oval, to 2 × 1.2 cm, leathery, adaxial surface glossy dark green, abaxial surface pale green, top apiculate. Stipular spines, simple, thin, 1–1.5 cm long. Inflorescence subterminal, or lower along the branches, dichasial, very loose, 4–6 cm long, 2–3 times dichotomous, with 4–8 cyathia, branches sticky, green. Cyathophylls erect, only slightly spreading near the top, ca. 5 mm wide, short apiculate, pale yellowish green, often slightly reddish at the margin (Fig. 42A). Cyathium ca. 2 mm long. Cyathial glands 5, pale yellow. Style trichotomous from the base, each branch split dichotomous in their upper half.

Etymology:—The species is named after the late Prof. Werner Rauh of the Heidelberg University and Botanical Garden, who opened up the succulent world of Madagascar for scientists and plant lovers alike and published numerous papers and books on the subject, presenting also a great number of new species.

Note:—Only known from the type. This taxon is the plant Rauh (1967b) wrongly identified as E. delphinensis , see E. delphinensis entry in the present paper.

Revised taxonomy in Euphorbia L. subgenus Chamaesyce Raf. section Bosseriae Haev. & X.Aubriot (Yang et al. 0)

A small group of four taxa of Euphorbia has long been thought to be part of the informal coralliform (coral-like) euphorbias ( Cremers 1984a). Recent molecular studies ( Yang et al. 2012) demonstrated the group’s monophyly and allowing its own subsection within Euphorbia subgenus Chamaesyce Rafinesque (1817: 119) . The four taxa known to date are E. bosseri Leandri , E. platyclada Rauh , E. platyclada var. hardyi Rauh and E. bemarahaensis Rauh & Mangelsdorff. They all share a small shrubby habit, narrow cylindric or flattened cladodes, tiny, bractless cyathia, and tuberous hypocotyls and root bases. They are restricted to southwestern Madagascar. We re-examined all available original and additional herbarium material, living plants and literature to establish the correct names, the application of the names and their taxonomic status in this group. We propose one new species ( E. cylindroclada Haev. & Hett. ), reduce one to synonymy ( E. platyclada ), re-establish the correct application of one name ( E. bosseri Leandri ) and raise one variety to species status level ( var. hardyi Rauh ) and leave one name as is ( E. bemarahaensis Rauh & Mangelsdorff ).

Euphorbia bosseri Leandri (1965: 212) . (Figs. 43B & 40A). Type:— MADAGASCAR. Toliary: Anosy, Betroka , “[Cultivated in Tananarive, originally from] environs de Betroka, 1963”, [23°15′50″S 46°5′30″E], March 1964, Bosser 17621 (holotype: MNHN-P-P00077904!, isotype: MNHN-P-P00077905!) .

— Euphorbia platyclada Rauh (1970b: 46) , syn. nov. Type :— MADAGASCAR. Toliary: “in sylva generis Alluaudiae inter Amboasary et Fort Dauphin apud km 434”, 25 April 1969, Rauh 22170 (holotype: HEID!, in spiritu).

Note:— Leandri (1965) described E. bosseri , a species found sterile by Jean Bosser near Betroka in 1963 and brought into cultivation in Antananarivo where it flowered in 1964. Leandri prepared dried specimens of it for the Paris herbarium from plants growing in the Botanical Garden of Tsimbazaza and brought there by Bosser in 1964. The type material is very scanty and represents a few cylindrical twigs, without leaves and with one tiny cyathium. Luckily, we recently found a picture of the terminal end of a branch of one of the flowering living plants at Tsimbazaza made by Bosser himself, which he provided to Leandri. Bosser wrote on the slide “ E. bosseri Type”. We reproduce that slide here (Fig. 43B), where it clearly represents the species published by Rauh (1970b: 46) as E. platyclada Rauh. The Bosser slide shows a flattened cladode with its complex rugosity and red color, along with pink cyathia, so typical of Rauh’s E. platyclada , but also clearly described as such in Leandri’s protologue of E. bosseri . Rauh (1970b) not only presents his species E. platyclada but also presents a plant identified as E. bosseri Leandri , with a short description and a drawn figure of close-up details (1970b: Fig. 10), appearing more extensively in later publications with habit photographs (e.g. Rauh 1998: 117, Figs. 415–418, 1999a: Figs. 1 & 2). It would appear that Rauh used the name E. bosseri for this plant, based on an interpretation of Leandri’s description of that species but may have overlooked Leandri’s mentioning of a few characters that are not present in Rauh’s plant but identical to Rauh’s characters defining E. platyclada . Also, the type material of E. bosseri in its desiccated state may have led him to believe that E. bosseri is a plant with exclusively cylindric cladodes and so he associated the name E. bosseri with the wrong plant and then redescribed E. bosseri Leandri itself under the name E. platyclada Rauh. Here we correct the error by synonymizing E. platyclada under E. bosseri and describe the “false” E. bosseri of Rauh as a new species ( E. cylindroclada Haev. & Hett. , see below). We noticed from a plant grown in the Arboretum de Chèvreloup (MNHN, Paris) under the name E. platyclada , that E. bosseri grown in the shade tends to produce elongate, cylindric to subcylindric cladodes, which may explain the shape as seen in the holotype and which may have been part of the confusion surrounding its proper identity. Rauh (1970b) described his E. platyclada as a plant with erect and plagiotropic branches, but in Rauh (1998: 118), he mentioned that the species mutated in Heidelberg to produce a second type of erect branches, thinner and greener than of the typical plant. The second author has grown this type of plant and noted that it eventually produces basal plagiotropic branches like the typical plant, and new erect branches may also be broader than initially described and eventually also turn red in brighter daylight. Rauh (1998: 118) suggested to name the mutated plant as a cultivar, E. platyclada var. platyclada ‘Erecta’ (“cv. Erecta”) but it doesn’t comply with the stability and uniformity criterion for cultivars as per the International Code of Nomenclature for Cultivated Plants ( Brickell et al. 2016), and it is also not established accordingly to Brickell et al. (2016: Art. 12.11) due to the use of a Latin designation for a cultivar name. We also noted that the tuberous hypocotyl does not develop when plagiotropic branches are rooted and grown separately. This seems to have an analogy with E. cylindrifolia Rauh & Marn. -Lapost. (1961: 148), where typical plants have a large tuberous main body, whereas rooted, cut-off branches, never develop the tuber.

Euphorbia cylindroclada Haev. & Hett. , sp. nov. (Figs. 43ACD). Type :—[ MADAGASCAR?]. Cultivated, “From a plant cultivated by W.L.A. Hetterscheid originating from the Heidelberg Botanical Garden as Rauh 70591a (HEID140166) and there received from Les Cèdres Botanical Garden in France in July 1998 ”, 13 August 2020, Hetterscheid H.EU.056-T (holotype: P!, in spiritu).

— E. bosseri auct . non Leandri in Rauh (1970b: description and Fig. 10 under E. bosseri Leandri ), Rauh (1998: 117 description, and Figs. 415–418, under E. bosseri Leandri ) and Rauh (1999a: description in the discussion and Figs. 1–4 under E. bosseri Leandri ).

Diagnosis:— Species similar to E. bosseri Leandri , but differs in the completely dark green, terete or slightly compressed stems, with a non-rugulose surface, completely green cyathia and glands, leaves not borne on a tubercle and gland surface smooth, not pitted.

Description:—A very loose bush, straggling, sparingly branched from the base or higher up, to 1 m high in cultivation, base irregularly tuberous, branches and stems 3–5 mm thick, at first erect, then arching and spreading over surrounding objects, articulated, cylindric to slightly flattened in basal parts of new branches, surface with distant, very short, bristle-like, hairs, dark greyish green. Leaves at the tip of the branches, scale-like, quickly drying and caducous, very short petiolate. Petiole ca. 0.3 mm long. Lamina orbicular, vestigial, 0.4 mm long, 0.4 mm in diam., with bristlelike hairs. Cyathium solitary, on a short peduncle, without cyathophylls, cup-shaped, 2 × 3 mm, outside with short, bristle-like hairs. Cyathial glands 5, dirty dark-greenish, trapezoid. Style very short, 0.5 mm long, trichotomous at the top. See also Rauh (1999a: description in the discussion pp. 3–5, Figs. 1–4 under E. bosseri Leandri ) [in French], shortened descriptions in Rauh (1970b: 51, Figs 10–11 under E. bosseri Leandri ) [in German], and Rauh (1998: 117 description, and Figs. 415–418, under E. bosseri Leandri ) [in English].

Distribution:— Unknown, very likely Madagascar. Acquired from the Les Cèdres Botanical Garden collection by Werner Rauh for the Heidelberg Botanical Garden and distributed to other growers from there. All material in cultivation is likely to be from this one clone (see note below).

Notes:—The characters defining this new taxon as presented by Rauh (1970b, under E. bosseri non Leandri) distinguish it well from E. bosseri , and we consider them suited to distinguish E. cylindroclada from E. bosseri . The database of the Heidelberg Botanical garden collection (https://gartenbank.cos.uni-heidelberg.de/public/ gb.php?sid=SID) mentions with the entry HEID140166 E. bosseri (sensu auct. non Leandri), that the plant was originally collected from “ Madagaskar, Betroka, Bachufer [river bank]” but these data likely originated from Leandri’s protologue of the true E. bosseri (see above) and were added in Heidelberg. As for the actual source locality of the E. cylindroclada plant in Les Cèdres, no records of its collector or locality exist. We also know from a personal communication by Marc Teissier, who worked at Les Cèdres, that the plant presently cultivated there originated from Heidelberg. We can conclude that a piece of the original plant was donated to Heidelberg Botanical Garden and that this original plant then died in Les Cèdres, and a piece of the Heidelberg plant was subsequently given back to them. Our suggestion that E. cylindroclada originates from Madagascar is based on its very close affinity with the three other species of this, seemingly natural group, all from Madagascar.

Euphorbia hardyi (Rauh) Haev. & Hett. , comb. & stat. nov. (Figs. 11B & 44). Basionym :— Euphorbia platyclada var. hardyi Rauh (1970b: 46) . Type:— MADAGASCAR. Toliary: “ 50 km in directione septentrionali-orientali a Tuléar distante in sylva sicca inter rupes”, 4 April 1969, Rauh 21849a (holotype: HEID!, in spiritu).

Note:— This taxon was collected 50 km north of Tulear (and 300 km NW of the type locality of E. platyclada ) and described as a variety of E. platyclada Rauh (originally from Fort-Dauphin area) and represents in our opinion, a distinct species due to its distinctive morphology, especially its very flattened and rhomboid base of the new shoots; see Fig. 44E , here, or Rauh (1998: 120, Fig. 428) for an illustration of this peculiar character.

Figures

FIGURE 4. Euphorbia milii. A, B. details of flowering branches, near Taolagnaro (Lake Lanirano); C. epitype of the species, specimen by Des Moulins from plants brought back by the Baron Milius, Des Moulins s.n. (BORD_DESM_081_Euphorbe exotique), with the writing “Euphorbia milii nobis. nov. spec. à Madagascar”. Credits: P. Pavelka (A and B) and Jardin Botanique de la Ville de Bordeaux (C).

FIGURE 8. Distribution map of E. hislopii (red), E. splendens (blue), and E. tananarivae (green). Full symbols indicate a specimen of wild origin, hollow symbols indicate a cultivated plant occurrence.

FIGURE 40. A. Distribution map of Euphorbia psammiticola (red), E. werneri (blue), E. bosseri (yellow), E. delphinensis (green); B. detail of the stem spination and leaves insertion of E. psammiticola in cultivation in the National Tree Museum Gimborn, The Netherlands; C. detail of the cyathium and cyathophylls of E. psammiticola, showing the red-tinged filaments and whitish pedicels of staminate flowers. Credits: W.L.A. Hetterscheid (B) and P. Pavelka (C).

FIGURE 44. Euphorbia hardyi in cultivation in the National Tree Museum Gimborn, The Netherlands. A. habit; B. flowering cyathium, pistillate flowers stage; C. flowering cyathium, young fruit stage; D. detail of the succulent rootstock at branch insertion level; E. young flattened shoot. Credits: W.L.A. Hetterscheid (A–E).

FIGURE. A. distribution map of Euphorbia betrokana (red), E. neobosseri (yellow), E. pachyspina (green) and E. tenuispina (blue); B. drawing of E. milii var. tulearensis from Ursch & Leandri (1954) (=E. tenuispina). Credits: Muséum national d’histoire naturelle, Paris (B).