80 AROIDEANA, Vol. 24 A Revision of Alloschemone Schott (Araceae:Aionstereae) J. Bogner Botanischer Garten Miinchen Menzinger Strde 63 D-80638, Miinchen, Germany P. C. Boyce Herbarium, Royal Botanic Gardens, Kew Richmond, Surrey, TW9 3AB, u.K. C. M. Sakuragui Departamento de Biologia Universidade de Maringa Av. Colombo, 5790 CEP 87020-000 Maringa PR, Brazil ABSTRACT premnum Schott, Monstera Adans., Rha­ phidophora Hassk., Rhodospatha Poepp., The genus Alloschemone Schott is re­ Scindapsus Schott and Stenospermation vised. One new species, A. inopinata Schott. The genera are mostly separated Bogner & P. C. Boyce, is recognized, and on gynoecial characters, notably locule an expanded description of A. occidentalis number, ovule (and thus seed) number, (Poepp.) Eng!. & K. Krause is presented. placenta position, seed shape and endo­ The distinguishing characters of the genus sperm presence or absence. Due to their are discussed, especially with regard to the great plasticity, vegetative characters alone genus Scindapsus Schott, from which Al­ are of little use for generic determination. loschemone has been considered insepa­ In the past some botanists have sought to rable. Both genera have unilocular ovaries reduce several genera into one (e.g. Bak­ with basal placentation with one ovule; Al­ huizen van den Brink 1958; Hotta 1970). loschemone differs from Scindapsus by a However, recent work on vegetative anat­ pinnatifid leaf lamina (entire in Scindap­ omy (e.g. French & Tomlinson, 1981) and sus), shoot architecture, anatomical differ­ floral anatomy (e.g. Carvell, 1989) has ences, and Neotropical distribution (Scin­ shown that these 'weak' genera are prob­ dapsus Paleotropical). Our new species ably soundly based. has fused filaments, a unique character in Previously Alloschemone has been treat­ Monstereae, and further differs from A. oc­ ed as part of Scindapsus (e.g. Madison, cidentalis by the narrower leaf pinnate 1976) due to shared gynoecial characters. separated by an oblong sinus. Seeds, often However, Madison (1979), French & Tom­ diagnostically very useful in Monstereae, linson (1981) and Carvell (1989) have are still unknown in Alloschemone and are demonstrated convincingly that Allosche­ highly desirable. mone is not only distinct from Scindapsus but also probably not closely related. INTRODUCTION IDSTORY Generic delimitation within Araceae tribe Monstereae has long been subject to Schott (1858) established Alloschemone differing interpretations. Mayo et al. (1997) (as Alloschemone poeppigiana Schott, recognize eight genera for the tribe: Allos­ nom. illeg.) for Scindapsus occidentalis chemone Schott, Amydrium Schott, Epi- Poepp. (Poeppig 1845) based on a collec- J. BOGNER, P. C. BOYCE, C. M. SAKURAGUI, 2001 81 tion made by Eduard Poeppig in 1831 at cluding Alloschemone. Unfortunately the Ega (modern Tefe) in Brazilian Amazonas. distribution of species with vestigial tepals Poeppig produced a good description of is, in this instance, of little taxonomic sig­ his new species, which is fortunate since nificance. the single spadix collected was lost. In Alloschemone has been seldom collect­ separating Alloschemone from Scindapsus, ed. Following Poeppig's original gathering Schott (1858, 1860) cited the disjunct dis­ it was recollected by B. A. Krukoff in 1934 tribution of Poeppig's species (Scindapsus and then by M. Madison in 1978. The same is wholly Paleotropical) and the divided year it was also found by staff from the leaf lamina (Scindapsus leaf laminae are Instituto Nacional de Pesquisas de Ama­ always entire). zonia (INPA) while working on the Projeto Koch (1856) and Bentham & Hooker Flora Epffita; in 1982 T. Plowman made (1883) included Alloschemone in the Neo­ what was then the best collection. In 1977 tropical genus Monstera but this was never sterile specimens were collected by Kirk­ universally followed. Engler and others bride et at. in Rondonia state, Brazil and (Engler, 1879, 1889; Engler & Krause, in 1987 Solomon made the only known 1908) recognized Alloschemone but with collection from Bolivia. Recently a fine the caveat 'genus dubium, imperJecte cog­ collection has been made by C.M. Sakur­ nitum'. Madison (1976) argued that the agui at Tefe. This collection, which in­ leaf shape and disjunct distribution of At­ cludes for the first time pre-adult shoots loschemone were secondary to the gyn­ and inflorescences in spirit, forms the basis oecial characters. Alloschemone and Scin­ for the expanded deSCription presented dapsus have a unilocular ovary with a sin­ below. All collections seen by us originate gle ovule on a basal placenta and Madison from Brazilian Amazonia and Bolivia concluded that the genera should be unit­ (Dept. Pando). ed. Later, based on observations along the Alloschemone has been introduced into Rio Negro in Brazilian Amazonas, Madison cultivation only once, from the 1978 Mad­ reported that the growth habit of Allosche­ ison collection (Madison 6310, see Croat, mone was so different to that of Scindap­ 1986). However, there has been some con­ sus that he subsequently considered Allos­ fusion as to the origin of the plant in cul­ chemone to be a distinct genus (Madison, tivation under this number and it seems 1979). that it may have become mixed. A speci­ French and Tomlinson (1981) surveyed men of this Madison collection grown by vascular patterns in the stems of Monster­ one of us (J.B.) flowered after several oideae. They found that in Alloschemone years in cultivation and proved to be Mon­ the course of the vascular bundles in the stera spruceana (Schott) Eng!. stem central core differed from that of all other vining Monstereae while Scindapsus REIATIONSIllPS showed similarities to Monstera, Rhaphi­ dophora, Epipremnum and A mydrium. The maintenance of Alloschemone as a Carvell (1989) showed the floral vascu­ genus is well supported, but its generic re­ lature of Alloschemone to be substantially lationships are less clear. Gynoecial char­ different to that of Scindapsus but similar acters are important for the delimitation of to that of Monstera deliciosa Liebm. Allos­ genera in Monstereae. Alloschemone and chemone also shares some features in its Scindapsus, the only genera with a single vasculature typical of Rhodospatha and basal ovule in a unilocular ovary, would Stenospermation. Much of Carvell's work appear to be closely allied. However, oth­ involved studying vestigial tepals at the er characters do not support this assump­ base on the flowers. Monstereae has na­ tion. Alloschemone and Scindapsus differ ked flowers (i.e. lacking a perigon), al­ in features of vegetative and floral anato­ though Carvell found that many species my, shoot architecture, leaf shape and studied have traces of vestigial tepals, in- geographical distribution. Gynoecial char- 82 AROIDEANA, Vol. 24 acters aside, Alloschemone is most similar arrangement of characters is unknown to the Neotropical genera of Monstereae. elsewhere in Monsteroideae. Alloschemone has a chromosome number After R. Keating's unpublished anatom­ of 2n = 84 and would be hexaploid if a ical data the following leaf data can be basic number of x = 14 is considered. summarized here (with permission). Two Neotropical genera, Stenospermation Leaf lamina: surface with a smooth, thin and Rhodospatha, have 2n = 28, Rhodos­ cuticle, epidermis cells undulate adaxially, patha also 2n = 56, while Monstera and but polygonal abaxially, with straight an­ the other genera of the Monstereae have a ticlinal walls at the cell base. Stomata ab­ certain number of 2n = 60; some other axial, level with the surface and brachy­ numbers have been published but must be paracytic to brachyparahexacytic; subsidi­ confirmed. The chromosome number of ary cells narrow and rounded. Palisade 2n = 84 for Alloschemone suggests a clos­ cells of the mesophyll short, with 6-8 er relationship to Rhodospatha and Sten­ spongy layers, air cavities accounting for ospermation. All counts cited are from up to 600Al of this layer. There are small work by Dr Gitte Petersen (Petersen 1989). cavities at the palisade/spongy boundary and larger cavities regularly spaced within VEGETATIVE ANATOMY the spongy layer. Collenchyma weakly de­ veloped abaxially at the mid-rib, with thin French & Tomlinson (1981) investigated wall layering on epidermal and subepi­ the vascular patterns in stems of subfamily dermal walls. Mid-rib with one to two vas­ Monsteroideae. Their findings mostly sup­ cular bundles. The xylem has small pro­ port the infrageneric classification pro­ toxylem cells and two to five medium­ posed by Engler & Krause (1908) and also sized metaxylem cells. The phloem is a Mayo et al. (1997). French & Tomlinson small irregular strand. Sclerenchyma fibers defined three groups; 'Group l' and '2' to­ ensheathing larger vascular bundles. Tri­ gether equating to Monstereae, 'Group 3' chosclereids are present in small clusters to tribe Spathiphylleae. Alloschemone has in air cavities just beneath the pallisade vascular patterns not fitting one of these layer, larger trichosclereids (up to 40-50 groups. j.Lm) in bigger air cavities deeper in the 'Group I' has simple vascular bundles spongy layer. Tannin cells with dark con­ with the axial bundles formed by the ag­ gregation of small traces branching from tents are scattered and very common in existing bundles. Bud traces are promi­ the mesophyll. Raphide cells are large, nent, enter the central cylinder at the thin-walled and elongate, filled with mul­ node, are superficially attached and fre­ tiple crystals. Druses are very