A Nomenclator of Croton (Euphorbiaceae) in Madagascar, the Comoros Archipelago, and the Mascarene Islands

Abstract All published names of Croton from Madagascar, the Comoros, and the Mascarenes are treated here. We indicate which names are currently accepted (123 native species and 1 introduced), which ones we consider to be heterotypic synonyms (188), which ones are doubtful (25), and which ones should be excluded (5). We newly designate lectotypes for 108 names, and epitypes for C. anisatus Baill., C. nobilis Baill., and C. submetallicus Baill. A total of 133 names are newly treated as synonyms. One new combination is made, Croton basaltorum (Leandri) P.E.Berry for C. antanosiensis var. basaltorum Leandri, and one new name is proposed, Croton toliarensis B.W.vanEe & Kainul. for C. tranomarensis var. rosmarinifolius Radcl.-Sm.


Introduction
This work is part of an effort to lay the taxonomic foundation for a thorough phylogenetic-based revision of Croton (Euphorbiaceae) in the Western Indian Ocean Region (WIOR). Madagascar alone harbors around one third of the roughly 450 species recognized to occur in the Old World . There are also several species known to occur on the Mascarene Islands (Mauritius and the French Department of Réunion), as well as on the Comoros Islands (Union of the Comoros and the French Department of Mayotte). There are no known species of Croton in the Seychelles or in any of the smaller, "Scattered Islands" of the WIOR. Because of the relative proximity of these islands to Madagascar, and the phylogenetic findings of Haber et al. (2017) that recovered all sampled Croton species from the Comoros and Mascarenes within a Malagasy clade, all described Croton taxa from this region are treated together here. The main purpose of this paper is to provide a comprehensive nomenclatural index of the published names of Croton from the Western Indian Ocean Region and to indicate their current taxonomic and nomenclatural status, based on our ongoing taxonomic and phylogenetic work on the genus.
Until 2016, taxa of Croton native to the Comoros, Madagascar, and the Mascarenes were published by just eight botanists in the following 27 books or journal articles: Lamarck (1786); Geiseler (1807); Baillon (1861Baillon ( , 1890Baillon ( , 1891a; Müller (1864Müller ( , 1865Müller ( , 1866; Baker (1882Baker ( , 1883Baker ( , 1887; De Candolle (1901); Leandri (1931Leandri ( , 1935Leandri ( , 1939Leandri ( , 1948Leandri ( , 1957Leandri ( , 1970aLeandri ( , 1970bLeandri ( , 1970cLeandri ( , 1972aLeandri ( , 1972bLeandri ( , 1973aLeandri ( , 1973bLeandri ( , 1974Leandri ( , 1976; and . In 2016, a large number of new Croton taxa from the WIOR were described. Ten new species of Croton from Madagascar were published by Kainulainen et al. (2016) and Berry et al. (2016aBerry et al. ( , 2016b. Then, taking advantage of an overly literal interpretation of Article 29.1 of the International Code of Nomenclature for algae, fungi, and plants (McNeill et al. 2012), Martin Cheek of the Royal Botanic Gardens, Kew, submitted three "preprint" copies of a long-dormant manuscript on Malagasy Croton by the late botanist Alan Radcliffe-Smith to the libraries of Kew, Wisley, and the Natural History Museum in London toward the end of December 2016. By early January 2017, most of the names from that manuscript had been posted on the International Plant Names Index website (www.ipni.org), with an effective publication date of 23 December 2016 (Radcliffe-Smith 2016). This manuscript added 150 newly described names of Croton from Madagascar and the Comoros Archipelago, consisting of 26 new species, three new subspecies, and 121 new varieties. All of these recent names have been evaluated here, and our assessment of their current taxonomic status is presented in Tables 1 and 2 and in the "Incertae Sedis" section at the end of the nomenclator.

Materials and methods
The main herbaria consulted for the WIOR Croton types were G, K, MICH, MAO, MO, P, TAN, and TEF. Scanned images of types from these and other herbaria available on JSTOR Global Plants (http://plants.jstor.org/) were also consulted. All original literature sources were also reviewed.
Specimens are considered holotypes when a single gathering in a particular herbarium is cited in the protologue, and there is only one specimen of that gathering housed there, or if there was a single gathering with no herbarium mentioned in the protologue, or else a single specimen exists in the herbarium where the author was based (ICN Art. 9.1, McNeill et al. 2012;McNeill 2014). In some publications, such as Leandri (1970bLeandri ( , 1973aLeandri ( , 1974 and Radcliffe-Smith (2016), the protologue indicates a holotype, but there is more than one sheet of that collection at the herbarium cited. In such cases, we rely on the notations on the sheets to determine whether a holotype can be identified; if not, a lectotype is designated. When two or more syntypes were cited in the protologue, a lectotype is designated, using our informed assessment of the most appropriate specimen. In a few cases, an epitype is also newly designated, when the condition of the holotype or lectotype is so poor as to make the identity of the species unclear.
Since the name Croton is derived from the Greek κροτων (tick), which is masculine, this is the proper gender for Croton. However, it was sometimes treated as neuter or as feminine by Baillon and Leandri, and this now requires that the epithets with those endings be orthographically corrected to the masculine ending. In such cases, the original spelling of an epithet is given, and users should be aware of the need to search databases using both the original spelling and subsequent, corrected spellings.
Within the taxon citations, if a lectotype is designated among the extant syntypes, the remaining syntypes are also cited. Type locality information is taken directly from the type specimen labels, even if it does not entirely agree with the location given in the protologue. If we add information not present on the specimen labels, such as the province name, or a more modern or accepted spelling of a place name, that information is placed within brackets. Barcode numbers of type specimens are cited when available, but as of late October 2017, the type specimens cited by Radcliffe-Smith (2016) have not yet been assigned barcodes and do not yet appear on the JSTOR Global Plants website. Whenever we newly treat a taxon as a synonym, we include the annotation "syn. nov." at the end of the entry. Lastly, under the "Habit and Distribution" section of each entry from Madagascar, we list the general distribution of the species on the island and include the former province names where they occur in parentheses.

Results and discussion
In the present treatment we recognize 114 accepted species native to Madagascar. All of these species are endemic to the island, with the exception of C. adenophorus Baill., which also occurs on Mayotte in the Comoros. Besides this species, there are four species endemic to the Comoros (C. bifurcatus Baill., C. emeliae Baill., C. humblotii Baill., and C. mayottae P.E. Berry & Kainul.). Another four species are endemic to Mauritius (C. fothergillifolius Baill., C. grangerioides Bojer ex Baill., C. tiliifolius Lam., and C. vaughanii Croizat), and a fifth (C. mauritianus Lam.) is endemic to Réunion. Croton bonplandianus Baill., a native of South America, is the only documented non-native, naturalized species of Croton in the WIOR, having previously been reported from the Mascarenes (Mauritius, Réunion, and Rodrigues; , Coode 1982. We also report here the occurrence of C. bonplandianus from Mayotte. Other species that are likely to become naturalized weeds, but have not been reported from the region, include Astraea lobata L. (formerly Croton lobatus (L.) Klotzsch), C. hirtus L'Hér., and C. glandulosus L. Several non-native species have been cultivated on the islands in the past, presumably for their medicinal or shade properties, such as C. aromaticus L., C. haumanianus J.Léonard, C. laccifer L., and C. tiglium L., but there is no evidence that they have persisted.
For the 114 species that are native to Madagascar, only six are truly widespread throughout the island, occurring in all six former provinces, namely C. catatii Baill., C. goudotii Baill., C. macrobuxus Baill., C. mongue Baill., C. myriaster Baker, and C. stanneus Baill. Three others occur in five provinces (C. chrysodaphne Baill., C. hypochalibaeus Baill., and C. nitidulus Baker), five in four provinces, six in three provinces, and 29 in two provinces. Sixty-five species are known from just a single province. Of the six provinces, Toliara is the richest in Croton species with 53, followed by Antsiranana with 43, Toamasina with 39, Mahajanga with 35, Fianarantsoa with 27, and Antananarivo with 15. While we realize that these six administrative provinces of Madagascar have been superseded by a system of 22 administrative regions that represent subdivisions of the provinces, we did not attempt to further refine the distribution of species, although this will be a promising avenue to pursue in the future for conservation planning efforts. See Fig. 2 for a summary of these figures.
Given the small size of the islands in the Mascarenes and the Comoros Archipelago, and the small area of existing natural habitats, we believe that it is fairly unlikely that any additional Croton taxa will be discovered there. On the other hand, Madagascar harbors a large number of undescribed species, which are currently under study by the authors and awaiting formal description. We anticipate the description of at least 20 more new species from Madagascar based on the material we now have on hand. Many of these come from recently explored areas in the eastern part of the island, but also from relatively well-explored areas in the north and south but belonging to difficult species complexes, such as the coppery, lepidote-leaved species.
The sometimes extensive synonymy presented in this nomenclator (viz., C. adenophorus, C. catatii, C. chapelieri, or C. stanneus) suggests that many species have been poorly understood in the past, sometimes known only from sparse descriptions or poor type material, and numerous taxa are based on a single or just a few specimens. For instance, from Baillon (1861) to Leandri (1939) to Radcliffe-Smith (2016), all three authors recognized Croton chapelieri as an accepted species, but known only from the type, which was of unspecified provenance. The re-evaluation of this species by Kainulainen et al. (2017a), and again in this paper, reveals that C. chapelieri is in fact a widespread species along the eastern littoral zone, and we now consider that it has eight heterotypic synonyms. We should point out that without first-hand knowledge of the species in the field, it may have been impossible to sort out the variability of this species and recognize its restriction to a fairly narrow ecological and elevational zone in sandy coastal habitats. Among earlier workers on the genus in the region, the only one who had any first-hand knowledge of Madagascar was Jacques Leandri. In contrast, we have made four collecting trips to Madagascar dedicated to studying Croton, covering much of the geography of the island, although there are still numerous areas we have not been able to visit. Especially for deciduous species, it is important to collect specimens at different times of the year, since some species flower when leafless and produce their leaves during the rainy season. Ultimately, the greatest progress in understanding the diversity of a large, complex genus like Croton on Madagascar will be achieved by local botanists who are able to study populations in the field, revisit sites at different phenological stages, and assess their local and regional variability.
Another instructive example of a previously misunderstood species is Croton chrysodaphne Baill., a tree species that Baillon (1861) described based on three syntypes from Madagascar. Baillon's concept of this species was followed by both Leandri (1939) and Radcliffe-Smith (2016), but  determined that the three syntypes in the protologue actually represent three different species, and while one of the others had a valid name (C. argyrodaphne), it took several more years to determine with certainty that the third one was a new species, C. cupreolepis P.E.Berry, B.W.van Ee, & Kainul. (Berry et al. 2016b). As a further example,  also deciphered the history of C. multicostatus Müll.Arg., a Malagasy species which had erroneously been attributed to Hispaniola in the Caribbean, and included as synonyms two species names from Madagascar, C. vernicosus Baker and C. sclerodorus Baill. Figure 1A shows an example of foliage that is similar in all these coppery-lepidote tree species, while Fig. 1B-G show major differences in flowers of four species in this assemblage in terms of sepal shape, stigma morphology, and presence or absence of petals in the pistillate flowers, as well as differences in the number of stamens in the staminate flowers.
The recent, hastily published work of Radcliffe-Smith (2016) appears to us to illustrate the problem of a lack of field knowledge in the region and the consequent lack of understanding of the variability in many Croton species. In some cases, Radcliffe-Smith picked up on important differences in specimens, but he was very inconsistent in how he treated them. Out of the 150 new taxa that he described, 89 were known only from the type collection, and another 29 were known from just two collections. Such a narrow circumscription tends to recognize every minor variation as a separate taxon. On the other hand, in some cases very distinctive new species were treated as varieties of previously known species. For example, Croton bemaranus Leandri is a small shrub, and what Radcliffe-Smith (2016) described as C. bemaranus var. pseudolepidotus Radcl.-Sm. is a very distinctive large tree that was rediscovered in the field and described by Berry et al. (2016a) a few months earlier as C. aleuritoides P.E.Berry. Our general stance for recognizing varieties for Malagasy crotons is that we do not yet have a sufficient level of understanding of the majority of the species to make meaningful subspecific designations. This may come in time, but the fact that more of Radcliffe-Smith's (2016) published varieties fell into synonymy under different species than the ones to which they were assigned (see below) is an indication that we are not yet at the stage where widespread designations of infraspecific taxa are advisable. The overall disposition of the taxa described by Radcliffe-Smith (2016) is given in Tables 1 and 2. Of the 26 species and three subspecies that he described, only four are maintained here as accepted taxa (Table 1), and all others are considered to be There are 123 native species overall in the region, with 114 native to Madagascar; one of them (C. adenophorus) is shared with Mayotte, and the single species in the Union of the Comoros (C. humblotii) also occurs on Mayotte. The species on Mauritius and Réunion only occur there. The map of Madagascar shows how many species occur in each of the six former provinces (there are varying levels of overlap between provinces; see text for details). synonyms of previously described species. Of the 121 new varieties described, we synonymize 97, more often than not under different species than the ones to which they were assigned (Table 2). Of the remaining 24 varieties described, 22 are included in the 'Incertae Sedis' section and two are not validly published (see Names Not Validly Published). Based on this outcome, we contend that the manner in which the new taxa in Radcliffe-Smith (2016) were brought to effective and valid publication, namely by the hand-delivering of three copies of a minimally reviewed and edited manuscript to several London libraries while simultaneously having the taxa entered into IPNI at Kew, avoided a badly needed, more rigorous review of the manuscript and led to the useless creation of a great many names, in contravention of Preamble paragraph 1 of the ICN (McNeill et al. 2012). In our view, the Radcliffe-Smith (2016) publication Table 1. New species and subspecies of Malagasy Croton described by Radcliffe-Smith (2016) and their treatment in this paper. Radcliffe-Smith described 26 new species and 3 new subspecies, besides the 121 varieties treated in Table 2. The four Radcliffe-Smith names in bold below are the only ones maintained here as accepted taxa.
is not at all an accurate reflection of our current knowledge of Malagasy Croton taxonomy, and it should not be consulted as such. Rather, with this paper, previous ones we have published Berry and Kainulainen, 2017;Berry et al. , 2016aBerry et al. , 2016bKainulainen et al. 2016Kainulainen et al. , 2017aKainulainen et al. , 2017b, our initial molecular results Haber et al. 2017) and with more complete molecular studies forthcoming shortly, as well as additional taxonomic novelties and revisions in preparation, we are generating a significantly different and better substantiated vision of the rich diversity of Croton in the Western Indian Ocean Region.  Leandri (1939) named the species after the type locality, Anadabolava, but he may have inadvertently or even purposely omitted the first two letters of the name. The different varieties described by Radcliffe-Smith (2016) all appear to be just minor variants within the normal variability of this species that do not merit taxonomic status, but more intensive work is needed on this and the other small-leaved species from Toliara Province. Notes. Croton adenophorus was substantially recircumscribed from the concept of Leandri (1939) and Radcliffe-Smith (2016) by Kainulainen et al. (2017b). In both earlier publications, C. loucoubensis was treated as a synonym of C. adenophorus, and C. subaemulans was recognized as a distinct species. This earlier synonymy was due to a fundamental misunderstanding of C. adenophorus. See Kainulainen et al. (2017b) for further details and the distinguishing features of C. adenophorus and C. loucoubensis.

Croton adenophorus
The sheet P00466148 is listed in Sonnerat under Croton tulasnei, but without an image. We never saw this specimen at P either before or after the herbarium renovation in 2015, and the whole folder of C. tulasnei was missing during visits to P in 2016 and 2017. Radcliffe-Smith (2016) mentioned Boivin 2187 at P as the lectotype for Croton adenophorus but failed to state "designated here" or an equivalent statement, so his designation was not validly published. Notes. This is a distinctive high-elevation species with well-developed petals in the pistillate flowers and large capsules, belonging to the Mongue Group of Leandri (1939 Notes. There are two sheets of Goudot s.n. at G within the same jacket, meaning they are considered as being part of the same collection (see Gautier et al. 2016). However, they do not fit the criterion of bearing a single label in common. Each sheet has its own label, and they are not identical. One, the lectotype, has a precise collecting date and the locality name of Ambanivoules, whereas the other sheet (G00446337) does not, and the written field description of the plants differs between the two. Also, the lectotype has mainly pistillate flowers open whereas the other syntype has only staminate flowers open. This leads us to conclude that the two sheets are actually different gatherings, even though both were annotated in Baillon's hand as "Croton ambanivoulense." We have designated the sheet that is most consistent with the protologue (G00446336) as the lectotype.

Croton alchorneifolius
Ambanivoules were an ethnic group of eastern Madagascar located approximately 80-100 km west of Tamatave, their name derived from the Malagasy "Antanbanivolo," or "people living at the base of the mountains covered with bamboos" (Schatz 2013).

Habit and distribution. Shrubs; eastern coast of Madagascar (Toamasina).
Notes. Since the holotype of Croton anisatus has only young inflorescences in bud and is known only from cultivation on a quite different island, we designate an epitype with open flowers. This littoral species can be characterized by its very congested inflorescences and the pseudoverticillate, anisophyllous, and sparsely lepidote-pubescent leaves with an entire margin. The plant is apparently quite aromatic as indicated by the descriptions of both the holotype and epitype, as well as by its specific epithet.

Notes.
In its stellate-pubescent and cordate leaves, Croton anosiravensis is superficially similar to some of the species in the Adenophorus Group. However, it does not have opposite leaves or laminar glands, and is probably not closely related. It appears to be a rare species, because besides the type from the northern slopes of Montagne des Français, it is otherwise only known from Analamera (Humbert 19141 [P]) and Befarafara in Daraina (Rakotonandrasana et al. 1048 [CNARP, MICH, MO, P, TAN]). Radcliffe-Smith (2016) stated "holo: P" for var. pilosus, but there are two sheets of the type collection there, so we have selected the more complete of the sheets at P as the lectotype. Habit and distribution. Shrubs to small trees; southern Madagascar (Toliara). Notes. We believe that the lectotype chosen here best conforms to the protologue among the syntypes cited by Leandri (1939). Although most of the other syntypes correspond to the same species, one of them, Humbert 6416, appears to belong instead to C. trichotomus. Notes. Leandri (1972b) designated Richard 218 as the type of Croton argyrodaphne. Given that he did not specify an herbarium in his selection of this collection, we complete the lectotypification here by designating P00127450 as a second-step lectotype. The type of C. argyrodaphne var. orientalis comes from an area in northern Toamasina that is well south of the range of most other C. argyrodaphne specimens. However, there is a second collection of the species from the same area, SF-10816 (TEF), which confirms its occurrence near Fénérive.

Croton argyrodaphne
The type of Croton vohemarensis consists of two small twigs with leaves that are unusually small and wide for C. argyrodaphne. However, the low stamen number (11) and the characteristic stylar column are very typical of C. argyrodaphne (see Fig. 1E-F), and it falls within the geographic and altitudinal range for the species. An additional paratype cited by Radcliffe-Smith (2016), Meyers & Bolz 170 (G, MO), comes from the type locality and is a tree 7 m tall, again with unusually wide and long-petiolate leaves for C. argyrodaphne, but it only has young floral buds.
What Leandri recognized as Croton argyrodaphne var. occidentalis is a rather distinctive element of this species, with a western, subcoastal distribution. The plants are small trees, and the leaves have yellowish pigmentation along the midvein on the adaxial leaf surfaces, but other than that they conform well to the general aspect of C. argyrodaphne. Habit and distribution. Shrubs; southwestern Madagascar (Toliara). Notes. Leandri (1970b) designated Cours 4641 as the holotype of C. aubrevilecta but there are three sheets of this collection at P. One of them, P00312369, has a label in Leandri's handwriting saying "type," and the other two duplicates at P have labels stating "isotype." The isotypes at P have preprinted labels stating "Itinéraire de Didy à Brickaville (forêt orientale)," but P00380444 has a note added later stating "Localité très douteuse, voir récolte Homolle 1944 (avec M.G. Cours)." The holotype has a penciled note stating "Probablement région de Tranoroa." In the protologue, Leandri (1970b) also alluded to the erroneous labels from Didy and states that the collection likely came from far southern Madagascar close to Lake Tsimanampetsotsa. Notes. Although the sheet designated here as lectotype has a sticker stating "TYPE" and the other sheet at P has one stating "ISOTYPE," it is not clear who applied those labels and if it was done after Leandri's publication. In any case, there is no annotation in Leandri's hand on either sheet to indicate which of the two he intended to be the holotype. Notes. The sheet chosen here as lectotype bears two labels, one on the left that lists Pervillé 648 as the collector, and one on the right in Baillon's hand that seemingly attributes the collection to Boivin, stating "ex Ambongo, cum cl. Pervillé et Bernier comm. (1846)." Baillon lists this in his protologue as a separate collection, but it looks identical to the other syntypes, so we believe it is actually part of the same collection by Pervillé.

Croton bernieri
The types lack pistillate flowers, but they are distinctive from C. brevispicatus in bearing a pair of sessile acropetiolar glands and in having relatively long petioles for the size of the leaf, and blades with a rounded-cuneate base and an acuminate apex.  (1861) cited Boivin 2183 and Pervillé 267 when he transferred the taxon to Croton. Both specimens are annotated in Baillon's hand as "Croton (Furcaria) boivinianum H. Bn., Et. Gen. Euph., 356," so we therefore consider them to be original material for the taxon. Notes. Croton bonplandianus is currently the only non-native, naturalized species of Croton in the Western Indian Ocean Region. Oddly, it is known from Mayotte and the Mascarenes, but it has not yet been observed or collected in Madagascar. Notes. Boivin's collection number 2658 was applied to several different collection events. Baillon (1861) was apparently aware of this given the greater detail in which he cited Boivin's specimens, such as "Boivin (1848), n. 2658, Madag., baie de Rigny (h. Mus.)" under Croton brevispicatus and "Boivin, n. 2658, cap d'Ambre" under C. squamiger Baill. Müller (1866) cited C. brevispicatus Baill., as well as the type of that species ("Boivin n. 2658! in hb. Mus. Paris"), in his treatment of C. brachybotryus, which we interpret as an illegitimate name for C. brevispicatus. The isotype at P was a gift from the Caen herbarium received by the Paris herbarium in 1974, and it was never annotated by Baillon, although it had been annotated by Müller. Notes. Unlike most other Malagasy Croton species described by Lamarck from Commerson specimens, there is no specimen of C. cassinoides found in the Lamarck Herbarium at P, so we designate the sheet in the general herbarium at P as lectotype. Most of Commerson's collections from Madagascar came from the Fort Dauphin area in Toliara Province, and all other specimens of C. cassinoides that we have determined are from near Fort Dauphin, so this is likely where the type came from. In the description of C. delphinianus (Baillon 1891a), there was no collection or locality cited (it was the last in the section of Baillon's text and appeared to be cut off). In the descriptions of other new species in the same publication, the Latin description was always followed by a paragraph containing specimen information. At P there is a sheet [P00133052] that has a label with a note in ink stating "Croton cassinoides Lamk. et type du C. Delphinianus H. Bn." It also has a small envelope that reads "Croton Delphinianus Scott-Elliot. n. 1557 Fort-Dauphin." Based on that information, we designate it as lectotype for C. delphinanus. Notes. In this circumscription, Croton catatii is a variable but distinctive species. It is usually a tree and typically occurs in montane habitats, but in the drier areas of Isalo and Zombitse area in Fianarantsoa and Toliara Provinces, it can be shrubby, and plants there have more verrucose fruits than elsewhere. Although plants of C. catatii are typically finely lepidote, plants such as what was desribed as C. catatii var. setosus from Sahafary in Antsiranana Province can be softly pubescent, with trichomes having long-protruding central rays. Notes. Croton chapelieri was accepted by Leandri (1939) but was restricted to the type specimen, due at least in part to the meager type specimen and the lack of any reported collection locality. Similarly, C. aymoniniorum has been recognized only from the type collection. Extensive collections and our own field studies from coastal areas of southeastern Toliara Province (Mandena, Petriky, and Sainte Luce) show that both of these type specimens correspond to a locally common species found in sandy, littoral forests to the west and north of Fort Dauphin, in the Mahabo area of Fianarantosa Province, and then in Toamasina Province much farther north. Chapelier collected mainly in the Foulpointe and Tamatave area (Dorr 1997), and his specimen is similar to the type of C. louvelii, from a nearby area. Many herbarium specimens of this species were identified by the late Alan Radcliffe-Smith and subsequent botanists as Croton daphniphyllus Radcl.-Sm., or sometimes as Croton rhododendroides Radcl.-Sm. for a more pubescent form. See Kainulainen et al. (2017a) for more details about the circumscription of C. chapelieri. Newly added to the synonymy here are C. domohineifolius Radcl.-Sm. and C. macrobuxus var. polygynus Radcl.-Sm., both from littoral sites on the east coast where C. chapelieri is one of the few species to occur. Notes. Baillon (1861) included material of Croton argyrodaphne (Bojer s.n.) and C. cupreolepis (Dupetit-Thouars s.n.), among the syntypes of C. chrysodaphne, which was lectotypified on a Chapelier specimen by Leandri (1972b). See  and Berry et al. (2016b) for further discussions.

Croton chauvetiae
Both Croton meeusei and C. submetallicoides fit well within the variation of leaf size and shape of C. chrysodaphne, but they are reported to have the normal five petals in staminate flowers, whereas specimens from the Foulpointe area (the presumed lectotype locality as well as the epitype locality) have the unusual number of ten petals. We judge this to be somewhat of an anomaly and not a feature that has been fixed in the species overall. The type of C. meeusei lacks pistillate flowers altogether, whereas the type of C. submetallicoides has the short, curved inflorescence and pistillate flowers that are typical of C. chrysodaphne. Habit and distribution. Shrubs; mainly southern and western Madagascar (Mahajanga, Toliara), but also in Fianarantsoa Province.

Croton chypreae
Notes. The Geneva sheet was chosen as the lectotype of Croton crocodilorum, since it is the most complete sheet among the syntypes and contains many seeds in the packet. It was annotated by Leandri as 'Croton crocodilorum Leandri n. sp.,' before it was acquired by Geneva as part of the Herbier Delessert in 1949.
The type of Croton bathianus var. toliarae supposedly differs from the type of C. crocodilorum in its rugulose (vs. smooth) seeds (Radcliffe-Smith 2016). However, wrinkled seeds were also seen in the type of C. crocodilorum, whereas the type of C. bathianus has no fruits. The type of C. crocodilorum var. meridionalis supposedly differs from the type of C. crocodilorum by its denser cinereous indumentum on the lower side of the leaves and larger capsules and seeds (Radcliffe-Smith 2016), but we consider this merely an extreme pubescence outlier in the morphological variation of the species. Notes. Leandri (1939) considered Croton dissimilis to be a synonym of C. ambanivoulensis, but C. dissimilis differs in its larger, usually crenate leaves and echinate ovary with trichomes with a long, porrect central ray. Although the type of C. echinatus has entire leaves (the type of C. dissimilis has crenate leaves), it has similar whitish bark with contrasting tufts of brown trichomes, acropetiolar glands that are cylindrical and shortly stipitate, and the ovary shares the distinctive feature of being covered by long porrect trichomes that give it the appearance of being "echinate." Similarly, the type of C. alaotrensis var. integrifolius has entire to slightly crenate leaves, and it comes from eastern lowland Toamasina Province, near the presumed type locality of C. dissimilis. Based on this, the name "C. muricatus Bojer, nom. nud." has been included in several indices, including IPNI, Tropicos, and Govaerts et al. (2000). In the same manner as Leandri (1939), we interpret Bojer (1837) as a misidentification of what Baillon (1861) later described as C. fothergillifolius, rather than the publication of a nomen nudum.  (1877) distinguished Croton boutonianus from C. grangerioides in his key by the former having entire leaves and pistillate flowers with petals and the latter obscurely crenulate leaves and pistillate flowers that lack petals. Leandri (1939) accepted C. grangerioides, but appears to have overlooked mentioning C. boutonianus, even as a synonym. Coode (1982) treated C. boutonianus as a synonym of C. grangerioides, describing the petals of the staminate flowers as small and delicate, and the crenulation of the leaf margins as variable. We follow here the taxonomy of Coode (1982) in treating C. boutonianus as a synonym of C. grangerioides. Notes. The type of Croton belintae differs from the typically lepidote plants of C. hildebrandtii only in the presence of prominent porrect rays emerging from the center of the lepidote scales, which gives the plant a more fuzzy-pubescent appearance. In all other characters, however, such as leaf shape, petiolar glands, and the small flowers with somewhat recurved pedicels, they are identical. We therefore treat C. belintae as a synonym of the earlier name C. hildebrandtii.

Croton hovarum Leandri, Ann. Mus. Colon. Marseille, sér. 5, 7(1): 40. 1939
Croton rubricapitirupis Leandri, Adansonia, sér. 2, 13: 173. 1973 Notes. Leandri (1973a) distinguished Croton rubricapitirupis from C. hovarum almost exclusively by the sparser, lepidote pubescence on the leaf undersides of C. rubricapitirupis. Croton hovarum is quite variable in leaf size and degree of indumentum, and we do not consider this a sufficient distinction at the species level. In its monopodial branching, large accrescent female calyx, and finely crenate to serrate leaf margins, C. hovarum is a readily recognizable species in upland Madagascar Notes. Leandri (1939) considered Croton hypochalibaeus to be a synonym of C. noronhae, whereas Radcliffe-Smith (2016) considered it to be a synonym of C. jennyanus. Croton hypochalibaeus was accepted by Kainulainen et al. (2016), based on a number of distinguishing morphological and ecological criteria. It is one of the most wide-ranging Croton species in Madagascar.  Govaerts et al. (2000), have listed Baillon (1891b) as the publication in which C. incisus was described; however, the correct citation is Baillon (1861).  Leandri (1972a) called the Perrier de la Bâthie 9788 specimen at P the holotype. We interpret this as a lectotypification of the taxon, and given that there appears to only be a single duplicate of Perrier de la Bâthie 9788 at P there is no need for a second-step lectotypification. Habit and distribution. Large shrubs to small trees; southeastern Madagascar (Toliara).

Croton isomonensis
Notes. Leandri (1939) cited four Humbert collections as syntypes of Croton isomonensis: 13247, 13367, 14070, and 13768. We have not located a Humbert 14070 specimen, but the Humbert 14076 collection has the verbatim collection locality as given in Leandri (1939) for Humbert 14070. We surmise that the '14070' in Leandri (1939)  Notes. We follow here the precedent of Leandri (1939) and Radcliffe-Smith (2016) in treating Croton squamiger as a synonym of C. jennyanus, but we differ in treating C. hypochalibaeus as a distinct, more highland species rather than as another synonym of C. jennyanus (see Kainulainen et al., 2016). According to our interpretation, C. jennyanus is restricted to lower elevations in northern Madagascar (Montagne des Français, Sahafary, Daraina, Ankarana), as well as in far midwestern Madagascar on or near tsingy formations (Bemaraha).
Concerning the type of Croton squamiger, three other sheets of Boivin 2658 correspond to C. brevispicatus, so this is clearly a mixed collection. In the description of C. squamiger, Baillon (1861) divided the species into two infraspecific taxa, a and b. These were subsequently named by Müller (1866) as C. squamigerus var. obtusifolius (the typical variety) and C. squamigerus var. acutifolius, respectively. The latter has been recognized here to be synonymous with C. hypochalibaeus.
Radcliffe-Smith (2016)  Notes. See the note above under Croton cassinioides and its synonym C. delphinianus regarding the Scott-Elliot 1557 specimen (P00133052, upper right), which was also cited by Baillon (1891a) as a syntype of C. lasiopyrus. Croton cassinioides and C. lasiopyrus are sufficiently different that they are not easily confused; the former has smaller (1.5-6 ×0.7-3 cm) elliptic leaves with dentate to subentire margins and grows in littoral zones near sea level, while the latter has larger (4-15.5 × 2.5-7 cm) obovate leaves with entire margins and grows in moist montane forests. Notes. Leandri (1939) treated Croton loucoubensis as a synonym of C. adenophorus, but his concept of C. adenophorus, as shown in his key and description, conforms to the type of C. loucoubensis as treated here. The syntype of C. loucoubensis from Mayotte is sterile and cannot definitively be placed in this species; it could also potentially belong to C. mayottae. Notes. Lamarck clearly attributed this species to the Île de Bourbon, the former name of Réunion, so it is unclear why he named it "mauritianus." At the time of Lamarck's publication, both Réunion (Île de Bourbon) and Mauritius (Île de France) were occupied by the French, and Réunion was administered out of Port Louis, Mauritius. So perhaps Lamarck used the name in a general sense for the islands administered out of Mauritius.

Notes.
It is surprising to us that neither Leandri (1939) nor Radcliffe-Smith (2016) realized that Croton oreades is not distinct from C. mongue. This is the largest tree species among Malagasy Croton, and it is widespread in moist, montane forests.  (1865) mistook the type locality of Croton multicostatus for the Caribbean (Hispaniola) rather than for Madagascar. In that paper, we attributed both the P-JU and P-LA sheets as being a Philibert Commerson collection, but more likely these were collected by Franz Wilhelm Sieber, who visited Madagascar between 1822 and 1825, and included them under his series "j. maut.", as indicated on the P-LA sheet. Notes. The existing syntypes of Croton nobilis at P are all large-leaved (most around 18 × 7 cm), sterile branches. They are consistent with oversized sucker shoots that can sometimes be found on basal growth or regrowth of cut trees. The leaves on these specimens also resemble sucker leaves of C. chrysodaphne, which comes from eastern coastal Madagascar. However, Baillon (1861) clearly stated in the protologue that the staminate flowers of C. nobilis have five petals (the normal state in Croton), whereas C. chyrsodaphne usually has ten . Also, the pistillate sepals of C. chrysodaphne are described by Baillon (1861) as being oblong and entire, whereas those of C. nobilis are broad, squat, and reduplicate (Berry, pers. obs.). The lectotype chosen here contains a packet labeled "Flores," with only a small fragment of an inflorescence with an irregularly flattened and straight rachis, but lacking any recognizeable floral parts that can be reconstructed. The inflorescences of C. chrysodaphne tend to be more slender, subterete, and curved, and this fragment does not fit that profile. Finally, Du Petit-Thouars spent six months in the Fort Dauphin area between 1792 and 1793 (Dorr 1997), so this would be consistent with the limited localities where C. nobilis occurs to the northwest of Fort Dauphin. We therefore conclude that the Dupetit-Thouars collections are consistent with the tree species that has subsequently been collected in moist, mid-elevation forests in the area that is now part of Andohahela National Park, named C. nobilis. To stabilize better this concept of C. nobilis, we designate here a modern epitype that has also been photographed in situ (viz., Fig. 1B) and sequenced for molecular phylogenetic studies (Haber et al. 2017). Notes. A Bojer s.n. specimen at G-DC (G00311742) has a label that states "Insula Mauritius, M. Bojer 1833," and is indicated as a type. The only Bojer specimen cited in the protologue of Croton noronhae (Baillon 1861) is identified as being from Foulpointe, Madagascar, and is cited from P ("h. Mus."). Although the Geneva specimen appears to be correctly identified as C. noronhae, that species is not known from Mauritius. Perhaps it was from a plant native to Madagascar that was cultivated in Mauritius, as was the case with the type of C. anisatus. Notes. The type of Croton submetallicus is deficient for characterizing this species properly; it lacks pistillate flowers, which are very diagnostic. Therefore we are designating an epitype that is well distributed, has photographs in Tropicos®, and is also sampled molecularly. The only other described species name that could be applied here is C. macrochlamys, but it has a very poor type specimen, and we are treating it here as a synonym of C. nitidulus because of its smaller leaf size.

Croton nudatus
For Croton nitidulus var. acuminatus, Radcliffe-Smith (2016) designated Cours 4871 at P as the holotype, but there are four nearly identical specimens of this collection at P, so we selected one of them here as lectotype. Likewise, for C. nitidulus var. macrophyllus, Radcliffe-Smith (2016) designated Cours 4660 at P as the holotype, but since there are three duplicates of that collection there, we selected one of them as the lectotype. Notes. In the protologue of Croton trichotomus, Geiseler (1807) listed two different elements, C. trichotomus from Madagascar and C. punctatus from the Caribbean. We choose here as lectotype the collection in P-LA that Leandri (1939) attributes to P. Commerson. There is no sheet in P-JU that matches the plant of C. trichotomus on the P-LA sheet, but P-JU Catal. 16347 is a mixed collection that bears a label on the left-hand specimen [P00674048] that states, possibly in Geiseler's hand, [Croton] "punctatum Jacq." followed below by "trichotomum Geiseler, Crot. Monogr." However, this is a completely different plant from the one represented on the P-LA sheet, instead belonging to the Caribbean C. flavens L.

Croton tanalorum
In our view, Croton trichotomus is primarily a littoral species of the eastern coast, but it also occurs in a number of more inland and higher elevation situations, which is an exception among Malagasy Croton species.
For Croton antanosiensis var. ambohibyi, Radcliffe-Smith (2016) designated Leandri et al. 1775 at P as the holotype, but since there are four sheets of this number at P, we select one of them as lectotype.
[WIS00000605MAD] Notes.  referenced collector information for characteristics of C. vaughanii that are not observable on the holotype at A ("fide collectoris"), and the holotype is accompanied by a copy of field notes referencing that the species drops its leaves in November and December, and that flowering takes place immediately afterward. According to information in the Mauritius Herbarium (MAU) database, in this case "863" is the accession number rather than the collection number. Given this, the sheets at K and MAU with the accession number 863 are not a part of the same gathering as the holotype at A and therefore are not syntypes. Even though  did not explicitly refer to the "Vaughan 863" collections at K and MAU, we interpret them as being paratypes. Furthermore, the sheet at K labeled with the barcode K000422600 represents material from two gatherings on different dates and should be considered two distinct specimens.
Croton vaughanii is listed by the IUCN as Critically Endangered (Strahm 1998), and is known from only a single individual according to information on the label of Haevermans et al. 558 (P [P00696110]), making it one of the rarest of all Croton species.

Incertae Sedis
This section includes 22 taxa described by Radcliffe-Smith (2016) and three by Leandri (1939) for which we have not yet been able to determine if they are worthy of recognition; they will require further evaluation to decide whether they should be considered synonyms of earlier-described taxa, or if in some cases they are actually distinct species rather than mere varieties of existing species. Notes. Croton bemaranus is restricted to tsingy habitats in Antsiranana and Mahajanga Provinces and has entire leaves and elongate stipules. This variety does not fit the species at all and requires further study to determine its affinities. Notes. This is a very scrappy specimen, making it difficult to determine what it may be. It does not appear to belong to Croton betiokensis, which is confined to southern Toliara Province. Notes. This variety has much smaller leaves than typical Croton catatii and is only a small shrub, so it is not clear yet whether it fits within another species or of it may represent an undescribed species. Notes. It is unlikely this belongs to C. chapelieri (in which C. daphniphylloides is treated here as a synonym); it occurs in the Marojejy Reserve at 500 m elevation and is a 8 m tall tree, whereas C. chapelieri is confined to the eastern littoral zone near sea level. It could also be a northern form of C. submetallicus, with more acute leaves and lacking the submetallic leaf undersides of that species. Notes. This is an interesting specimen that bears some resemblance to Croton tanalorum. Additional material is needed to assess this further. Another specimen of it is Cours 631 (P [P00133297]). Notes. The type is somewhat similar to Croton ericius, but the leaves are much narrower and it is less hirsute than that species. Notes. This variety differ from Croton manampetsae in that the leaves are mostly alternate (vs. opposite) and in the lepidote indumentum on the abaxial side of the leaves (vs. stellate trichomes). It will likely be described as a distinct species. Notes. This plant is similar to C. meridionalis, except that it has considerably wider leaves that are silvery-lepidote on the lower surface and turn dark blackish green on the upper surface when dried. Further study is needed to determine if it should be recognized as a distinct species. Notes. This is likely the same as the preceeding variety. Notes. The leaves of these specimens are too small to conform to Croton meridionalis, and Radcliffe-Smith (2016) also expressed doubt that it belonged there but did not know where else to place it. Notes. The leaves of the type of this variety are too small to conform to Croton meridionalis (for which C. tranomarensis is a synonym). Radcliffe-Smith (2016) expressed doubt that it belonged here, but he did not know where else to place it other than to suggest that it may be close to C. salviformis (presumably because of the small leaves and venose pistillate sepals). However, the holotype appears to have entirely pistillate inflorescences, suggesting that the plants may be dioecious. If so, it may correspond to an apparently undescribed species that we have also found south of Ranopiso at Mt. Vohitsiandriana in Toliara Province.  Leandri (1939, pg. 58) published "Croton arenicola f. pubescens Leandri," but as seen in the preceding paragraph, it was not validly published. Radcliffe-Smith (2016, pg. 188) later attempted to transfer it as "Croton leandrii var. pubescens (Leandri) Radcl.-Sm." Given that the intended basionym is not a validly published name, Radcliffe-Smith's new combination at a different rank is also invalid. Radcliffe-Smith (2016, pg. 188) did include a short diagnosis, which would have qualified as a new name if he had designated a holotype, but since he called the Decary 9031 (K) specimen the "lectotype," Art. 40.6 is not satisfied for valid publication of that name.  Schmelzer (2007), C. haumanianus is commonly used as a shade plant in coffee and cacao plantations, which is likely the context of this report from Madagascar. No additional information is known as to whether this species has persisted in Madagascar.