Introduction

INTRODUCTION

VEGETATIVE GROWTH AND (see below) to detail the shoot architectural patterns in MORPHOLOGY the family. Since then, the complex architectural habitsof this The Araceae occupy a relatively isolated position among family have been given improved descriptive tools by ¯owering plants due to their possession of a number of Ray (1986, 1987a,b,c). Ray (1987c) engaged the com- distinguishing vegetative and reproductive characters plexity of aroid shoot diversity and produced a scheme (MBB). Within the family a number of distinctive char- that allows structural comparisons across taxa support- acter states occur, especially in the larger or more basal ing evolutionary and systematic analyses. Beginning with genera, but few are diagnostic in themselves. It remains the whole shoot, the article (axis) is produced by the best to combine as many characters as possible when activity of a single meristem from origin to termination, making determinations. Particularly in the more derived abortion, or onset of sexuality. The shoot is a single genera, anatomical charactersare diminishedin variety (unbranched) article. The segment is an internode plus and must be combined with other morphological and associated leaf (or leaves) and bud. reproductive charactersfor thispurpose. In aroids, segments can be monopodial or branching, Growth form and habitat variety exceedsthat of any the latter being a junction between two articles. Ray other monocot family. The range includesvinesor large (1987b) investigated shoot organization, leaf types and herbs in tropical forests and borders, and plants with growth correlations of the segment. The two shoot types rhizomes or corms in seasonal dry or cold habitats (Bogner found in the Araceae are (1) monopodial, in which the 1987). Hemiepiphytesand epiphytesare not common but base of the petiole ensheaths the stem, and (2) sympodial occur in thirteen tropical genera (Benzing 1990). A useful where the shoot is continued after a foliage leaf by a bibliography, including anatomical citationsfor epiphytic sylleptic axillary bud; the shoot apex terminates by diff- species, can be found in Watson et al. (1987). Habitat ele- erentiating into an in¯orescence, and the petiole base does vation rangesfrom sealevel to greater than 3±4000 m. not encircle the stem. Plants with elongated stems are res- For a predominantly terrestrial family, aquatic adapta- tricted to taxa with predominantly monopodial segments. tions of several types are widespread. Steenis (1981, 1987) Ray examined the allometry among the componentsof and Cook (1999) have listed numerous observations on the segments and attempted correlations between parts. the family; together the lists include 29 genera distrib- His conclusions yielded no immediate systematic applic- uted among at least six subfamilies (including the four ation but have potential to contribute to understanding lemnoid genera), making 26% of the genera having the evolution of growth formsin the family. aquatic species. Most aroid aquatics are rooted plants Ray (1987c) studied shoot organizational diversity of with emergent stems and/or leaves. Some are submerged 83 species representing 27 genera and 20 tribes, plus (Cryptocoryne, Jasarum), or are rheophytes( Aridarum, Acorus. He noted that ¯owering isalwaysterminal in Bucephalandra, Piptospatha). Some prefer brackish aroids, and that sympodial and monopodial axes are dis- water (Aglaodorum and some Cryptocoryne). They may tinguished by whether or not the shoot is renewed after form rosettes (Anubias, Cryptocoryne, Jasarum), be free- ¯owering. After reviewing the early German and more ¯oating (Pistia) or planktonic (lemnoids). Cook pro- recent literature, Ray codi®ed terminology and revised posed that these adaptations have arisen independently Engler's diagrams that use a series of symbols to represent in the family many times. segment parts. They are useful in correlating structure Meusel (1951) reviewed the work by Engler (1877, and with phylogeny and systematics. His architectural analysis see Engler 1990 with translation by Ray and Renner) on of an unbranched system ®ts Chamberlain's Model in the the growth formsof aroidsand noted that sympodial system of Halle et al. (1978). Ray and Renner (in Engler growth isthe principal mode of extensionin genera with 1990) noted that leaf divergence (phyllotaxy) isgreatly elongated internodes, as illustrated by eight examples. in¯uenced by shoot diameter. Small shoots have a leaf Holttum (1955) supported these observations and called divergence approaching 1 : 2; larger diametersproduce a attention to the precision (exact number of leaves) of spiral arrangement. branch architecture in the climbing genera. Engler Leaf typeswithin the Araceae vary considerably devised the symbol and line diagrams later used by Ray (cf. Hay and Mabberly 1991) but were divided by 2 Introduction

Ray (1987b) into four classes: those associated with (1) Hydrocharitaceae. However, ascurrently understood,it initiation of new shoots; (2) continuation axes; (3) resting would appear that she was describing the maturation of axes; and (4) termination and replacement of axes. The cells divided within the procambium. All tissue matures terms leaf, prophyll, mesophyll, bracteole, mesobracteole, at the xylem±phloem boundary asoriginally blocked out blastophyll, and cataphyll are used conventionally. When by the procambium; therefore, thisinterpretation of combined with modi®ers, such as foliar, ¯agellar, pro- the observation may not be credible. It is interesting that leptic, sylleptic, etc, the terms are de®ned and used in a the pattern of maturation she noted was also found in the much more precise way than before. Ray has provided apparently neighbouring Alismatales, which share a tools by which an evolutionary survey of shoot archi- similarvascularbundleconstruction(cf.Tomlinson1982). tecture of the family could be attempted. REPRODUCTIVE MORPHOLOGY Germination and growth The family is strikingly set off from other ¯owering plants The Araceae are found to produce a larger variety of by its signature feature, the in¯orescence. A terminal seedlingformsthanmostmonocotfamilies,variationwell- branch, the spadix, is subtended by a single leaf-like illustrated by Tillich (1985, 1995). In general, seedlings organ, the spathe. The spadix may be covered with closely are said to be endospermless, with reduced hypocotyl, packed bisexual ¯owers, or the plants may be monoecious having weak development of the cotyledonary sheath and with gynoecial ¯owers crowding the base of the spadix primary root. While Acorus seems easily excluded from the and androecial ¯owers formed distal to them. Distally, a Araceae, it appears we are not at all close to systematic sterile appendix may occur which may be variously col- use of seedling variation within the family. For that we oured. The spathe may be simple and re¯exed, or form an will need comparable stages and morphometric data on erect sheath, variously elaborated into a tube, and often relative expansion of mass of embryonic organs. Much of variously pigmented (MBB). the apparently marked variety in germinating embryosis probably due to heterochrony (cf. Gould 1977: 209 ff) In¯orescences and ¯owers and the variation might be organized using an approach that addresses this possibility. Flowers of the family are usually small and prism-shaped Shoot ontogeny of Calla wasstudiedby Scribailo and due to their packing geometry and usual unextended or Tomlinson (1992). The shoot apex was investigated by sessile form. Because of the inclusion of numerous crys- Hacciusand Lakshmanan(1966) in Pistia and Lemna. talsand/or mucilaginousmaterials,microtechnique Jaeger (1968) reported on vegetative growth in Remusatia. requiresunusualcare and patience. Therefore, detailed A study of meristematic activity of internodal meristems anatomical knowledge hasbeen slowin coming. Eyde in monocots(Fisherand French 1976, 1978) included et al. (1967) made the ®rst survey of ¯oral anatomy since several genera of Araceae. Both localized intercalary meri- Engler's period, where they described anatomy and vas- stems (interrupted) and uninterrupted internodal mer- culature of 18 genera. A series of detailed studies in ¯oral istems are found in the family. Anthurium (three species) vasculature were made by Barabe (1982, Symplocarpus; and Zantedeschia (one species) have uninterrupted meris- 1987, neoteny), Barabe and Chre tien (1985a, Spathicarpa; tems on reproductive axes as detected using surface 1985b, Monstera; 1986, Spathiphyllum), Barabe , Chre tien marking data. Aglaonema (one species) and Pistia stra- and Forget (1986, Anchomanes; 1987, gynoecia), Barabe tiotes have interrupted meristems on reproductive and and Forget (1987, Calloideae; 1988a, Zamioculcas; 1988b, vegetative axes, respectively. Spathiphyllum (three species) Aglaonema; 1992, Culcasia), Barabe , Forget and Chre tien has uninterrupted and interrupted meristems. No system- (1986, 1987, Symplocarpus), Barabe and Labrecque (1983, atic value can be imputed asyet. Calla; 1984, Lysichiton; 1985, Orontium), and Barabe , An intrafascicular cambium (as seen in TS) has been Labrecque and Chre tien (1984, Anthurium). Carvell's reported within vascular bundles of some Araceae (1989) large thesis also considered in detail ¯oral anatomy (Nicolas1916; Arber 1922a). Asdescribedby Arber, in of numerousgenera, emphasizingEngler'sPothoideae immature vascular bundles of Calla palustris and Arum and Monsteroideae. Its numerous detailed drawings of italicum, a cambium can be seen producing short ®les ¯oralsectionsofthefamilyandresultinganalysisareworth of derivativesthat mature between the protoxylem consulting. Mayo (1986, 1989) investigated ¯oral and and protophloem points. She considered the small in¯orescence anatomy of Philodendron. resulting cells to be secondary xylem. Arber also reported Mercado-Noriel and Mercado (1977±8) studied `radial seriation' in Alismataceae, Aponogetonaceae, and the ¯oral and seed anatomy of Pistia. In¯orescence Reproductive Morphology 3 morphology and ¯ower development in Pistia were `mess and spoil' pollinators, but are probably a compon- investigated by Buzgo (1994). The in¯orescence and ent of a highly co-evolved system that involves special- ¯owersare highly reduced in this¯oating rosetteplant. ized morphology, physiology and behaviour of both The spathe and spadix show fusion, congenital at the beetlesandbeetle-pollinatedplants(Young1986).Pollina- base and postgenital at the apex. Buzgo believed the tion phenomena have also been reviewed by Grayum genus to be closest to Cryptocoryninae and Pinellia. (1986b, 1990) and Bay (1995). Croat's(1998) review Several other authors have made useful observations on updated the effortsin thisactive ®eld. the in¯orescence (Engler 1884; Manya-Chernej 1978; Chernej 1977; Barahona Carvajal 1977). French (1986b) made the ®rst comprehensive investiga- Fruit tion of stamen vasculature involving observations Pijl (1969) discussed fruit dispersal in the family and Spjut on 100 genera and 350 spp of Araceae. Stamen traces (1994) provided a systematic treatment of fruit types. In vary in their capacity for branching and anastomosing, Spjut'sthorough review, good illustrationsare given for while synandrium construction complicates vascular each type, including the bacca (berry): Acorus calamus; organization. His ®ndings are systematically arranged. catoclesium (compound fruit of indehiscent fruitlets, French (1985a,b) also studied the endothecium in a total enclosed by leaves, bracts or fused perianth parts): Typho- of 59 genera. Manning (1996) noted that the Araceae nodorum; foraminicidal capsule (capsular fruit opening show variation on the annular±helical endothecium by irregular diverging cracks or slits): Alocasia macro- pattern. Acorus with itspalmate-baseplateendothecium rrhiza, Lasia spinosa; sorosus (compound fruit of many is distinct. Sampled plants of the Alismati¯orae show succulent pericarpia that developed on a peduncle): more diversity with their U-shaped, helical and baseplate Symplocarpus foetidus. types. Little is known of sex expression variables within the family. Lovett-Doust et al. (1986) studied fecundity and Pollen and pollination size relationships in Arisaema triphyllum, a gender-labile woodland herb. In that species, sex expression correlates Beginning with Erdtman's(1952) compendium, a number with the abundance of stored resources in the under- of observations on araceous pollen have been produced; ground corm. The plants are obligate outcrossers, having see especially Thanikaimoni (1969) and Grayum (1984, a self-incompatibility system, therefore neighbouring 1992) for comprehensive surveys using light and electron male ramets will not affect the pollination success. The microscopy, respectively. Also, Grayum (1985) has larger the corm, the smaller the percentage of successfully reviewed pollen starch and pollen nuclear number in the matured ovules, suggesting nutritional competition family (Grayum 1986a). See Kuprianova and Tarasevich between corm and ovules. Several other signi®cant cor- (1984) and Tarasevich (1990) for pollen surveys of relations are discussed. Fruit and seed structure are also Lemnaceae. discussed by Kulkarni et al. (1990) and by Bochenska and Grayum (1986b)notedthatpollen morphologyishighly Kozlowski (1974). Grubert (1981) included Araceae in a variable in Araceae. Sculpturing ismore variable than survey of angiosperms for mucilage. vegetative morphology for advanced genera. Grayum offered correlationsbetween sculpturingand weight with different families of insect pollinators, and concluded that Seeds and embryos sculpturingisprobablymostcloselycorrelatedwithpollin- ator type. Unusual pollination syndromes of Araceae Seubert (1993) made a comprehensive survey of seed involve in¯orescences acting as phototropic attractants, structure of 92 genera. Her work included a description insect traps, the formation of nectar-like substances of seed and some embryonic tissues with illustrations. through degeneration of cell contentsof stigmatic The Araceae are notable for presenting a large amount of papillae, and the generation of variousodoursthat act total monocot seed variation. She produced comparisons asbeetle or ¯y attractants,asreviewed by Faegri and of her data against four prominent classi®cations. Among Pijl (1971). Young (1986) made observations on beetle the conclusions, she noted the genus Gymnostachys is pollination (cantharophily) in Dieffenbachia longispatha, very isolated and most resembles Typhales rather than supporting arguments that beetle pollination is an early Araceae. Also Lemnaceae should clearly be a tribe within syndrome in angiosperms. Young noted the work of the Araceae. Later, Seubert (1997b) studied seeds of the Croat (1983) that included observations on ¯oral biology 10 genera and 18 species of the tribe Lasieae; these seeds of Dieffenbachia. Beetle visitors have been thought to be are distinguished from those of all other aroids by being 4 Introduction kidney- or horseshoe-shaped, with a warty or smooth Pandaniidites (Elsik 1968) and which resembles pollen of surface, and with a curved or straight embryo. extant Lemna (see Hotton et al. 1994). Because of the variation, morphoclines of these data The cladistic analysis by Stockey et al. (1997), based must be interpreted carefully since the trends are not on 34 vegetative and reproductive structures, nested synchronized or unidirectional. Takhtajan (1985) Limnobiophyllum in a position between Pistia and the described the seed anatomy of many genera. four lemnoid genera in a pectinally branched tree. This Grayum (1991) surveyed the systematic embryology of suggests an unexpected closeness of Pistia to the Lem- the family. Johri et al. (1992) reviewed the systematic naceae. However, the cpDNA analysis separated Pistia embryology of the family, asdid Seubert (1993), who from the lemnoid genera; they are paraphyletic rather demonstrated a close relationship between the Araceae than adjacent in the same clade. Further, if Pistia and the and Lemnaceae; the authorsprovided a comparisonwith lemnoidsareindependentlyembeddedwithintheAraceae, the Helobieae. the family Lemnaceae cannot be recognized without making the Araceae paraphyletic. I would note that the morphological character list used by Stockey et al. (1997) PALAEOBOTANICAL RECORD is an excellent structural list and the character polarities are most probably interpreted correctly. However, these The most recent review of the aroid fossil record is found reduction trendsare potentially homoplasiouseventsof in MBB and additional reviewsare given in Gregor and such character that homologies are very dif®cult or Bogner (1984, 1989) and Grayum (1990). The following impossible to verify. All evidence at this time favours the summary will emphasize several interesting aspects of the position of including these reduced ¯oating lemnoid history of vegetative and reproductive materials. genera within the Araceae. On the basis of cpDNA sequences in plants, dicot± The leaf record wasreviewed by Daghlian (1981) and monocot divergence hasbeen placed at 200 mya Æ 40 my, by Dilcher and Daghlian (1977). The family'srecord a suggestion that is also supported by an analysis of dating from the Paleocene isgood. Specimensknown as nuclear genesencoding 26S and 18S ribosomalRNAs Nityophyllites from the Paleocene of Kazakhstan consist (Li et al. 1993). This would suggest emergence of the of fragments that seem araceous based on venation and angiosperm lineage in Triassic±Jurassic time, which con- cuticular features. siderably predates its appearance in the fossil record. One of the interesting disjunctions among extant Much of what we wish to understand of the genesis of this aroidsisthat of Typhonodorum of Madagascar and the distinct family must have occurred within this early apparently related Peltandra of eastern North America. period. No pre-Cenozoic aroid remainsare known and This disjunction can only be understood if one or both no evidence isavailable from the period when diversi- genera were widely dispersed at one time, as now appears ®cation of subfamilies would have taken place. Grayum likely based on the following evidence. In 1977, Dilcher (1990) also noted the paucity of early aroid records. and Daghlian described a large well-preserved leaf However, by early Tertiary, macrofossils are relatively from the Middle Eocene of Tennessee which they widely dispersed (Daghlian 1981). named Philodendron limnestes and placed in section In the subfamily Aroideae we have an example of fossil Meconostigma, with which it seemed to agree in venation intermediates that shed some light on generic evolution. and cuticular features. Later, Mayo (1991) suggested that Since 1821, the small, ¯oating, aquatic lemnoid genera the specimen would be better placed near Typhonodorum, have usually been segregated as the family Lemnaceae. tribe Peltandreae. Hickey (1977: 114) described Peltandra However, the cpDNA results of French et al. (1995) nested primaeva, based on venation, from early Eocene rocks of Lemna clearly within the subfamily Aroideae; in partial North Dakota. Therefore, it appearsquite likely that con®rmation of these results are studies of Limnobio- these two genera once had congeneric ranges. Wilde phyllum expansum, originally described by KvacÏ ek (1995) (1989) described cuticular features of three araceous from the Miocene of Europe. Stockey et al. (1997) recov- leavesfrom the Middle Eocene of Germany. ered more than 200 specimens of L. scutatum (Dawson) Krassilov from the Paleocene of Alberta, specimens that include the ®rst known ¯owering materials. In size the Reproductive materials plantsare transitionalbetween modern lemnoidsand Sahnipushpam shuklai, a possible spadix from Eocene Pistia. This study also described the ®rst fossil aralean Deccan Intertrappean of India, hasuncertain af®nities pollen, a monoporate, spheroidal grain with echinate and may not be araceous. Berry (1930) described Acorus spines. It is similar to grains previously described as heeri, now the spadix form genus Acorites (Crepet 1978), Biogeography 5 occurring in the Middle Eocene of Tennessee. Araceites most subfamilies, making the cytological history of the hungaricus Rusky is an in¯orescence that can be included family complex and not well understood. in the Araceae (Crepet 1978). Bogner's(1976) reexamin- ation of Acoropsis Conwentz showed that Acoropsis eximia of the Eocene Baltic Amber isan infructescence BIOGEOGRAPHY that should be assigned to Monstereae. Permineralized seeds and fruits of the Princeton chert For a large, overwhelmingly tropical family, the nearly (Middle Eocene) Allenby formation were assigned to cosmopolitan presence of Araceae is unusual. The distrib- Lasioideae by Cevallos-Ferriz and Stockey (1988) and ution of genera, subfamilies, and tribes has been treated Cevallos-Ferrizetal.(1991).TheirspecimenKeratosperma by Willis(1949), Croat (1979), Grayum (1984, 1990), allenbyensis Cev. & Stoc. gen. et sp. nov. appears closely Bogner (1987) and MBB, and noted with generic treat- related to Cyrtosperma which constitutes evidence of mentsin thisvolume. subtropical elements in that ¯ora. These are the oldest Most of the nine subfamilies are widely dispersed in known fruit and seed remains of Lasioideae. The authors' the tropics and southern hemisphere including Pothoideae table 2 provides a useful list of reproductive characters of (including Monstereae), Lasioideae, Philodendroideae, Monsteroideae/Lasieae from several sources. They review Schismatoglottidoideae, and Aroideae. The Gymno- reports on Oligocene seeds of Monstereae, Lasieae, stachydoideae (Gymnostachys) are eastern Australian, the Pothoideae, and Pistioideae, and a report by Bown et al. Orontioideae (three genera) and Calloideae (Calla) are (1982) on fossil fruit from the Egyptian Oligocene. northern. The Lemnoideae are cosmopolitan aquatics. Madison and Tiffney (1976) found fruits and seeds in At the generic level, the majority of tropical rangesare the Oligocene representing Monstereae, Pothoideae and restricted to major ¯oristic regions, rather than cosmo- Lasioideae. These tribes are also recorded in Miocene and politan. Figuresgiven for tropical genera are Asiantro- Pliocene sediments in a variety of Laurasian localities. pics: 44 genera; Neotropics: 36 genera; Africa: 20 genera. In reviewingrecords offossil aroid pollen, Muller (1981) Neotropical genera have a higher percentage of endemic noted a report of a Spathiphyllum type from the Upper ranges and signi®cantly more species. Miocene, and pollen similar to Ambrosina, Amorpho- The distribution and subfamilial structure of the family phallus, Arisarum, Holochlamys, Pseudodracontium and con®rmsitsgreat age and the point of origin, or pointsof Thomsonia. Graham (1976) reported a Spathiphyllum origin, of most subfamilies has not been solved. In their type in the Upper Miocene of Mexico asdid Leopold analysis of ¯oristic origins, Raven and Axelrod (1974) (1969) from the Miocene of Palau (W. Paci®c). At least considered the Araceae to be `West Gondwanalandic- four modern subfamilies were present by Oligocene Laurasian'. However, the possibility of an east times. Gondwanan, or perhapsLaurasianorigin hasnot been ruled out and will be discussed below. An analysis attempting to explain the family's CHROMOSOMES contemporary distribution, as well as the fossil record, hasbeen provided by Grayum (1990) and will be brie¯y The family was ®rst widely surveyed for chromosome summarized here. The distribution of Engler's Lasieae countsby Marchant (1970, 1971a,b, 1972, 1973), who and Monstereae strongly suggests that the family was accounted for the existing literature and added many new well-established in west Gondwanaland before the sep- counts. Petersen (1989) wrote a comprehensive review for aration of Africa and South America. However, Grayum the family and evaluated all published counts; new pointed out that this inference does not establish west numbers were later added (Petersen 1993a,b). The most Gondwanaland asa point of origin for the family, and he updated report with all dubiouscountsappropriately proposed that the family originated in temperate or sub- marked wasprovided in MBB with new information tropical Laurasia. Later, it reached west Gondwanaland supplied by Petersen; counts are missing for only four via Eurasia, perhaps in late Cretaceous whence it under- genera. These counts are included in this volume. went a major radiation. In support of this concept, he Palaeopolyploidy isrampant in the family and the noted that the four primitive (long-branch) taxa Calla, original basic number is rare or non-existent, perhaps Lysichiton, Orontium and Symplocarpus are north tem- occurring in Ulearum or Filarum (n ˆ 7). The presumed perate, and that there are no south temperate genera, basic (x) numbersaslistedfor the genera in MBB are unspecialized or not. 10±21 (continuous series), and 27. One or more genera By this scenario, elements of this ¯ora migrated sec- with an extensive aneuploid or polyploid series occur in ondarily into Eurasia and some perhaps reached North 6 Introduction

America (Typhonodorum, Peltandra, and perhaps classi®cation in the family. Allergenes capable of causing Orontium) via a north Atlantic route, thishaving been contact dermatitis are present in some Philodendron open prior to the Eocene (Raven and Axelrod 1974; Wolfe species. Proanthocyanidins (condensed tannins and their 1975). Con®nement of Peltandra and Orontium to eastern precursors) are found in about half of the genera sampled. North America isperhapsdue to the barrier presentedby Essential oils, abundant in the outgroup, Acoraceae, are the north±south epicontinental sea in the great plains only known from two species of Homalomena within the region that lasted up to the Paleocene. Araceae. Cyanogenic glucosides, especially prussic acid, The®nalelementofthisproposalisthatTethyandistrib- are found in many genera. They can produce enough ution of certain Areae (Engler'sArinae) may re¯ect a HCN to be toxic. Finally, carbohydratesare abundant Laurasian origin of the group. These taxa may have and are found aseither mucilaginousglucomannans,or undergone a radiation on the northern margin of that sea starch, or both. extending from western Europe to Indo-Malaysia. Also Laticifer contentsare chemically complex, asusualfor included were the genera Ambrosinia, Ariopsis, Arisaema, this cell type, but they show systematically useful variation Arisarum, Pinellia, Pistia and perhapsThomsonieae.As in the family that should be pursued (Fox and French evidence, Grayum (1990) put forward the poor repres- 1988). New World `colocasioids' Xanthosoma and entation of these groups in south-east Asia, Africa, and Syngonium (two species each) have latex sterols which their absence in the Neotropics. He hypothesized that seem to vary systematically by genus. The Old World Arisaema only recently reached North America from genera Colocasia and Alocasia (two species each), contain Eurasia. sterol esters, with no sterols detected (Fox and French However, Hay (1992b) disagreed with the concept of a 1988). The authors suggested that the antiherbivory Laurasian origin for the Areae and he endorsed Riedl's hypothesis needs to be investigated regarding the role of (1980) suggestion of a Gondwanan origin for the tribe. sterols and related compounds. Hay described Lazarum, new for the Areae, and consi- A study of leaf ¯avonoids and anthocyanins of 61 dered it to be an autochthonouselement of the Australian genera of Araceae wasmade by Williams et al. (1981). ¯ora. Later, French et al. (1995) contributed an import- Other studies of chloroplasts or their ontogeny include ant piece of evidence supporting the idea of an east those of Guillot-Salomon et al. (1978), Chow et al. (1988), Gondwanan origin for the Areae. The six genera of Areae Stewart and Dermen (1979) and Buinova (1988). Bate- sampled for their cpDNA study are arranged topo- Smith (1968) found representation of a wide list of logically in a strict pectinal relationship, which, from phenolic compoundsshowingno obviousrelation to eco- base of the clade to distal-most are arranged as follows: logy or systematics in the family. Later, an analysis of the Typhonium (south-east Asia, Australia), Theriophonum systematic distribution of tannins in angiosperm leaves (India), Sauromatum (Africa, India), Biarum (Medi- was accomplished by Mole (1993). However, the sys- terranean), Arum (Mediterranean), and Helicodiceros tematic versus ecological signi®cance of tannin presence (Mediterranean). Lazarum is considered most closely hasnot been resolved. related to Sauromatum. ThissupportsHay's(1992b)idea The phenolic constituents of cell walls when studied by that the tribe is southern in origin. He suggested that the ¯uorescence microscopy (UV) form two major groups progenitors of the northern genera, already seasonally within the monocots(Harrisand Hartley 1980), and the adapted, were rafted northward on the Indian plate. The authors compared the results with Cronquist's (1968) evidence certainly makesit lesslikelythat the tribe is classi®cation. The Araceae (in group 2, three genera northern in origin. Li (1986) also argued for a south sampled) tested negative for the four wall extracts exam- Asian (southern Laurasian) origin for the family. ined in this study, as did most of the Alismatidae. Acorus differedfromtheAraceaeintestingpositiveforp-hydroxy- benzoic acid. Groupsreacting positively(group 1) CHEMISTRY included most of the Commelinidae and many Liliidae. Also, Harborne (1974), surveying ¯avonoids, found four Chemical compoundsin the family were surveyedby monocot groupings. The one group having both ¯avonols Hegnauer (1963, 1986) and brie¯y reviewed in Hegnauer and glycosyl-¯avones comprised Araceae, Lemnaceae, (1987). The family isquite active in the production of and Commelinaceae. secondary compounds. Among those widely detectable Dring et al. (1995) surveyed chemistry in 55 genera throughout the family are alkaloids and amines, saponins, from all aroid subfamilies. They stated that the family's cinnamic acids, and ¯avonoids. The latter are diverse chemistry is highly diverse and only a small part has enough to make some contribution to understanding been investigated; they studied alkyl resorcinols and Aroids and Humans 7 polyhydroxyalkaloids.Thevarioustypesreportedshowed are very valuable carbohydrate sources in many tropical some correlation with classi®cation. In particular, the nationsand severaltaxa have been carried far from their close relationship of Anchomanes to Nephthytis and originalranges.Thetwomostimportantspecies,Colocasia Pseudohydrosme was strongly supported by the alkaloid esculenta (taro of south-east Asia), and Xanthosoma sag- evidence. Schmid et al. (1997) made a survey of seed lipids ittifolium (South American yautia or cocoyam), are the of 54 species representing 20 genera of the family. Using source of ancient cultivars. They are well-established GC mass spectrometry, they reported ®nding 13-phenyl- staple crops throughout the tropics. More locally, tridecanoic acid (13-PTDA) in the seeds of seven genera some species of Alocasia, Amorphophallus paeoniifolius of aroids, a ®rst report from any plant family of a lipid (campanulatus) (elephant yam), and Cyrtosperma with a substituted aromatic phenyl ring. All genera with chamissonis (giant taro) are also cultivated for starchy thiscompound are in subfamilyAroideae ( sens. str.). tubers. Monstera deliciosa (ceriman), often grown asan ornamental, isprized in variousplacesfor itsedible fruit, said to taste like pineapple. It is eaten only fully ripe to Thermogenesis avoid the raphides and other irritants. The seeds of A metabolic process that generates heat in in¯orescences Typhonodorum (Madagascar) and Montrichardia (South has been observed and studied for two centuries (Bay America) also provide food. 1995). At least seven apparently unrelated thermogenic Various lemnoid species are used as high-protein plant familiesare known but genera of Araceae provide the food for humans and domesticated animals, ®sh, and best studied examples. The process is complex, energy shrimp. They have a very favourable amino acid com- intensive, and apparently allows volatilization of odours position and high productivity, up to 40 t dry weight per that attract pollinators. The family shows at least four hectare (Landolt 1998). In Thailand, Wolf®a globosa is patterns of expression (Leick 1915). These vary from a produced from pond culture asa vegetable for humans generalized heating of the spadix for as long as two weeks (Bhanthumnavin and McGarry (1971), and various at 35 C above ambient in the cold season-blooming species of other genera are being evaluated for larger Symplocarpus, to a process in Arum where four distinct scale production. in¯orescence zones volatilize distinct compounds during Purseglove (1972) and Coursey (1968) have discussed short but intense periods of heating. The increase in the edible aroids. metabolism that accomplishes thermogenesis is triggered Regarding edibility, with very few exceptions, aroid by salicylic acid (Raskin et al. 1989) and made possible partsmustbe cooked or heated before eating. Probably by a very ef®cient electron transport process called the the large majority of species, including the `edible' ones `cyanide resistant pathway' (Meeuse 1975). The resulting listed above, have a powerful irritant in virtually all plant high energy cost to the plant is thought to be justi®ed on parts. It is certainly associated with the ubiquitous raphide the basis of high pollinator speci®city and pollination crystals, which penetrate soft epithelial surfaces of the timing, yielding better genetic ®tness of the seeds (see gastro-intestinal tract. In the case of Xanthosoma, the reviewsby Bay 1995, and in MBB). barbed raphideswill magnify mechanical damage (Sakai et al. 1972). However, Walter and Khanna (1972) argued that proteolytic enzymes(named `dumbcain') work AROIDS AND HUMANS together with raphides. The combined mechanical and chemical damage probably activateskinin-releasing In a review of aroid morphological variation, Bogner mechanisms in the body, triggering the in¯ammatory (1987) also surveyed economic uses. A chapter on the response. Dieffenbachia iscommonly called `dumb cane' topic isincluded in MBB, and the family isnoteworthy since chewing on the plant parts may cause swelling of for itseconomic value. Many of the genera are import- the throat suf®cient to prevent speech and possibly cause ant ornamentals, either as foliage plants, or based on death due to a blocked airway. The attractive seeds their white or colourful in¯orescences. Especially prom- of Arisaema are also notorious in this respect, as are inent ornamentalsare speciesof Aglaonema, Alocasia, the leavesof Pistia stratiotes, `water lettuce' (John Amorphophallus, Anthurium, Caladium, Colocasia, MacDougal, personal communication). Raw araceous Dieffenbachia, Monstera, Philodendron, Spathiphyllum, plant parts should be considered dangerous to humans in Xanthosoma, and Zantedeschia, but many othersare also spite of the fact that uncooked starchy rhizomes are fed collected and grown for display throughout the world. directly to livestock in various parts of the world. Equally important isthe cultivation of severalgenera Plowman (1969) and Bogner (1987) described folk uses for their edible starchy tubers, corms or rhizomes. They of aroids, with examples as follows. In tropical Asia, 8 Introduction

Typhonium blumei isusedasa cure for diarrhoea. Schultes (1977±88). They are noteworthy for their wide generic (1963) stated that a number of native tribes in north- sampling and are reviewed below. western South America reported the use of parts of certain species of Anthurium, Philodendron, and Urospatha as female contraceptives. Other miscellaneous examples Tissues include basket-weaving materials (Heteropsis spruceana Collenchyma roots) and arrow poisons. See also the discussion in The ®rst account of collenchyma strands in the family Bown (2000) and the listing in Duke and Vasquez (1994). wasby Ambronn (1879±81). He reported that collen- chyma development beginsin the epidermal region in Engler'sAroideae and Amorphophallineae and isof REVIEW OF VEGETATIVE subepidermal origin in Colocasia esculenta petioleswith ANATOMICAL LITERATURE stranded collenchyma. He called attention to ring- forming(ˆbanded)collenchymainPhilodendroneximium. This review necessarily omits discussion of numerous con- Ittenbach and Boecker (1997) described the in¯orescence tributions listed in the bibliography. The present inten- axisanatomy of 13 African Amorphophallus species and tion isto provide an overview of diversecontributions commented on the support provided by collenchyma. and some other characteristics of the family. However, the ®rst surveys of this tissue and its trends in the family have only recently been published (GoncËalves General sources et al. 2002; Keating 2000). Numerousanatomical detailsof the family are found Sclerenchyma throughout the monographic treatmentsof the family by Engler, often with Krause (1905, 1908, 1911, 1912, 1915, Sclereids have long been observed in araceous tissues and 1920a,b). Engler frequently used anatomical features in were summarized ®rst by Solereder and Meyer (1928) de®ning major subdivisions of the family. Solereder who listed the genera bearing trichosclereids. These cells, (1919) reviewed the anatomy of about 23 genera of aroids growing among ground tissue, can be branched, knobbed, and Solereder and Meyer (1928) compiled observations or H-shaped, and are recorded at lengths of up to 1500 mm. on the general anatomy of about half the genera of They have only been reported from nine genera in the Araceae, together with a compilation of early literature, subfamily Pothoideae of Grayum (1990) and Keating although the treatment is not particularly systematically (2002b; thisvolume) and have been describedby Bloch oriented. (1946), Rao (1954, 1964, 1977, 1991), Nicolson (1960), The history of systematically oriented research on Singh (1968) and Hinchee (1983); within the nine genera Araceae wasreviewed by Croat (1998). In spiteof his they vary from rare to numerousin one or all organs. stated emphasis, Croat's comprehensive treatment covers Later, Seubert (1997a) provided observations and a history and in¯uential workers, regional studies, as well reviewofsclereidoccurrenceinthefamilyandtheirdistrib- as anatomy, cytology and other disciplines. One of the ution within organs. She reported them present in eleven best general reviews of this or any family is MBB which genera and argued for acceptance of the subfamilies provided a detailed account of essential literature and all Monsteroideae and Pothoideae and perhaps raising aspects of classi®cation, morphology, geography, and Spathiphyllum to that level of recognition. However, note illustrated generic concepts. In an introductory chapter to that the present study only accepts Pothoideae for the 16 MBB, French (1997) reviewed anatomical literature and genera (see discussion, p. 23). Brachysclereids occurring gave a comprehensive, though not illustrated, summary at the central cylinder boundary in Pothoideum were not of the anatomy of all organs and tissues. Grayum (1984) mentioned in Seubert's review. Fibres in roots possessing also reviewed anatomical literature in his thesis empha- isotropic layering were investigated by Mueller and sizing palynology. Other recent surveys include Dahlgren Beckman (1979) (see roots, p. 16). and Clifford (1982), Dahlgren et al. (1985), and an abstract of The genera of Araceae by Mayo et al.in Xylem Kubitzki (1998). Ko et al. (1990) looked at leaf sheaths, petioles, petiolules The most proli®c researcher in systematic anatomy and roots of seven Korean spp of Arisaema, and found of the family hasbeen JamesFrench, whosework, often that all xylem is composed of tracheids. Tracheid walls with collaborators, has covered many vegetative and are variously thickened and show no trend; examples reproductive tissues in a lengthy series of papers from several genera, including Pistia, Colocasia, and Review of Vegetative Anatomical Literature 9

Scindapsus, are described in detail. A survey of xylem for French et al. (1995) implied independent (homoplasious) the presence of vessels has been accomplished by Carlquist origins of Old World and New World anastomosing and Schneider (1997: Acoraceae; 1998: Colocasioideae: types. This begs some interesting evolutionary questions ®ve genera) and Schneider and Carlquist (1998: that have not yet been investigated. Philodendroideae: ®ve genera) as part of a series of studies Va gu jfalvi (1971) recorded alkaloidsaspresentin latex of monocots. All genera studied have vessels in roots, as of ®ve genera of Araceae and Fox and French (1988) previously reported, and most genera have less specialized investigated the contents of laticifers in a larger generic vessels in rhizomes (stems) as well. The absence of pit sample. Typical of laticifers in general, their contents membranes, diagnostic for vessels, was con®rmed by included polyisoprene hydrocarbons among other com- the authorsusingSEM. In the genera Anthurium and pounds, and the authors found considerable variation Zantedeschia (one species each studied), it could not be across the family in latex chemistry, much of which could con®rmed that stem vessels were present. No systematic be systematically useful. Most groups have clear latex meaning isclear but the Araceae have reached a higher but milky latex characterizes Dieffenbachia. New World level of physiological ef®ciency than if only tracheids colocasioids possess sterols and sterol esters, which are were present. also common in dicots.

Phloem Secretory tissue and glands There are very few histological descriptions of phloem in This topic subsumes a variety of different cells and tissues. the family. Shah and James(1971) describedthe ontogeny Tieghem (1872) described secretory tissue in seven aroid of sieve elements and companion cells of stolons in Pistia. genera. He noted that the presence of laticifers and canals Sieve mother cellsproduce sieveelementsand companion with secretory epithelium in the same organ is very uncom- cells. The latter continue to divide, producing up to 20 mon among ¯owering plants. Belin-Depoux (1978) has companion cellsalong a mature sieveelement in up to described small circular glands near vascular bundles two ®les. Sieve elements are short with oblique or trans- occurring on leavesof Alocasia; they consist of super®cial verse end walls having simple sieve plates, and containing columnarepidermalcellsunderlainbyparenchyma.Later, protein bodies. In Dieffenbachia sp., Parthasarathy (1980) Belin-Depoux (1989) described foliar hydathodes or foliar described phloem as being up to 4 years old and little or nectariesin Alocasia. Lindorf (1980) described secretory no P-protein was detected in mature sieve elements. It was glandular cavitiesin the adaxial epidermisof Anthurium noted that sieve tubes in monocot axes appear to function bredemeyeri that were 4±5 timesthe sizeof epidermal throughout organ life span. Behnke (1969a,b,c) described cells. Schnepf (1965a), in a TEM investigation, described phloem contentsin monocotsand Behnke (1972) found the secretion of odorous eccrine type oil by the spadix that all sampled monocot species have exclusively P-type apex of Typhonium divaricatum. plastids in their sieve elements.

Laticifers Cell inclusions Early descriptions of laticifers in leaves were produced by Hanstein (1864), Tre cul (1866, 1866±70), and Scott Starch (1889). Tieghem (1872) described secretory tissue for Reserve photosynthates in the family are quite variable. seven genera. More recently, Pant and Kidwai (1966) While of probable systematic signi®cance, they have described secretory canals and laticifers in the family. not been subject to any broad survey. Starch granules, Articulated laticifers of the anastomosing type are said to certainly the most common reserve, were examined by begin as separate cells which then start lobing. Elongat- Reichert (1913, 1919). He described physical, optical and ing tubular branchesmeet similarbranchesfrom adja- chemical properties, with illustrations of starches of cent cells and `fuse by dissolution of the common walls to seven genera of Araceae. Observations of starch of seven form articulate tubes'. araceousgenera have been recorded from cormsand French (1988) surveyed about 75 genera of Araceae for rootstocks by Allen (1929), Winton and Winton (1935, laticifer presence. He noted that most genera have the non- 1945), and Fujimoto et al. (1990). Sakai and Hayashi anastomosing type. He concluded that the anastomosing (1973) recorded that nine aroid genera store carbo- type characterized the Colocasioideae (except Ariopsis hydrates as sugar rather than starch. The sugar-reserve which has non-anastomosing laticifers), which by implic- genera are scattered throughout four subfamilies and ation were monophyletic. Later, the cpDNA study by Acoraceae. 10 Introduction

Crystals Arisarum, Arum, Dracunculus and Zantedeschia. While Nearly all of the known calcium oxalate crystal con®g- there often appeared to be a close relationship between urations, raphides, druses, prismatics, and crystal sand, protrusions and groups of mitochondria, no function for have been reported in Araceae (Genua and Hillson them has been demonstrated. Such protrusions were 1985), and their presence was con®rmed in this study. frequently encountered in the present study on plastids Among monocot families such diversity is only found in and nuclei. Kisser (1928) recorded pectic warts within Araceae (Prychid and Rudall 2000). In the earliest ana- mesophyll cells of Arum and Dieffenbachia. Also, slime tomical study on the family, Turpin (1836) described a cells have been found suspended from neighbouring cells unique and specialized form of raphide cell, since then into ground tissue air cavities in Dieffenbachia (Kisser referred to asthe biforine. Thisthick-walled, spindle- 1928;Lindorf1980;PotgieterandVanWyk1992).Alarger shaped cell has papillate ends capable of ®ring a stream survey of intercellular pectic strands in leaf palisade of raphidesout at either end asdescribedby Middendorf parenchyma (Carr et al. 1980b) showed them to be present (1968, 1982, 1983). Among systematic accounts of crystals in Arum sanctum. Such strands apparently form the prin- are those of Buscalioni (1895±6: four genera), Solereder cipal connection between palisade cells. As they occur at and Meyer (1928: 16 genera), Genua and Hillson the limit of resolution of light microscopy, they were not (1985:14 genera), and Keating (2002a: family). con®rmed in the present study. Carr et al. (1980a) also Development andchemical analysisofaraceouscrystals studied intercellular strands associated with stomata. have been investigated by Pfeiffer (1925), Wakabayashi Wakabayashi (1957b) reported on mucilage present in (1957a),MollenhauerandLarson(1966),PantandKidwai Amorphophallus. (1966), Ledbetter and Porter (1970), Al-Rais et al. (1971), Rakovan et al. (1973), Sakai and Hanson (1974), Cody Leaves and Horner (1983), Genua and Hillson (1985), and General compilationsof aroid leaf anatomy were ®rst Traquair (1987). Franceschi and Horner (1980) provided made by Tieghem (1866), de la Rue (1866), and Dalitzsch a thorough review of the topic. Where the chemistry has (1886). The early work wassummarizedby Solereder been examined, raphideshave been found to be mono- (1919) who described the general leaf anatomy, stomata, clinic calcium oxalate monohydrate (Whewellite) while crystals, and secretory tissue of 23 genera of aroids. druses are mostly calcium oxalate dihydrate (Weddelite). Solereder and Meyer (1928) compiled observations on Sakai et al. (1972) and Cody and Horner (1983) described the general anatomy of about half the genera of Araceae, deep groovesin raphide surfacesof Colocasia, Alocasia including all types of crystals; the treatment is not sys- and Xanthosoma; such crystals may also have backwardly tematic in organization, nor accompanied by tablesor oriented surface barbs which would certainly increase index. damage to mouth tissues when plants are grazed upon. Investigation into what, in addition to raphides, causes Ontogeny the irritating effectsexperienced when chewing or swal- Leaf ontogeny wasdescribedby Troll (1932b) for lowing aroid tissues was undertaken by Saha and Hussain Eminium, Helicodiceros, and Sauromatum. He described (1983). They studied varieties of Colocasia and Alocasia shield-form leaf ontogeny (1932a) in Alocasia, Ariopsis, and concluded that a non-volatile glycoside of 3, Arisaema, Dracunculus, Orontium, Pothos, Remusatia, 4-dihydroxybenzaldehyde isthe irritant. It isheat-labile; and Syngonium. Troll and Meyer (1955) studied unifacial cooked or dried samples are considerably less acrid. development in 11 genera of aroids. Marginal growth in Philodendron oxycardium leaveswasexamined by Pray Silica (1957), who noted that super®cial initials rarely if ever Netolitzky (1929) listed no Araceae as having siliceous contribute to internal leaf tissue. Kaplan (1970, 1973a,b) bodiesbut Gertz (1916a,b) reported them aspresentin made a series of studies of events occurring during Pothos ventricosus. None was found in the present study. shoot expansion of Acorus, providing observations more detailed than available for any genusof Araceae. Murata Protrusions, pectic materials and warts, (1990a) investigated the morphological development of slime cells, mucilage pedate leavesin sevengenera of Araceae. The chemistry and systematic importance of these materials is not well understood at present and they The phyllode theory should be further investigated. Morassi Bonzi and Fabbri Arber (1918a) elaborated DeCandolle'sphyllode theory (1975) described chloroplast (stromatic) protrusions in of the monocot leaf and proposed that the lamina of Review of Vegetative Anatomical Literature 11 bifacial leaves, such as found in the Araceae, represents ontogenyofseveralgeneraincludingtheparallelodromous the secondary expansion of the petiolar region. Kaplan Pistia (four to ®ve secondaries along one side of a primary (1973b) re®ned thistheory by adding developmental vascular bundle); the basally actinodromous Colocasia observations of monocot dorsiventral leaves including esculenta (peltate lamina, three primariesdiverging from Zantedeschia aethiopica. Asin other monocots,leavesof the base; six to eight secondaries along one side of the Araceae present leaf primordium segmentation into upper primary); and the campto-brochidodromous Scindapsus and lower leaf zones, as opposed to the three zones of aureus (®ve to six secondaries along one side of the pri- dicot primordia. However, in Zantedeschia and other mary). Ertl's (1932) techniques should be revisited. They aroid genera sampled by Kaplan, aroids appear to be have the capacity to produce important trends, given the unique among sampled monocots in having acroplastic morphological variation in the family and our improved maturation after the lower and upper leaf zoneshave been understanding of its classi®cation. initiated. This makes them surprisingly more similar to A number of studies have outlined informative vascular the dicot maturation pattern than the other monocots patterns of smaller groups of taxa as shown by the fol- sampled. Nevertheless, the dicot lamina always differen- lowing examples. Davie's (1917) early study of venation tiatesfrom the upper primordium zone, and in monocots in Philodendron selloum concluded that leaf and pinna from the lower primordium zone. The reticulate venation traces are related to the vascular system found in the stem of aroid leaveshasto be viewed asconvergent with the in such a way as to be independent of the size of the leaf dicot leaf pattern via a basically different ontogeny and itsmanner of development. Additional venation pat- (Kaplan 1973b). Therefore, termsfrom Hickey's(1979) ternswere describedby Knecht (1983) in African (Ivory original leaf venation scheme, later modi®ed to include Coast) Amorphophallus, Anubias, Cercestis, Culcasia, the net-veined monocots(Leaf Architecture Working and Xanthosoma. Other genera treated include Lemna Group 1999), should be used only in a topographical minor (Fletcher and Arnott 1963), and Arum, Monstera, sense and not suggestive of homologies. Philodendron and Pinellia among observations on numer- Venation. The ®rst comprehensive survey of leaf ven- ousfamiliesby Freundlich (1908). ation in the family wasby Ertl (1932) who sampled37 Cuticle and blade surface. Barthlott and Wollenweber genera, and described the ontogeny for many. He con- (1981) examined the taxonomic signi®cance of epicuti- cluded that higher ordersof venation are alwaysreticul- cular waxes. A comprehensive inventory of epicuticular ate except in Gymnostachys. Ertl also described vascular waxesin monocotsby Fro È lich and Barthlott (1988) bundle con®guration in petiole transverse sections and revealed a wide variety of sculpturing, extrusions, rodlets, related the patterns positionally to vascularization of the and irregular particles that have some systematic value. lamina. In general, the dorsal-most petiolar bundle Arales(sixteentaxa examined) were found to have either becomesthe central vein of the midrib. The lateral or no wax secretions, or unspeci®c patterns, or non-oriented ventral-most bundles diverge into the lamina closer to its platelets( Alocasia). Non-oriented typesare widespread base. However, the degree of bundle branching and cros- and not helpful for interfamilial comparisons. There is sing over (superpositioning) of other veins varies consider- experimental evidence that cuticle development isin¯u- ably among genera. Ertl held that most genera show enced by the environment, which suggests that cuticle basiplastic maturation (but see Kaplan 1973b) and con- thickness should be interpreted cautiously (Sutter 1985). nection of vascular bundles in the base of the lamina and Epidermis. P®tzer (1872) noted the occurrence of a petiole. Where pinnate venation occursin monocots,as water-bearing epidermisin Philodendron and Anthurium, in Araceae, Kaplan (1973b) concluded that lamina con- and Mittman (1888) described the spur or prick on the struction is basically linear due to the presence of the petiole of Lasia. Numerous observations of cell shape in multistranded `midvein'. The wider blade is the result of surface view are available. Examples include the works of differential expansion between the peripheral and central Linsbauer (1930) and Webber (1960) who also observed regions of the blade. Where pinnate secondary veins show stomata. Pant and Kidwai 1966) noted that in the genera an arcuate or diagonal course, they arrive at that orienta- Caladium, Schismatoglottis, and Zamioculas, whose poly- tion by developmental displacement of an originally lon- gonal epidermal cells show straight anticlinal walls in sur- gitudinal alignment (Ertl 1932; Troll 1939; Kaplan 1973b). face view, the wallsappear sinuousina lower plane of There are aroidswith true reticulate higher order vas- focus; they described papillate and other forms of cells. culature in addition to other typical dicot-like features Pant and Kidwai (1966) recorded surface orientation includingareoles,freeveinendingsandmarginalvenation. of stomata as mostly irregular or random in all aroids, but Kaplan (1973b) and Inamdar et al. (1983) offered reviews longitudinal in Gymnostachys and Lemnoideae; stomatal of leaf venation literature, describing in detail venation frequency wasfound to be 18±159 per mm 2, with no mode. 12 Introduction

Average guard cell length, stomatal index, and cuticle aerenchyma. Pant and Kidwai (1966) found that arm- thickness was also calculated by Pant and Kidwai (1966) cells are present in the spongy mesophyll in all aroids and all showed a broad range and mode. Shaw (1992) studied, except Pistia which hasunlobed cells.Ogden's calculated mean stomatal frequencies for different parts (1974) study of aquatic plant anatomy included obser- of Monstera deliciosa leavesand found broad variances vationson the leavesof four aroid genera; nearly all genera of frequencies depending on leaf form, position and size. in the study have type 3 aerenchyma (classi®cation of this In a given leaf, stomatal dimensions vary little, i.e., they volume). Evolutionary trendsare not well understood belong to a single size class as is typical for monocot leaves regarding aerenchyma development. (Dunn et al. 1965). Fryns-Claessens and Van Cotthem Specializations, ecological anatomy. Miscellaneous (1973) found Araceae to have no clear boundary between studies have addressed ecological variation. An altitudinal well-de®ned subsidiary cells and non-distinctive neigh- gradient study of Xanthosoma by Korner et al. (1983) in bouringcells.Whilemoststomatainthefamilyarebrachy- New Guinea provided data on leaf size, stomatal fre- paracytic, there are percentages of non-typical subsidiary quency, cuticle thickness and cell-wall thickness. But the cell arrangement in most genera observed. Non-typical limits of variation are not well-understood for most includes two polar subsidiary cells over guard cells at one systematic anatomical investigations. Lee and Richards end, or at both ends, or tricytic. Grau (1983) looked at (1991) noted substantial age-related variation when cell orientation on upper and lower surfaces of 24 species describing Syngonium and Monstera leaves, and Sutter in 15 genera, and Czaja (1962) brought polarization (1985) found environmental-based variation in cuticle techniques to bear on guard cell and subsidiary cell studies. Kasapligil (1961) investigated the seasonal geo- orientation. phytic Arum orientale of the Anatolian steppe ¯ora and Developmentally, stomata in most Araceae sampled found that leavesshowedvery little anatomical variance have been characterized asperigenousby Pant (1965) when compared to normal mesophytic anatomy. and Tomlinson (1974). Guard cell ontogeny was observed by Pant and Kidwai (1966) for eleven genera of Araceae. Guard cell initialsdivide once producing guard cells.An Sun/shade leaves adjacent sister cell, usually unrecognizable from other Lindorf (1980) examined shade leaf anatomy in Anthur- neighbouring cells, becomes one of the subsidiary cells. ium,DieffenbachiaandMonstera,andInaba(1984)studied However, these authors, as well as Fryns-Claessens and the effectsof shadingon the leavesof Amorphophallus Van Cotthem (1973), found Lemnaceae and Pistia sto- konjac. Araus et al. (1986a), who included Philodendron mata to be anomocytic, since perigene neighbours remain scandens in a larger study of shade plants, noted that the undivided. The survey by Stebbins and Khush (1961) internal mesophyll area in this C3-photosynthesis type considered Araceae, Cannaceae, and Musaceae to have leaf is correlated with the light-saturated photosynthesis eumesogenousontogeny.NyawuaneandGill(1990)found rate. Shade leaveswere often found to have low palisade the family to be mostly eumesogenous although eumeso- volume but higher spongy tissue volume, which caused the perigenousontogeny wasreported in Dieffenbachia and shade leaves to be thicker rather than thinner than sun Scindapsus. They found no correlation between onto- leaves in this species. Chloroplast density was found to be genetic type and the presence of straight-walled poly- lower but the chloroplasts larger. These ®ndings imply gonal or sinuous-walled mature epidermal cells. that volumes and proportions of tissues, and air cavity Systematic use of stomatal types in subfamilies and space, would be dif®cult to use systematically. However, tribes should be interpreted cautiously until more is the typology and unusual cell shapes noted in the present learned of the contribution of ecological versus genetic study probably have some systematic utility. factors. Numerous informative contributions include those of Porsch (1905), Mashima (1964: Arisaema), Grear (1973: Orontium), Roth and Merida de Bifano (1979), Perforations (fenestration) and Gill and Karatella (1983: Colocasia). Stomatal Tre cul (1854) made the ®rst descriptions of the ontogeny hydathodeshave been reported in about 65 speciesof ofleafperforationsin PothosandMonstera.LaterMelville aroidsby Ziegenspeck(1949). and Wrigley (1969), studying Monstera deliciosa, deter- Mesophyll. This tissue has been little studied sys- mined that small necrotic spots occur in intercostal tissue tematically, although the current survey offers typology when primordia are c. 14 mm in length. The spots expand and shows some intriguing differences among tribes. The toward the margin becoming sinuses or perforations ¯oating leavesof Pistia were reported by Sifton (1945) depending on their original position. (See also Melville to have 713 parts/1000 air volume in its highly organized 1971.) Review of Vegetative Anatomical Literature 13

Variegation and colouration Niklas (1991) measured the elastic modulus of a Hara (1957) studied variegation in leaves of 55 species in Spathiphyllum petiole and found that valuesincreaseas 24 familiesand determined four typesto exist. Arisaema the petiole dehydrates. was found to be of type II, having air spaces just beneath the lamina epidermis. Coloured leaf mottling in Arum Stems maculatum isaccompanied by a thinner lamina and more anthocyanin (Pethybridge 1903), a type IV variegation Investigations on stem anatomy constitute the largest (Hara 1957). Hara'stype I, having white streaksdueto body of work on aroid organ structure, much of which chlorophyll de®ciency, hasbeen found in Zantedeschia wasaccomplishedby J. C. French, often with collaborat- (Saxton 1913) and in six other genera by Bergdolt (1955). ors. Fisher and French (1978) investigated stem growth In white areas, the mesophyll is not differentiated and in 1±3 species of Aglaonema, Anthurium, Pistia, Spathi- thinner than in adjacent green tissue, or may be normally phyllum, and Zantedeschia. The two typesof meristems developed. Araus et al. (1986b) have described the struc- found are uninterrupted meristems (UM) and intercalary tural differencesbetween green and white sectorsof meristems (IM). In the former, cell divisions are con®ned variegated leavesin Scindapsus aureus. Optical properties progressively to the distal region of a developing inter- of the whitish segments of variegated leaves of Dieffen- node, causing acropetal tissue maturation. In the latter, bachia sp. and Spathiphyllum wallissii were measured localized regionsof cell divisionsoccurin the lower by Eller and Flach (1991); they found that absorption (proximal) regionsof developing internodes;maturation of energy took place in proportion to the amount of is basipetal. Frequently, species with sheathing leaf bases chlorophyll present. A number of studies reported investi- have IMs, while species with UMs have stems unpro- gationson the genetic basisof variegation including tected by leaf bases. those of Zettler and El-Nil (1979) and Henny (1983a, French and Tomlinson (1980) began studies in stem 1986). vasculature with a broad generic sample including a species each of Culcasia, Epipremnum, Philodendron, and Syngonium. Basing their approach on Tomlinson's earlier Foliar glands work on palm stem nodes and internodes, they character- In leaf tips(VorlaÈ uferspitzen), Volkens (1883) described ized the vascular patterns and the resultant changes in water excretion asa liquid in Calla and Gentner (1905) vascular bundle histology as bundles change position in observed the structures in eleven genera of aroids. Most the stem and node. They concluded (see also French and genera have a cylindrical hooded tip with the abaxial Tomlinson 1986) that compound bundles are temporarily surface oriented outward on all sides. Guttation in juxtaposed separate strands, and that aroid vasculature Anthurium occursthrough a large stomawith approx- is¯exible, often irregular, i.e., lesspredictablethan the imately seven subsidiary cells, not paracytically arranged. Rhapis-type of palms. This re¯ects the morphological Acorus leaves have a basically different leaf tip with slime diversity in the family (Tomlinson 1984). papillae (Gentner 1905). Following thiswork, French and Tomlinson(1981a,b, c,d, 1983, 1984) studied the vascular patterns of a much larger sample of araceous stems, organized by subfamily, Petiole with the ®nal paper dedicated to Philodendron alone. Ertl's(1932) ontogenetic studyof araceousleaf venation Another contribution (French and Tomlinson 1986) remainsthe mostthorough to date concerning the rela- reviewed compound vascular bundles in monocotyledon- tionship between petiole and lamina venation patterns. ous stems. They found bundles in aroids to be only He determined that the apparently scattered pattern of super®cially similar to those in the compared families petiole venation, as seen on a transverse section, is related Cyclanthaceae and Pandanaceae. Compound bundles to lamina venation in a precise way. Early in develop- were reported in Cercestis, Dieffenbachia, Montrichardia, ment, in all genera examined, the young lamina shows a Philodendron, Rhodospatha, Stenospermation, and moretypicalmonocotyledonousparallelvenation.Insome Zamioculcas. cases the medial to marginal relationship of petiole veins Asreviewed by French (1997, in MBB, p. 14) a to the lamina ispredictable and parallel (e.g. Crypto- demarcation of the inner boundary of a stem cortex by coryne, Orontium, Philodendron). In others, the veins anendodermishasbeenreportedin Amydrium,Anthurium, may cross over each other and branch more frequently Orontium, Monstera, Pistia, Rhaphidophora, Scindapsus, (e.g. Anthurium, Arum, Typhonodorum) producing a more Spathantheum,Symplocarpus,andinthetribesPeltandreae complex dorsiventral layered vascular development. and Schismatoglottideae. It was not encountered in most 14 Introduction genera (French and Tomlinson 1980, 1981a±d, 1983, Hinchee (1981) reported structural transitions as aerial 1984).VanFleet(1942)studiedtheendodermisandoxidase rootsenter soilin Monstera. The basic features of aerial system in hundreds of monocot genera, including the and transitional roots are found to be quite similar and centripetal rate of deposition of endodermal walls against they are not reliably distinguished microscopically. certain variables. Casparian strips were found to originate Aerial roots were ®rst characterized for 20 species in as extracellular deposits ®rst formed in intercellular space. thegeneraAnthurium,Calla,Philodendron,Rhaphidophora Radial cell growth caused casparian strips to take a posi- and Scindapsus by Olivier (1881). Goebel and Sandt (1930) tion between radial walls. Van Fleet (1942) observed four made similar observations as did Went (1895) for seven conditionsin Araceae: (1) endodermispresentin hori- genera of climbersand epiphytes.Sinnott and Bloch zontal and erect underground stems: Symplocarpus, and (1946) and Rakovan et al. (1973) recorded observations also Acorus (Acoraceae); (2) endodermisin aerial portions for Monstera, asdid Napp-Zinn (1953) and Serobjan of erect, prostrate or decumbent aerial stems and in (1974: Philodendron). Guttenberg (1940, 1968) reviewed culms and scapes as cylinder endodermis: Scindapsus available observations on aerial roots for over 30 genera. pictus; (3) endodermis absent in sub-aerial stems which do Among terrestrial roots, the contractile type has been not grow up to the ground line: Arum (3 spp.), Colocasia described for a few cormose genera. They are distin- esculenta; (4) endodermis absent in aerial or sub-aerial guished by unusual form and structure, particularly by stems, culms, scapes, or leaves: species of Aglaonema, their wrinkled surface after contraction as reported in Amorphophallus, Arisaema, Peltandra, Philodendron, and detail by Scott (1908). Lamant and Heller (1967) des- Symplocarpus. Wilson and Peterson (1983) made obser- cribed them for Arum. Contractile rootsconstitutea vationson the additionsof lignin or suberinto the principal morphological adaptation of geophytesin the endodermis or other tissues. dormant season as they draw corms down to their appro- Marlatt (1970) described wound cork formation from priate `physiological depth' (PuÈ tz 1991). A `channel stem cuttings of Dieffenbachia, documenting eventual effect' occurswhere contractile rootspull the organ dir- formation of a full periderm. Phellem tissue was shown ectly following it, thusclearing the way for the corm or by Keating (unpublished results) to bisect damaged offset (PuÈ tz 1992). Contractile rootsoriginate from the Epipremnum stems isolating necrotic lesions. Additional summit of the corm in Sauromatum guttatum and then literature on cicatrice formation, lenticelsor periderm is exert downward pull against full organ resistance, hence found in Klebahn (1884), Muhldorf (1927), Bolli (1953), having 0% channel effect. Contractile rootsin Arum and El-Hadidi (1969). italicum exert a 20% channel effect. Several miscellaneous studies exist on tuber and corm anatomy but no larger survey. Pinellia cordata tuberswere Root zonation and histology found to possess a central ring of amphivasal vascular Structurally, aroid rootshave many typical monocot bundles surrounding a circular phloem strand (Higuchi featuresand the following zonation or layershave been and Okada 1980). The ground tissue contained enlarged reported for most of those studied: epidermis, exodermis, circular cellseach bearing a singleraphide bundle, or mul- hypodermis, cortex, endodermis, and the stele, consisting tiple bundlesin a clumped arrangement; the outer surface of a pericycle surrounding equal numbers of xylem and was protected by periderm. Ohtsuki (1960) investigated phloem strands on alternating radii. In older roots, the tuber reserve photosynthate, mostly of Amorphophallus stele centre may consist of ligni®ed ®bres or mature rivieri. In that species the tuber ground tissue consisted of metaxylem. French (1987c), in his study of the sclerotic large cellseach containing one large mucilage grain, hypodermis (see below), described various anatomical 0.5±1.0 mm diameter, composed mostly of mannose and featuresin rootsof 280 speciesin91 genera. some d-glucose; small cells surrounding them were ®lled Ontogeny of young rootshasbeen describedby withsmallcompoundstarchgranules.Itappearsthattuber Berquam (1972) for Syngonium and the quiescent centre structure and its storage tissue can be quite specialized; it in the Pistia root primordium wasdescribedby Clowes is suf®ciently interesting to warrant a general survey. See (1984, 1985, 1992). Majumdar (1929) provided observa- additional contributionsby Paliwal and Kavathekar tionson the secondaryroot growth of Amorphophallus. (1972), Anandakumar et al. (1982), and Hather (1993). Root surface tissue varies according to age of the organ. In young roots, the epidermis is usually a single transient layer. The velamen, de®ned by Esau (1960) as a multiple Roots epidermis, has been described in aerial roots of aroid All roots of araceous genera investigated are adventitious. epiphytes by Deshpande (1956), who also recorded one They are usually divided among aerial or terrestrial, and asoccurring in Spathiphyllum, a terrestrial aroid. The Review of Vegetative Anatomical Literature 15 velamen has3±4 cell layers®lled with granular matter. epidermis that is distinct from underlying tissues. They Root hairs were described by Pinkerton (1936) as arising proposed that the exodermis be de®ned as a hypodermis from the hypodermisin Philodendron cordatum. containing a casparian band. That interpretation is As reviewed by French (1987c) outer cortical tissues followed in the present study but other de®nitions of are divided among hypodermis, exodermis and cortical exodermisappear in the literature. sclerenchyma. The hypodermis is highly localized and Shishkoff (1987) surveyed 148 angiosperm families for therefore potentially useful systematically. It forms the the presence of a dimorphic hypodermis (DH), a structure outer boundary of the cortex, distinguishing it from the that has suberized and non-suberized cells in an alternat- epidermis. In most Araceae, the cortex consists entirely ing regular pattern. More than half the genera sampled of thin-walled unligni®ed parenchyma. Some genera have have the DH including one or more species of Anthurium, peripheral collenchyma (e.g. Anadendrum, Anthurium, Arum, Caladium, Dieffenbachia, Monstera, Philodendron Aridarum, Hottarum, Monstera, Pothoideum, Pothos, and and Remusatia (Gonatanthus). Rhaphidophora, aswell as Acorus (Acoraceae)). Root periderm and related tissues have infrequently Trichosclereids were found by French (1987c) in some been investigated in the family. Weisse (1897) ®gured genera, where they develop in the intercellular ground lenticelson aerial rootsfor severalAraceae and reviewed space by intrusive growth (Holochlamys, Monstera, earlierreports.Dutt(1956)describedthelenticelsofPothos Rhaphidophora, Rhodospatha, Scindapsus, Stenosperma- asbeing `longitudinal' although Eamesand MacDaniels tion). Thick-walled pitted sclereids form narrow or wide (1947) had reported that lenticels are always transverse bandsadjacent and external to the endodermisin in roots. Causal factors in cork formation were studied Anthurium, variousMonstereae,and Heteropsis. Resin in the family by Priestley and Woffenden (1922). In canals with sclerotic sheaths form in Philodendron, Philodendron erubescens, a cortical cork layer arises Cercestis and Culcasia. Systematic conclusions on the beneath the sclerenchymatous exodermal region in aerial placement of eight genera are offered by French. roots and outside the incomplete secondary endodermis. It appearsthat rootshave a penetration or moisture In Monstera deliciosa, it wasfound that asaerial roots barrier at the hypodermisaswell asthe endodermis. age they never produce a secondary endodermis. As the Endodermal casparian bands also appear in all species protoplasts of the primary endodermis become more per- that have an exodermisor hypodermal casparianband. meable, cork appearsbeneath exodermis.Marlatt (1970) French (1987c) described the sclerotic hypodermis (SH) described details of periderm formation as wound cork in and other anatomical featuresin rootsof 91 genera and Dieffenbachia where suberin is deposited on wound sur- 280 species. The SH has various forms but often has faces up to 28 cells deep and starch granules disappear uniformly thickened elongated cellswith ligni®ed walls, from neighbouring cells. 1±5 cellsdeep, adjacent to the exodermis.The SH is Tracheary elementsin rootshave long been known usually sharply delimited from subadjacent parenchyma. to be vessels in most monocots that have been investi- The exodermisisde®ned ashaving either differentially gated (Cheadle 1942; Wagner 1977). Using the SEM, thickened periclinal walls, or lacking thickened walls. Schneider and Carlquist (1998) and Carlquist and Wilson and Peterson (1983) noted that the hypodermis Schneider (1998) studied root samples of the subfamilies of most roots has suberin and often lignin present. Philodendroideae and Colocasioideae, respectively. In all Perumalla et al. (1990) reported that most Araceae sur- cases, vessels were present although they might be called veyed for thisfeature have a uniseriatehypodermiswith incipient vessels. The absence or presence of pit mem- casparian bands and an epidermis. In the Araceae, cas- branes, a de®ning feature, cannot be distinguished by light parian bandsmay be suberized(S) or have lignin and microscopy.Perforationplatesarenoteasilydistinguished suberin (LS). Aroids examined included species of from wall pitting. Anthurium (S) or (LS), Arisaema (LS), Dieffenbachia (S) Tieghem (1870±1) supplied a list of c. 20 aroid genera or (LS), Philodendron (S), and Syngonium (S). The results that have secretory tissue. Resin canals were described suggested that bands are systematically variable and not by French (1987b) from a survey of roots of 91 genera ecologically related. Philodendron wendlandii wasfound in the family. They were found in all material examined to have a multiseriate hypodermis and both layers have a of Cercestis, Culcasia, Furtadoa, Homalomena and lignosuberized casparian band (Peterson and Perumalla Philodendron. Canal contents tested positively with Sudan 1990). They found that the multiple hypodermisbegins IV and canalsoften had ligni®ed or collenchymatous asone layer and maturescentripetally, and may develop sheaths that may be diagnostic. asynchronously. They followed Esau's (1977) de®nition The ontogeny of root branching wasobservedin about that the hypodermis consists of a layer beneath the 11 genera of Araceae by Tieghem and Duliot (1888). An 16 Introduction observation of root branching in Pistia (Charlton 1983) found in MBB wasalsocompared in detail. Anatomical revealed some regularity in lateral primordium spacing variation wasfound here to match the French et al. that began 100±350 mm behind the root cap junction. topology with little ambiguity and most trends of spe- Apogeotropic roots(rootsthat emerge from the cialization (listed below) can be interpreted parsimoni- ground) have been described by Sanford (1987). This ously. The family infrastructure presented by MBB is phenomenon of nutrient-poor areashasbeen recorded in close to the new proposed outline as far as the ®rst 31 twelve species of ®ve families. A Philodendron species genera, the bisexual-¯owered Araceae. The placement by from the Amazonian Tierra Firma and Anthurium MBB of the remaining 74 unisexual-¯owered genera ellipticum (Bruhn 1910) are given asexamplesbut with- within the subfamily Aroideae embeds a number of out observations on growth or anatomy. The total number distinct groupings which are addressed in the revised of apogeotropic species and their mycorrhizal status are classi®cation presented here. unknown. The bisexual (hermaphroditic) condition will be taken Mycorrhizal symbiosis is believed to be widespread as plesiomorphic in common with all recent classi®cations in both dicotsand monocotsbut only limited informa- of the family. Thiswill begin the processofpolarizing tion isavailable for aroids.Brundrett and Kendrick character states, many of which have not been analyzed (1988, 1990) surveyed temperate plants with vescicular- across the family or reliably polarized before. arbuscular mycorrhizae (VAM) and described Arisaema In spite of the great variety of habits and habitats in atrorubens asbeing obligately mycorrhizal; fungal colon- the family, most anatomical characters show gradual or ization occurred in 54% of roots sampled. In this species subtle variation across subfamilies and tribes. The family cortical air channelsfacilitated the spreadof VAM fungi, seems to have settled into a characteristic anatomical but when rootsbecame contractile, VAM colonieswere Gestalt at an early stage in its radiation. There are very few lost. Nasim (1990) described roots containing VAM in characters or character suites that change abruptly across Colocasia. tribal or subfamilial lines. The greatest character richness Only limited additional observations on cell-wall occursin the subfamilyPothoideae (including the mon- chemistry and structure are available for aroid roots. steroids as a tribe) with a loss of, or simpli®cation in, Wilson and Peterson (1983) found suberin and lignin in charactersin more derived tribes. Philodendron encom- epidermal wallsof Dieffenbachia picta, typical of half of passes the greatest anatomical variation of any single ¯owering plant roots so tested, and walls also tested genus. As one moves across the Aroideae as de®ned here, positive for cellulose. Central cortical walls were cellulosic there is often very little histological variation and hist- and no differenceswere detected between geophytic and ology alone isunreliable diagnostically.Trendsin con- non-geophytic plants. In a study of isotropic layering of spicuous and interesting tissues will be described as root ®bre walls of several monocot families, Mueller and follows. Beckman (1979) reported isotropic layers in the ®bres of only Monstera deliciosa among Araceae examined; the Collenchyma several ®bre layers were rich in pectin and lignin and corresponded to primary wall composition. Such layers This tissue, found in all subfamilies except Gymnos- were found only in mature root tissues, showing up tachydoideae, has emerged as the most interesting new infrequently in greenhouse-grown plants in the oldest character in thisstudy(Keating 2000; GoncËalves et al. roots. 2002). In general, collenchyma development isgreatest where ®bre development isleast.In the Potheae, collen- chyma does not coexist with ®bre-ensheathed vascular ANATOMICAL CHARACTER bundles. In Anthurieae and throughout the remainder of STATE TRANSFORMATIONS AND that clade, collenchyma usually occurs as bands around THEIR POTENTIAL USE IN the periphery of the petioles and midveins, and occasion- PHYLOGENETIC ANALYSIS ally as bundle caps. It is present as bands or caps in the orontioids, except Orontium. Bands are present in most The order of genera presented in this volume follows the lasioids and absent in Calla. revised outline of Keating (2002b; Tables 1 and 5, this Nearly all genera of philodendroidshave banded volume). It is based on a synthesis of data accumulated in collenchyma. Interrupted bandsare alsofound in Philo- thisstudycombined with the topology produced by dendron and in Zantedeschia. The African and Asian French et al. (1995), whose survey was based on cpDNA genera of the Aglaonemateae (including ®ve genera in the restriction site analysis. The morphology and geography present outline) have interrupted bands and strands that Anatomical Character State Transformations and their Potential Use in Phylogenetic Analysis 17

Table 1. Generic list ofAraceae and comparisons ofrecent classi®cations

Genera Keating 2002b MBB 1997 Grayum 1990 Bogner & Nicolson 1991

Acorus Acoraceae Acoraceae Acoraceae Acoraceae Gymnostachys GY GY PO Gymnost GY Orontium OR Oront OR LA Oront LA Oront Lysichiton OR Symploc OR LA Symploc LA Oront Symplocarpus OR Symploc OR LA Symploc LA Oront Pothos PO Poth PO Poth PO Poth PO Pedicellarum PO Poth PO Poth PO Poth PO Pothoidium PO Poth PO Poth PO Poth PO Anthurium PO Poth PO Anthur PO Anthur LA Anthur Holochlamys PO Monster MO Spathiph PO Spathiph MO Spathiph Spathiphyllum PO Monster MO Spathiph PO Spathiph MO Spathiph Rhodospatha PO Monster MO Monster PO Monster MO Monster Stenospermation PO Monster MO Monster PO Monster MO Monster Scindapsus PO Monster MO Monster PO Monster MO Monster Rhaphidophora PO Monster MO Monster PO Monster MO Monster Anadendrum PO Monster MO Anaden PO Anaden MO Anaden Monstera PO Monster MO Monster PO Monster MO Monster Alloschemone PO Monster MO Monster PO Monster MO Monster Epipremnum PO Monster MO Monster PO Monster MO Monster Amydrium PO Monster MO Monster PO Monster MO Monster Heteropsis PO Monster MO Heterop PO Monster MO Heterop Cyrtosperma LA LA LA LasLA Las Lasimorpha LA LA LA Las Podolasia LA LA LA LasLA Las Lasia LA LA LA LasLA Las Anaphyllum LA LA LA LasLA Las Urospatha LA LA LA LasLA Las Anaphyllopsis LA LA LA Las Pycnospatha LA LA LA LasLA Las Dracontium LA LA LA LasLA Las Dracontioides LA LA LA LasLA Las Calla CA CA CA Calle CA Montrichardia PH Philoden AR Montrich CA Montrich LA Montrich Anubias PH Philoden AR Anub CA Anub PH Anub Furtadoa PH Philoden AR Homalom CA Homalom PH Philoden Philodendron PH Philoden AR Philoden CA Philoden PH Philoden Homalomena PH Philoden AR Homalom CA Homalom PH Philoden Zantedeschia PH Zanted AR Zanded CA Zanted PH Zanted Callopsis PH Zanted AR Callop CA Callop LA Callop Stylochaeton PH Stylochae AR Stylochae LA Stylochae AR Stylochae Gonatopus PH Stylochae AR Zamiocul PO Zamiocul LA Zamiocul Zamioculcas PH Stylochae AR Zamiocul PO Zamiocul LA Zamiocul Nephthytis PH Aglaon AR Nephthyt CA Nephthyt LA Nephthyt Anchomanes PH Aglaon AR Nephthyt CA Nephthyt LA Nephthyt Pseudohydrosme PH Aglaon AR Nephthyt CA Nephthyt LA Nephthyt Aglaonema PH Aglaon AR Aglaon CA Aglaon PH Aglaon Aglaodorum PH Aglaon AR Aglaon CA Aglaon PH Aglaon Culcasia PH Culcas AR CulcasCA CulcasLA Culcas Cercestis PH CulcasAR CulcasCA CercesLA Nephthyt Bognera PH Spathicar AR Dieffen CA Bogner PH Anubiad Dieffenbachia PH Spathicar AR Dieffen CA Dieffen PH Dieffen Spathantheum PH Spathicar AR Spathicar CA Spathicar AR Spathicar 18 Introduction

Table 1. continued

Genera Keating 2002b MBB 1997 Grayum 1990 Bogner & Nicolson 1991

Gorgonidium PH Spathicar AR Spathicar CA Spathicar AR Spathicar Synandrospadix PH Spathicar AR Spathicar CA Spathicar AR Spathicar Gearum PH Spathicar AR Spathicar CA Spathicar AR Spathicar Spathicarpa PH Spathicar AR Spathicar CA Spathicar AR Spathicar Asterostigma PH Spathicar AR Spathicar CA Spathicar AR Spathicar Mangonia PH Spathicar AR Spathicar CA Spathicar AR Spathicar Taccarum PH Spathicar AR Spathicar CA Spathicar AR Spathicar Phymatarum SC Schismat AR Schismat CA Schismat PH Philoden Schismatoglottis SC Schismat AR Schismat CA Schismat PH Philoden Aridarum SC Schismat AR Schismat CA Schismat PH Philoden Piptospatha SC Schismat AR Schismat CA Schismat PH Philoden Hottarum SC Schismat AR Schismat CA Schismat PH Philoden Bucephalandra SC Schismat AR Schismat CA Schismat PH Philoden Heteroaridarum SC Schismat AR Schismat CA Schismat PH Philoden Cryptocoryne SC Cryptocor AR Cryptocor AR Cryptocor AR Areae Lagenandra SC Cryptocor AR Cryptocor AR Cryptocor AR Areae Spirodela LE Lemna LE Wolf®a LE Wolf®ella LE Amorphophallus AR Thomson AR Thomson AR Thomson AR Thomson Pseudodracontium AR Thomson AR Thomson AR Thomson AR Thomson Hapaline AR Calad AR Calad CO Calad CO Steudner Syngonium AR Calad AR Calad CO Calad CO Syngon Xanthosoma AR Calad AR Calad CO Calad CO Calad Chlorospatha AR Calad AR Calad CO Calad CO Calad Ulearum AR Calad AR Zomicar CA Callop AR Zomicar Filarum AR Calad AR Zomicar CA Callop AR Zomicar Zomicarpella AR Calad AR Zomicar CA Callop AR Zomicar Caladium AR Calad AR Calad CO Calad CO Calad Scaphispatha AR Calad AR Calad CO Calad CO Calad Jasarum AR Calad AR Calad CO Calad CO Calad Zomicarpa AR Calad AR Zomicar CO Zomicar AR Zomicar Ambrosina AR Arisar AR Ambros AR Ambros AR Areae Arisarum AR Arisar AR Arisar AR Arisar AR Areae Peltandra AR Peltand AR Peltand CA Peltand PH Peltand Typhonodorum AR Peltand AR Peltand CA Peltand PH Typhon Colletogyne AR Peltand AR Arophyt CA Arophyt AR Arophyt Carlephyton AR Peltand AR Arophyt CA Arophyt AR Arophyt Arophyton AR Peltand AR Arophyt CA Arophyt AR Arophyt Pistia AR Pist AR Pist AR Pist PI Arisaema AR Arisaem AR Arisaem AR Arisaem AR Areae Pinellia AR Arisaem AR Arisaem AR Pinell AR Areae Typhonium AR Areae AR Areae AR Areae AR Areae Theriophonum AR Areae AR Areae AR Areae AR areae Sauromatum AR Areae AR Areae AR Areae AR Areae Lazarum AR Areae AR Areae Biarum AR Areae AR Areae AR Areae AR Areae Arum AR Areae AR Areae AR Areae AR Areae Eminium AR Areae AR Areae AR Areae AR Areae Dracunculus AR Areae AR Areae AR Areae AR Areae Helicodiceros AR Areae AR Areae AR Areae AR Areae Anatomical Character State Transformations and their Potential Use in Phylogenetic Analysis 19

Table 1. continued

Genera Keating 2002b MBB 1997 Grayum 1990 Bogner & Nicolson 1991

Ariopsis AR ColocasAR ColocasARAriopsCO Ariops Alocasia AR ColocasAR ColocasCOColocasCOColocas Remusatia AR ColocasAR ColocasCOColocasCOSteudner Colocasia AR ColocasAR ColocasCOColocasCOColocas Steudnera AR ColocasAR ColocasCOColocasCOSteudner Protarum AR ColocasAR ColocasCOColocasCOProtar

Key to abbreviations: Subfamilies ˆ pair of upper case letters; Tribes ˆ more than two letters Subfamilies:ARˆ Aroideae, CA ˆ Calloideae, GY ˆ Gymnostachydoideae, LA ˆ Lasioideae, LE ˆ Lemnoideae, MO ˆ Monsteroideae, OR ˆ Orontioideae, PH ˆ Philodendroideae, PO ˆ Pothoideae, SC ˆ Schismatoglottoideae Tribes: Aglaon ˆ Aglaonemateae, Ambros ˆ Ambrosineae, Anadend ˆ Anadendreae, Anthur ˆ Anthurieae, Anub ˆ Anubiadeae, Areae ˆ Areae, Ariops ˆ Ariopsideae, Arisaem ˆ Arisaemateae, Arisar ˆ Arisareae, Arophyt ˆ Arophyteae, Bogner ˆ Bognereae, Calad ˆ Caladieae, Calle ˆ Calleae, Callop ˆ Callopsideae, Cerces ˆ Cercestideae, Colocas ˆ Colocasieae, Cryptocor ˆ Cryptocor- yneae, Culcas ˆ Culcasieae, Dieffen ˆ Dieffenbachieae, Gymnost ˆ Gymnostachydeae, Heterop ˆ Heteropsideae, Homalom ˆ Homalomeneae, Las ˆ Lasieae, Monster ˆ Monstereae, Montrich ˆ Montrichardieae, Nephthyt ˆ Nephthytideae, Oront ˆ Orontieae, Peltand ˆ Peltandreae, Philoden ˆ Philodendreae, Pinell ˆ Pinelleae, Pist ˆ Pistieae; Poth ˆ Potheae, Protar ˆ Protareae, Schismat ˆ Schismatoglottideae, Spathicar ˆ Spathicarpeae, Spathiph ˆ Spathiphylleae, Steudner ˆ Steudnereae, Stylochae ˆ Stylochaetoneae, Symploc ˆ Symplocarpeae, Syngon ˆ Syngonieae, Thomson ˆ Thomsonieae, Typhon ˆ Typhoneae, Zamiocul ˆ Zamioculcadeae, Zanted ˆ Zantedeschieae, Zomicar ˆ Zomicarpeae.

are not aligned with vascular bundles (type Sb). This rare below the leaf bundles. Its phloem strands are circular con®guration contributesto their being placed in a single and surrounded by deep ®bre development, unlike any tribe. other aroid. Separate phloem and xylem capsremain The Aroideae are delimited by the presence of stranded frequent in the pothoids and monsteroids, the lasioids, or (type Sv) collenchyma. At the base of the clade, the asphloem capsin Orontium. In the philodendroids, ®bre schismatoglottids show banded and banded-interrupted sheaths are found only in the Culcasieae, Asterostigma collenchyma that one could hypothesize as being trans- (Spathicarpeae), and Philodendron (Philodendreae), not itional to the stranded type Sv. particularly closely allied. Fibre caps occur sporadically However, there isanother trend in collenchyma evolu- in nearly all cladesof the subfamily. tion that might also be operating. In a few genera below Phloem ®bre caps are common in schismatoglottids the Aroideae (Lysichiton, Symplocarpus, Rhodospatha, and the Caladieae. They are occasionally present in the Scindapsus, Monstera, Gonatopus, and Dieffenbachia), distantly related Lazarum and Colocasia. In subfamily collenchymatousbundle capsdevelop asapparent Aroideae phloem and xylem capsare lesscommon. replacement of ®bre caps, suggesting a spatial homology The occurrence of trichosclereids in Araceae was (Patterson 1982), a different route to evolving type reviewed by Seubert (1997a) who reported them present Sv strands. Normally, where ®bre caps or sheaths occur, in one or several organs of eleven genera. She mentioned collenchyma, when present, is always banded. Stranded an additional report for Montrichardia by Solereder and collenchyma, aligned with vascular bundles, normally Meyer (1928) which wasnot con®rmed by Seubert or in occurswhen ®bresare absent,except in Caladieae where the present study. Here, the trichosclereid-bearing genera both may coexist. are placed mostly in the Pothoideae: Monstereae. There is an occasional report for genera of Potheae (Nicolson Sclerenchyma 1960:Pothos,2species;andinthepresentstudy:Anthurium pallens, Croat 35265). The distribution of this unusual Fibresare bestdeveloped in the tribesPotheae and cell type also contributed to the decision to place all of Monstereae where they frequently ensheath vascular these genera in one subfamily, rather than in two indis- bundles. The genus Gymnostachys shows a unique con- tinctly bounded subfamilies. The capacity to express ®guration for the family in having separate phloem and these cells is an apomorphy for this group that seems not xylem capsthat extend astranscurrentgirdersabove and to have been carried into more derived lines. 20 Introduction

Crystals shaped or biforine-like in descriptions, have the same general idioblastic shape but lack thickened walls or The four typespresentall appear to be formed of calcium terminal papillae. Spindle-shaped cells are found in oxalate asdetermined by polarization behaviour, and by Symplocarpus, Cyrtosperma and Calla of the hermaphro- reaction to the usual microtechnical reagents (see anato- ditic genera and are scattered through most tribes of the mical literature review). Druses and crystal sand (clusters Philodendroideae and Aroideae. of ®ne prismatics) are common throughout the family True biforinesare not encountered at all in the although not detected in all genera. Both of these as hermaphroditic aroids. They occur in only seven scat- recorded in descriptions show some architectural vari- tered genera of Philodendroideae, and 17 genera (seven ation but information was insuf®cient to allow analysis tribes) of Aroideae. They are certainly the most specia- of trends. Styloids, present in Anadendrum and Amor- lized crystal cell form because of this exclusive distribu- phophallus, are otherwise rarely found and are thus too tion in the monoeciousgenera. little known for further comment. Many questions remain regarding the development of The ubiquitousraphidescan be characterized both by these crystals as, for instance, whether biforine-like cells cell form and by con®guration of crystalline contents. are a phylogenetic or ontogenetic stage in the devel- They often occur in at least two forms in ground tissue of opment of true biforines. Raphides of a given mor- most genera and there is no certain morphocline among phology may occur in various sizes and relationships to these types. The simple, non-systematically useful form the surrounding tissue. In ontogeny, raphide cells can be of raphides: thin-walled, rounded or oval cells with a single identi®ed early cytologically asthey enlarge differentially raphide bundle within, isnearly universal.The cellsmay just a few hundred micrometres proximal to the apical be slightly or highly idioblastic, and variable in size with meristem or primordia surfaces. Beyond this little is respect to neighbouring ground tissue, but in the majority known regarding the scope of crystal cell initiation of genera the simple type coexists with additional raphide timing and expansion in this family. types whose distribution is of considerable interest (Keating 2002a). Summary oftrends ofspecialization First, the subfamily Pothoideae commonly presents elongated raphide cellswith ill-de®ned raphide bundles, For the above vegetative features and for those listed that is, crystals occur in elongated, overlapping, some- below, character state transformations (trends of spe- timeshelically oriented groups.Overlapping bundles cialization) are proposed re¯ecting anatomical and occur sporadically elsewhere in Lasioideae, Culcasieae, morphological evidence. In addition, the sequence and Aglaonemateae, Zamioculcadeae, Cryptocoryneae, topology of genera asgiven by French et al. (1995) Schismatoglottideae, Caladieae, Arisaemateae, and in helped con®rm the polarization of trends. The trends ®ve genera of Areae. listed below appear those most likely or determinable Mostly elongated cells bearing multiple discrete bun- from this study. In some cases, no trend is discernable, dles are common in bisexual-¯owered subfamilies except and in many cases the trends should be considered Calloideae. They also occur sporadically in the unisexual potentially homoplasious and not necessarily appro- Philodendroideae and Aroideae. Tubular cells, that priate for family-wide cladistic analysis. Most trends will extend greatly beyond the length of included raphide be useful for within-genus or within-tribe analysis. bundles, occur sporadically without any obvious trend. Venation (leaf paradermal): Trend 1: venation pinnate in Rarely encountered articulated raphide tubesare dis- leavesor segments ) acrodromous. Trend 2: secondary tinctive apomorphiesoccurring in Philodendroideae: and tertiary veinsreticulate ) parallel or converging Spathicarpeae: Taccarum and Asterostigma, and in secondaries and tertiaries. Trend 3: brochidodromous Biarum of Aroideae: Areae. loops ) straightened `collectors' (marginal or submar- Probably the most specialized and unusual crystal ginal). Trend 4: brochidodromy ) eucamptodromy. phenomenon, described early by Turpin (1836), is the Leaf structure: Trend: no differentiation ) dorsiventral biforine cell. These specialized raphide cells are heavy- ) isobilateral. walled, spindle-shaped or cylindrical, usually with thin- Trichomes: No trend. The several genera bearing tri- walled and pointed (papillate) ends. When deformed, chomes seem not to be closely related on the they are capable of shooting a ¯ow of raphides into the basis of other characters. Therefore, trichomes are mouth parts of grazing organisms (see Turpin 1836, autapomorphies. Middendorf 1983). Similar cells, which may represent Epidermal cells (paradermal): Trend 1: cells phylogenetic or ontogenetic stages, referred to as spindle- straight-sided polygonal and nearly isohedral ) cells Scope of Araceae and Comparisons of Recent Classifications 21

elongated. Trend 2: cell wallsstraight ) undulate ) (Anastomosing condition is an apparent homoplasious sinuous ) interlocking. apomorphy in Old World and New World groupsof Epidermis (TS): Trend: cellsshort ) taller ) columnar. Aroideae.) Stomata (PS): Trend 1: brachyparacytic (bpc) ) Tannin cells: Widespread and scattered in the family; anomocytic. Trend 2: brachyparacytic ) brachypara- evolutionary patternsobscure. tetracytic ) brachypara-hexacytic ) amphibrachy- Raphide cells (apomorphiesfor cellsand crystalsprob- paracytic. ably not due to a single morphocline). Trend 1: cells of Subsidiary cells: Trend: neighbouring cellsbroad and type 1 (thin-walled and irregular, slightly larger than irregular, little differentiated from epidermal cells ) surrounding cells of ground tissue) ) cellswider and broad trapezoidal, angular ) broad and rounded ) circular. Trend 2: cellsof type 1 ) cellselongate narrow and rounded. ) cellstubular. Trend 3: cellsof type 1 ) cells Stomata (TS): Trend 1: cellslevel with surface ) cellson articulated, usually end to end. Trend 4: cells of type mounds. Trend 2: guard cells with double cuticular 1 ) cells spindle-shaped ) spindle cells with thick ledges ) cells with single (outer) ledges. ligni®ed walls(biforines). Hypodermis: Trend: absent ) present. Raphide crystals: Trend 1: single bundles ) wide bundles. Palisade mesophyll: Trend 1: absent ) columnar ) Trend 2: single bundles ) crystals in oblique over- squarish. Trend 2: columnar ) T-shaped. Trend 3: lapping bundles. Trend 3: single bundles ) multiple columnar ) armed. Trend 4: columnar ) U-shaped. bundlesper cell. Spongy mesophyll: Trend 1: type 1 ) type 1a. Trend 2: Stem structure: Trend: central cylinder demarcated from type 1 ) type 2 ) type 3. Unknown: type 4 isapo- cortex ) central cylinder not discernable. morphic from an undetermined plesiomorphic state. Stem vascular bundles: Trend: collateral bundles ) Air cavities: Trend: low substomatal cavities, mostly asso- compound bundles ) amphivasal bundles. ciated with types1, 1a mesophyll ) high substomatal cavities mostly associated with types 2 and 3 mesophyll. Collenchyma: In general, collenchyma (in midribsand SCOPEOFARACEAEAND petioles) is least developed where ®bre caps or sheaths COMPARISONS OF RECENT are best developed. Trend 1: collenchyma absent ) CLASSIFICATIONS present. Trend 2: collenchyma absent ) banded (type B) ) bandsinterrupted (type Bi) ) strands on By any recent account, the family comprises over 100 alternate radii with vascular bundles (type genera placed in a variety of groupings. Four recent Sb) ) strands subtending peripheral vascular bundles schemes are compared in Table 1 and listed in Tables (type Sv). Trend 3: phloem ®bre caps ) collenchyma 2±5. The generic sequence for the anatomical descrip- cellsreplacing ®bre strands(typeSvc) ) rounded tions follows the proposed scheme by Keating (2002b, strands subtending peripheral vascular bundles Table 5), which attemptsto follow the topology of (type Sv). French et al. (1995, Table 6), and also most parsimoni- Vascular bundlesÐpattern in TS midrib: Trend 1: vascular ously aligns the array of anatomical data. bundles appearing symmetrically (disposed) organized The genus Acorus, traditionally treated asa member of ) irregularly scattered. Trend 2: vascular bundles the Araceae, isnow consideredon a variety of linesof numerous ) few. evidence to be in itsown family, Acoraceae (Grayum Vascular bundlesÐdegree of development in TS midrib or 1987; Buzgo 2001). Duvall et al. (1993), Davis petiole: Trend: type I ) type II ) type III. (1995), Soltis et al. (1997) and Stevenson et al. (2000), Sclerenchyma (®bres): Trend: ensheathing vascular using molecular evidence, concluded that Acorus is bundles ) capsover phloem and xylem ) capsover phylogenetically isolated and should be considered sister phloem ) absent. (see collenchyma Trend 3.) to all remaining monocots. Sclerenchyma (trichosclereids): Trend 1 (in Pothoideae The planktonic lemnoids, Landoltia, Lemna, Spirodela, only): absent ) present ) absent. Trend 2: solitary ) Wolf®a, and Wolf®ella, seem now to be clearly embedded aggregated. (Variationsin branching, if any, shape, within the Araceae. They are treated here assubfamily degreesof aggregation have unknown polarities.) Lemnoideae and inserted before the Aroideae as indic- Secretory ducts: Trend: absent ) present (scattered in 14 ated by French et al. (1995) and Maheshwari (1958). genera of 4 subfamilies) ) absent. Landolt (1998) was more cautious. First, he noted Laticifers: Trend: absent ) present, articulated and Grayum's(1992) observationsthatthe lemnoid uniporate non-anastomosing ) articulated and anastomosing. pollen with the ulcerate aperture type isunknown in 22 Introduction

Table 2. Classi®cation ofGrayum (1990) [ Acorus excluded from Araceae]

Family ARACEAE Juss.

Subfamily Pothoideae Engl. Tribe Gymnostachydeae Nakai Gymnostachys Tribe Spathiphylleae Engl. Spathiphyllum, Holochlamys Tribe Anthurieae Engl. Anthurium Tribe Potheae Engl. Pothos, Pedicellarum, Pothoidium Tribe Anadendreae Bogner & J. French Anadendrum Tribe Monstereae Engl. Subtribe Heteropsidinae Engl. Heteropsis Subtribe Monsterinae Schott Rhaphidophora, Monstera, Amydrium, Epipremnum, Scindapsus, Alloschemone, Stenospermation, Rhodospatha Tribe Zamioculcadeae Engl. Zamioculcas, Gonatopus Subfamily Calloideae Schott A. Calla Alliance Tribe Calleae Schott Calla B. NephthytisAlliance Tribe Nephthytideae Engl. Nephthytis, Anchomanes, Pseudohydrosme Tribe Callopsideae Engl. Callopsis, Ulearum, Filarum, Zomicarpella Tribe Montrichardieae Engl. Montrichardia C. Aglaonema Alliance Tribe Anubiadeae Engl. Anubias Tribe Zantedeschieae Engl. Zantedeschia Tribe Aglaonemateae Engl. Aglaonema, Aglaodorum Tribe Spathicarpeae Schott Mangonia, Asterostigma, Synandrospadix, Taccarum, Gorgonidium, Gearum, Spathantheum, Spathicarpa Tribe Dieffenbachieae Engl. Dieffenbachia Tribe Bognereae Mayo & Nicolson Bognera D. Peltandra Alliance Tribe Peltandreae Engl. Peltandra, Typhonodorum Tribe Arophyteae Bogner Arophyton, Carlephyton, Colletogyne Tribe Schismatoglottideae Nakai Schismatoglottis, Piptospatha, Bucephalandra, Phymatarum, Aridarum, Heteroaridarum, Hottarum E. Philodendron Alliance Tribe Culcasieae Engl. Culcasia Tribe Cercestideae Grayum Cercestis Tribe Homalomeneae (Schott) M. Hotta Furtadoa, Homalomena Tribe Philodendreae Schott Philodendron Subfamily Colocasioideae Engl. Tribe Zomicarpeae Schott Zomicarpa Tribe Colocasieae Engl. Subtribe Protarinae (Engl.) Grayum Protarum Subtribe Steudnerinae Engl. & K.Krause Steudnera Subtribe Remusatiinae Grayum Remusatia, Gonatanthus Subtribe Colocasiinae Schott Colocasia, Alocasia Tribe Caladieae Schott Subtribe Jasarinae Grayum Jasarum Subtribe Scaphispathinae Grayum Scaphispatha Subtribe Caladiinae Engl. & K. Krause Caladium, Xanthosoma, Chlorospatha, Aphyllarum Subtribe Syngoniineae Schott Syngonium Subtribe Hapalininae Engl. & K. Krause Hapaline Subfamily Lasioideae Engl. Tribe Symplocarpeae Engl. Symplocarpus, Lysichiton Tribe Orontieae R. Br. ex Dumort. Orontium Evolutionary Position of Araceae 23

Table 2. (Cont.)

Family ARACEAE Juss.

Tribe Lasieae Engl. Subtribe Dracontiinae Schott Cyrtosperma, Lasia, Anaphyllum, Podolasia, Urospatha, Dracontioides, Dracontium Subtribe Pycnospathinae Bogner Pycnospatha Tribe Stylochaetoneae Schott Stylochaeton Subfamily Aroideae Engl. Tribe Thomsonieae Blume Pseudodracontium, Amorphophallus Tribe Arisareae Dumort. Arisarum Tribe Pinellieae Nakai Pinellia Tribe Pistieae Blume Pistia Tribe Cryptocoryneae Blume Cryptocoryne, Lagenandra Tribe Ambrosineae Schott Ambrosina Tribe Ariopsideae Engl. Ariopsis Tribe Arisaemateae Nakai Arisaema Tribe Areae Engl. Arum, Dracunculus, Helicodiceros, Theriophonum, Typhonium, Sauromatum, Eminium, Biarum

Araceae, and so is their utricular fruit type. However, we Lemnoideae, recognized by several apomorphies. Finally, do know that pollen and fruit diversity is large in the a restricted subfamily Aroideae is based primarily on Araceae, as is secondary chemistry. Also, extreme habit stranded collenchyma (type Sv), holding eight tribes with specialization among lemnoids makes direct structural 37 genera. The genus Pistia, often put at the end of the comparisons dif®cult. As Landolt (1998) suggested, more family because of several structural autapomorphies, is detailed molecular genetic studies will be required among placed here asa monotypic tribe within the Aroideae the putatively related aroids, lemnoids and adjacent according to the cpDNA evidence (French et al. 1995). Alismati¯orae in order to clarify their relationships. Taxonomic notesare added at the end of generic The present total of 106 genera of Araceae includes anatomical descriptions, and for many suprageneric Gymnostachys as the basal and most isolated long-branch groupings. Where none appears, no distinctive features genus (subfamily Gymnostachydoideae). The other or relationships were evident from the sample. bisexual aroid genera are placed among four subfamilies. The main difference with past practice is the recognition of a subfamily Pothoideae that includes two tribes, EVOLUTIONARY POSITION OF Potheae and Monstereae, as there seems insuf®cient ARACEAE evidence for their recognition at the subfamily level. None Structural evidence of the structural data allows for mutually exclusive subfamilial groupings nor does the topology of French The earlier idea that Arales(Araceae and Lemnaceae) et al. (1995). Genera of the subfamily Lasioideae are belong near the Arecales, Cyclanthaceae and Pandanales compact enough to need no additional infrastructure was illustrated by Dahlgren (1975) and Cronquist (1968, above the generic level. 1981: subclass Arecidae). At about this time, it was In the MBB scheme, all remaining genera (over 2/3 of becoming clear on the basis of accumulating evidence that the total) are placed in the subfamily Aroideae. While the nearest neighbours are elsewhere. Dahlgren (1980) this line is monophyletic as they suggest, there are placed the Arales (Aranae) near the base of the monocots, branch-length differences(French et al. 1995) within that distal to the Alismati¯orae and Triuridi¯orae and prox- clade that readily support recognition of four subfamilies imal to the Lilii¯orae, probably the earliest such conclu- which are de®nitely supported by anatomical evidence. sion. On the basis of chemical characters, Dahlgren et al. In the outline by Keating (2002b) the subfamily Philo- (1981) reaf®rmed that position. The `Dahlgrenograms' of dendroideae includes six tribes based on the banded thisperiod are compatible with the idea that the mono- collenchyma pattern. The subfamily Schismatoglotti- cots arose as a clade from within the dicotyledons, as was doideae includesthe tribesCryptocoryneae (two genera) stated by Dahlgren and Rasmussen (1983). and Schismatoglottideae (seven genera described By 1992, Johri et al. had provided embryological evid- here, now merged to ®ve), followed by the subfamily ence that palmsand cyclanthscannot be closeto Araceae. Table 3. Classi®cation ofBogner and Nicolson (1991) [ Acorus excluded from family.]

ARACEAE

Subfamily Gymnostachydoideae Bogner & Nicolson Gymnostachys Subfamily Pothoideae Pothos, Pedicellarum, Pothoidium Subfamily Monsteroideae Tribe Anadendreae Anadendrum Tribe Monstereae Amydrium, Rhaphidophora, Epipremnum, Scindapsus, Alloschemone, Stenospermation, Rhodospatha, Monstera Tribe Heteropsideae Heteropsis Tribe Spathiphylleae Spathiphyllum, Holochlamys Subfamily Calloideae Calla Subfamily Lasioideae Tribe Orontieae Lysichiton, Symplocarpus, Orontium Tribe Anthurieae Anthurium Tribe Lasieae Subtribe Dracontiinae Cyrtosperma, Lasimorpha, Lasia, Anaphyllum, Anaphyllopsis, Podolasia, Urospatha, Dracontioides, Dracontium Subtribe Pycnospathinae Pycnospatha Tribe Zamioculcadeae Zamioculcas, Gonatopus Tribe Callopsideae Callopsis Tribe Nephthytideae Pseudohydrosme, Anchomanes, Nephthytis, Cercestis Tribe Culcasieae Culcasia Tribe Montrichardieae Montrichardia Subfamily Philodendroideae Tribe Philodendreae Subtribe Homalomeninae Furtadoa, Homalomena Subtribe Schismatoglottidinae Schismatoglottis, Piptospatha, Hottarum, Bucephalandra, Phymatarum, Aridarum, Heteroaridarum Subtribe Philodendrinae Philodendron Tribe Anubiadeae Anubias, Bognera Tribe Aglaonemateae Aglaonema, Aglaodorum Tribe Dieffenbachieae Dieffenbachia Tribe Zantedeschieae Zantedeschia Tribe Typhonodoreae Typhonodorum Tribe Peltandreae Peltandra Subfamily Colocasioideae Tribe Caladieae Xanthosoma, Chlorospatha, Caladium, Scaphispatha, Jasarum Tribe Steudnereae Subtribe Steudnerinae Steudnera, Remusatia, Gonatanthus Subtribe Hapalininae Hapaline Tribe Protareae Protarum Tribe Colocasieae Colocasia, Alocasia Tribe Syngonieae Syngonium Tribe Ariopsideae Ariopsis Subfamily Aroideae Tribe Stylochaetoneae Stylochaeton Tribe Arophyteae Carlephyton, Colletogyne, Arophyton Tribe Spathicarpeae Mangonia, Taccarum, Asterostigma, Gorgonidium, Synandrospadix, Gearum, Spathantheum, Spathicarpa Tribe Zomicarpeae Zomicarpa, Filarum, Zomicarpella, Ulearum Tribe Thomsonieae Amorphophallus, Pseudodracontium Tribe Areae Subtribe Arinae Arum, Dracunculus, Helicodiceros, Theriophonum, Typhonium, Sauromatum, Eminium, Biarum Subtribe Arisarinae Arisarum Subtribe Arisaematinae Arisaema Subtribe Atherurinae Pinellia Subtribe Ambrosininae Ambrosina Subtribe Cryptocoryninae Lagenandra, Cryptocoryne Subfamily Pistioideae Pistia Evolutionary Position of Araceae 25

Table 4. Classi®cation ofMayo, Bogner and Boyce (1997)

ACORACEAE C. Agardh Acorus ARACEAE Juss.

Major Group PROTO-ARACEAE Subfamily Gymnostachydoideae Bogner & Nicolson Gymnostachys Subfamily Orontioideae Mayo, Bogner & P.C.Boyce Orontium, Lysichiton, Symplocarpus Major Group TRUE ARACEAE Subfamily Pothoideae Engl. Tribe Potheae Engl. Pothos, Pedicellarum, Pothoidium Tribe Anthurieae Engl. Anthurium Subfamily Monsteroideae Engl. Tribe Spathiphylleae Engl. Spathiphyllum, Holochlamys Tribe Anadendreae Bogner & French Anadendrum Tribe Heteropsideae Engl. Heteropsis Tribe Monstereae Engl. Amydrium, Rhaphidophora, Epipremnum, Scindapsus, Monstera, Alloschemone, Rhodospatha, Stenospermation Subfamily Lasioideae Engl. Dracontium, Dracontioides, Anaphyllopsis, Pycnospatha, Anaphyllum, Cyrtosperma, Lasimorpha, Podolasia, Lasia, Urospatha Subfamily Calloideae Endl. Calla Subfamily Aroideae PARAPHYLETIC GROUP: PERIGONIATE AROIDEAE Tribe Zamioculcadeae Engl. Zamioculcas, Gonatopus Tribe Stylochaetoneae Schott Stylochaeton MONOPHYLETIC GROUP: APERIOGONIATE AROIDEAE Dieffenbachia Alliance Tribe Dieffenbachieae Engl. Dieffenbachia, Bognera Tribe Spathicarpeae Schott Mangonia Taccarum, Asterostigma, Gordonidium, Synandrospadix, Gearum, Spathantheum, Spathicarpa Philodendron Alliance Tribe Philodendreae Schott Philodendron Tribe Homalomeneae (Schott) M.Hotta Furtadoa, Homalomena Tribe Anubiadeae Engl. Anubias Schismatoglottis Alliance Tribe Schismatoglottideae Nakai Schismatoglottis, Piptospatha, Hottarum, Bucephalandra, Phymatarum, Aridarum, Heteroaridarum Tribe Cryptocoryneae Blume Lagenandra, Cryptocoryne Caladium Alliance Tribe Zomicarpeae Schott Zomicarpa, Zomicarpella, Ulearum, Filarum Tribe Caladieae Schott Scaphispatha, Caladium, Jasarum, Xanthosoma, Chlorospatha, Syngonium, Hapaline No Alliance Tribe Nephthytideae Engl. Nephthytis, Anchomanes, Pseudohydrosme Tribe Aglaonemateae Engl. Aglaonema, Aglaodorum Tribe Culcasieae Engl. Culcasia, Cercestis Tribe Montrichardieae Engl. Montrichardia Tribe Zantedeschieae Engl. Zantedeschia Tribe Callopsideae Engl. Callopsis Tribe Thomsonieae Blume Amorphophallus, Pseudodracontium Tribe Arophyteae Bogner Arophyton, Carlephyton, Colletogyne Tribe Peltandreae Engl. Peltandra, Typhonodorum Tribe Arisareae Dumort. Arisarum Tribe Ambrosineae Schott Ambrosina Tribe Areae L. Arum, Eminium, Dracunculus, Helicodiceros, Theriophonum, Typhonium, Sauromatum, Lazarum, Biarum Tribe Arisaemateae Nakai Pinellia, Arisaema Tribe Colocasieae Engl. Ariopsis, Protarum, Steudnera, Remusatia, Colocasia, Alocasia Tribe Pistieae Blume Pistia Table 5. Classi®cation ofAraceae and notes on certain distinguishing leafand petiole characters (Keating, this volume)

Family ARACEAE SeriesI Bisexual-¯owered aroids Subfamily GYMNOSTACHYDOIDEAE Collenchyma absent; vascular bundles unique, type I; ®bres unique: vascular bundle sheaths and girders. Gymnostachys Subfamily ORONTIOIDEAE Tribe Orontieae Collenchyma absent; ®bres as phloem caps; laticifers non-anastomosing. Orontium Tribe Symplocarpeae Collenchyma types Svc, B; ®bres absent; laticifers absent. Lysichiton, Symplocarpus Subfamily POTHOIDEAE Tribe Potheae Collenchyma absent (type B in Anthurium); vascular bundle type I; ®bres ensheathing bundles; laticifers absent; prismatic crystals. Pothos, Pedicellarum, Pothoidium, Anthurium Tribe Monstereae Collenchyma absent or type B; vascular bundle types I, II; ®bres as phloem or xylem caps or ensheathing bundles; trichosclereids usually present; secretory ducts in most genera; laticifers non-anastomosing, occasional. Holochlamys, Spathiphyllum, Rhodospatha, Stenospermation, Scindapsus, Rhaphidophora, Anadendrum, Monstera, Alloschemone, Epipremnum, Amydrium, Heteropsis Subfamily LASIOIDEAE Collenchyma absent or type B; vascular bundle types I, II; ®bres: caps or ensheathing bundles; midrib or petiole ground tissue type 4. Cyrtosperma, Lasimorpha, Podolasia, Lasia, Anaphyllum, Urospatha, Anaphyllopsis, Pycnospatha, Dracontium, Dracontioides Subfamily CALLOIDEAE Collenchyma absent; vascular bundle types I, II; laticifers non-anastomosing. Calla SeriesII Monoeciousaroids Subfamily PHILODENDROIDEAE Collenchyma type B; vascular bundle types I, II; ®bres as bundle caps; secretory ducts present; laticifers non-anastomosing Tribe Philodendreae Collenchyma type B; vascular bundle types I, II; ®bres ensheathing or as bundle caps; laticifers non-anastomosing. Montrichardia, Anubias, Furtadoa, Philodendron, Homalomena Tribe Zantedeschieae Collenchyma type Bi; vascular bundle types I, II, III. Zantedeschia, Callopsis Tribe Stylochaetoneae Collenchyma type B; vascular bundles type II; ®bres absent; laticifers absent. Stylochaeton Tribe Zamioculcadeae Collenchyma type B; vascular bundles type I, II; ®bres absent or occasional bundle caps; secretory ducts present or absent; laticifers absent. Gonatopus, Zamioculcas Tribe Aglaonemateae Collenchyma types B, Bi, Sb; ®bres absent, or bundle caps, or ensheathing bundles; laticifers non-anastomosing. Nephthytis, Anchomanes, Pseudohydrosme, Aglaonema, Aglaodorum Tribe Culcasieae Collenchyma type B; vascular bundle type I; ®bres absent, ensheathing, or xylem and phloem caps; secretory ducts present; laticifers non-anastomosing (Cercestis). Culcasia, Cercestis Tribe Spathicarpeae Collenchyma types B, Bi; vascular bundle types I, II, III; ®bres ensheathing or bundle caps; laticifers non-anastomosing or absent. Bognera, Dieffenbachia, Spathantheum, Gorgonidium, Syandrospadix, Gearum, Spathicarpa, Asterostigma, Mangonia, Taccarum Subfamily SCHISMATOGLOTTIDOIDEAE Vascular bundle types II, III. Tribe Cryptocoryneae Collenchyma type Sv; laticifers non-anastomosing or absent. Cryptocoryne, Lagenandra Tribe Schismatoglottideae Stomata mostly brachypara-hexacytic or brachypara-octocytic; collenchyma types B, Bi; laticifersnon-anastomosing. Phymatarum, Schismatoglottis, Aridarum, Piptospatha, Hottarum*, Bucephalandra, Heteroaridarum* Subfamily LEMNOIDEAE Collenchyma and ®bres absent; vascular bundle type II; laticifers absent. Spirodela, Landoltia, Lemna, Wolf®a, Wolf®ella Subfamily AROIDEAE Collenchyma type Sv; vascular bundle types II, III; laticifers: 2 types occur. Tribe Thomsonieae Laticifers non-anastomosing. Amorphophallus, Pseudodracontium Tribe Caladieae Laticifers anastomosing. Hapaline, Syngonium, Xanthosoma, Chlorospatha, Ulearum, Filarum, Zomicarpella, Caladium, Scaphispatha, Jasarum, Zomicarpa Tribe Arisareae Laticifers non-anastomosing. Ambrosina, Arisarum Tribe Peltandreae Laticifers non-anastomosing Peltandra, Typhonodorum, Colletogyne, Carlephyton, Arophyton Tribe Pistieae Collenchyma typesSvc, Sv; laticifersabsent. Pistia Tribe Arisaemateae Laticifers non-anastomosing. Arisaema, Pinellia Tribe Areae Laticifers non-anastomosing. Typhonium, Theriophonum, Sauromatum*, Lazarum*, Biarum, Arum, Eminium, Dracunculus, Helicodiceros Tribe Colocasieae Laticifers anastomosing or non-anastomosing. Ariopsis, Alocasia, Remusatia, Colocasia, Steudnera, Protarum

*Genera recently reduced to synonymy but treated separately in this volume Evolutionary Position of Araceae 27

The apparent isolation of the Arales among monocots (1983) shows the Ari¯orae to be sister to the Triuridi¯orae was expressed in Takhtajan (1997) where the families were and Alismati¯orae. They also noted that the lemnoids placed within the derived subclass, Aridae, that included probably arose within ancestral Araceae. orders: Arales, Acorales, Cyclanthales, Pandanales, Typhales, but this placement contradicted his discussion in which he stated that the Arales are closer to the alis- Molecular phylogenetic studies matidsand triurids. The detailed reviewsof monocotsby Dahlgren and Monocot±dicot divergence hasbeen placed by Li et al. Clifford (1982), Dahlgren and Rasmussen (1983), and (1993) at 200 mya, a conclusion supported by analysis of Dahlgren et al. (1985) evaluated the potential of struc- nuclear genesencoding 26S and 18S ribosomalRNAs tural characters in understanding the relationships of the (Wolfe et al. 1989). Studiesby Soltis et al. (1997) using monocot groups. These compendia present valuable 18S rDNA, Chase et al. (1993) and Duvall et al. (1993), arrays of structural and chemical character states, many using the plastid gene encoding for the large subunit of of which can be polarized within families. ribulose, rbcL, concluded that the monocotsare a well- The distinctive and specialized morphology of several supported monophyletic group and that Acorus isthe monocot groups including orchids and grasses suggests `primal extant monocot'. The monocotsare probably thatmonocotevolutionshouldbesubjecttodecipherment. derivedfromwithinthemonosulcateMagnoliidaewiththe But among the lower groups, the main phyletic lines paleoherbs as the immediate sister group. With Saururus remain obscure. Whole suites of characters, certainly asthe dicot outgroup, Acorus emerged as the sister group useful diagnostically, and for which morphoclines can be to all other monocots, and therefore certainly belongs in uncovered within families, are of essentially no use in its own family. The Alismataceae are sister to the Araceae. inferring the plesiomorphic condition of lower monocots, The genus Pleea (Melantheaceae) isbasalto the adjacent or understanding interfamilial relationships. Parallel remaining monocot lines. See also reviews by Chase et al. trends are very common within these orders or super- (2000) and Stevenson et al. (2000). orders. Examples of interesting characters which exhibit In a wider survey of the Magnoliidae, also using the homoplasious development or loss across families rbcL gene, Qiu et al. (1993) agreed that the monocotsare include: growth habits, ranging from vines to bulbs; vari- monophyletic and nested among ranalian orders (see also ations in atactostele structure; leaves varying from simple KaÈ llersjoÈ et al. 1998). Asconcluded by Burger (1977) on phyllomes to large compound structures; supervolute the basis of morphology and anatomy, Piperales were ptyxisin Araceae and many other families. found by Qiu et al. (1993) to be a sister group to the In histology, the occurrence of root-hair short cells is Laurales, monocots, Magnoliales and Nymphaeales. The widespread; vessels are widespread in roots where stems sequence of the six monocot genera sampled, beginning are vesselless; laticifers are scattered among several other- with the long-branch genera, is Pleea, Gymnostachys, wise unrelated monocot families; sieve-element plastids Spathiphyllum, Pistia, Sagittaria and Alisma. of the PIIc type (Behnke 1972, 1995) virtually de®ne the The Angiosperm Phylogeny Group (APG 1998) pro- monocotsasa synapomorphyfor all; oxalate raphides duced an ordinal classi®cation of ¯owering plant families. are common among alismatids, triurids, commelinids but It was based on a compilation of cladistic analyses using are present in all other superorders. They are probably structural and genetic sequence data, and resulted in common in monocot ancestors; perhaps a synapomorphy recognition of 462 familiesin 40 putatively monophyletic for the monocots and lost in some groups. Stomatal orders, and `monophyletic informal higher groups'. In subsidiary cells are virtually unusable as a taxonomic the monophyletic monocots, they began with the mono- character at this time. Rasmussen (1983) has shown that generic Acorales, followed by Alismatales consisting of there islittle correspondencebetween the appearance of the 13 families: Alismataceae, Aponogetonaceae, Araceae a mature stomatal complex and its ontogenetic type in the (including lemnoids), Butomaceae, Cymodoceaceae, Lilii¯orae. Therefore, one cannot use the mature stomatal Hydrocharitaceae, Juncaginaceae, Limnocharitaceae, complex as a basis for phylogenetic comparisons and we Posidoniaceae, Ruppiaceae, Scheuchzeriaceae, do not know the plesiomorphic state of development or To®eldiaceae, Zosteraceae. This order is followed by the mature appearance in the monocots. Intravaginal squa- Asparagales (29 families), Dioscoreales (®ve families), mules (scales in the axils of leaves) are present in Araceae Liliales (nine families), Pandanales (four families), and probably all alismatids. Epicuticular waxes are absent Commelinids(4 ‡ ordersincluding Arecales).At this in alismatids and Ari¯orae but scattered in many orders. time the identi®cation of the nearest dicot relative of these The cladogram offered by Dahlgren and Rasmussen monocot groupsremainsunsettled.It awaitsa combined 28 Introduction

Table 6. Strict consensus tree ofFrench et al. (1995), with the classi®cation ofKeating (2001b) shown on the right. Reproduced by permission ofJ. C. French et al.

Tribes Subfamilies of Araceae Acorus Acoraceae Gymnostachys Gymnostachydoideae 33 13 Lysichiton Symplocarpeae 38 Symplocarpus Orontioideae Orontium Orontieae 3 Amydrium 3 Epipremnum 3 Monstera 4 7 Anadendrum Rhaphidophora 10 Scindapsus Monstereae Pothoideae Stenospermation 26 Rhodospatha 37 16 Spathiphyllum Holochlamys 15 Pothos Anthurium Potheae 3 Dracontioides Dracontium 11 Anaphyllopsis Lasioideae Urospatha 41 Anaphyllum 46 Lasia Cyrtosperma Calla Calloideae 29 Zamioculcas Zamioculcadeae 11 Gonatopus 7 Stylochaeton Stylochaetoneae 9 Zantedeschia Zantedeschieae 14 Callopsis 4 Taccarum 4 Asterostigma 2 Spathicarpa 2 Synandrospadix Spathicarpeae 5 2 Gorgonidium 28 Spathantheum 14 Dieffenbachia Philodendroideae 6 Bognera Anchomanes 6 6 Nephthytis Aglaonemateae Aglaonema Aglaodorum 12 Cercestis Culcasieae Culcasia 13 Furtadoa 4 3 Philodendron Homalomena Philodendreae 3 Anubias 9 Montrichardia 3 Piptospatha 3 Aridarum 24 Schismatoglottideae Schismatoglottis Schismatoglottidoideae 29 Phymatarum 23 Lagenandra Cryptocoryneae Cryptocoryne Lemna Lemnoideae 31 Amorphophallus Thomsonieae 18 Pseudodracontium 13 Hapaline 25 Syngonium 11 5 Xanthosoma Chlorospatha 10 Caladieae 19 9 Ulearum Zomicarpella 8 Caladium Zomicarpa 6 Arophyton 17 Carlephyton 15 2 Colletogyne Peltandreae 4 Typhonodorum Aroideae 12 Peltandra 12 Arisarum Arisareae Ambrosina 1 Helicodiceros 4 Arum 4 Biarum Areae 6 Sauromatum 12 Theriophonum 18 Typhonium 10 6 Pinellia Arisaemateae Arisaema Ariopsis 15 Alocasia 4 Remusatia 10 Colocasieae 22 5 Colocasia Gonatanthus Steudnera Pistia Pistieae Materials and Methods 29 multigenomic and morphological character analysis MATERIALS AND METHODS (Duvall 2000). The consensus, then, of these lines of evidence is that: Approximately 380 specimens representing 105 out of (1) the monocotsare a monophyletic line embedded 106 currently accepted genera of Araceae plus Acorus within the lower dicots; (2) Acorus is sister to all monocots; were available for anatomical study. While most were (3) the Araceae are close enough to the alismatids for prepared from liquid-preserved materials, a few were them to be placed together within a large order; (4) the only available dry and had to be restored. Araceae are distant from the Arecaceae, Pandanaceae Early in this study, vegetative tissue samples were and Cyclanthaceae. prepared as leaf transsections (TS), through the midrib and margin, and oriented to cut secondary or tertiary veins in TS. Petioleswere cut in TS proximally and distally, Structural relationships with the Alismatidae and often in longisection (LS). Stems and roots were cut The thorough structural study of the Helobiae (Alisma- in TS and occasionally in LS. In this phase, section tidae) by Tomlinson (1982) makes possible detailed com- preparation involved conventional paraf®n microtech- parisons. In spite of their being largely aquatic, the nique, including Safranin-O and Fast Green FCF staining alismatid families show many similarities among them- (Berlyn and Miksche 1976; O'Brien and McCully 1981). selves and with the Araceae. In leaf tissue, the cuticle is The most recent third of the specimens were prepared by simple and thin, stomata when present are often para- hand sectioning and wet mounting (Herr 1992; Keating cytic, usually with narrow subsidiary cells. Leaf meso- 1996) and stained using mostly three dyes: iodine- phyll shows bifacial development, and palisade cells potassium iodide (IKI), Toluidine blue-O, and cresyl may have lateral projecting lobesmaking them H- or violet acetate (CVA). The sections were then mounted in U-shaped, a condition found in some genera of Araceae. 30% aqueouscalcium chloride, which rendersthe slides Theaerenchymashowsagenerallylacunosegroundtissue, virtually permanent if they are stored horizontally. common in emergent and ¯oating leaves. Using the A note regarding calcium chloride: thissaltishygro- aerenchyma classi®cation of the present study, the alis- scopic and its ®nal concentration and volume as a liquid matids are mostly type 3 with some development of type 4, slide mountant is dependent on equilibration with atmos- both well represented in terrestrial Araceae. Intravaginal pheric humidity. The salt solution is heavy, highly ionic, squamules are widespread in the alismatids and present and has the refractive index of glass. Because of its ability in a few aroids. Vascular bundle architecture can be very to weaken hydrogen bonds, it stimulates a marked different within monocot groupsbut the Alismatidae metachromatic (polychromic) effect when using CVA typesseemto be homologousin architecture with typesI, and a number of other commonly used aniline dyes, II, and III of the Araceae, the last type being especially especially of the quinone-imine group (Herr 1992; common among alismatids. Tannins are common or Keating 1996). In addition to indicating the presence and uncommon. Oxalate crystals are present as rhomboids, position of starch granules, IKI is polychromic on cel- styloids and occasionally druses in the alismatids. `Lati- lulosic and ligni®ed cell walls. In an excess of iodine cifers' appear to be present in some families although cellulose becomes magenta in a few days, a modi®ed they are described as having 4±6 sheath cells, i.e. they are zinc-chlor-iodide test (Artschwager 1921). Lignin and secretory ducts rather than true laticifer cells. Tieghem cutin become yellow-orange. These preparations are (1872) recorded circlesof secretorycanalsbordered by usually just as photogenic as thinner paraf®n sections 4±10 small cells in Alisma and fewer in Sagittaria; his and are often more informative. description resembled secretory canal structure found in Dried material wasrestoredto itsoriginal hydrated Philodendron or Homalomena. True raphidesof any type dimensions using a dilute surfactant such as Kodak 1 appear to be missing among alismatids. Photo¯o . If more drastic treatment was necessary, The alismatids are highly adapted aquatics in terms of 1±5% sodium hydroxide worked well with the covered habit, structure and ecology (Tomlinson 1982) and very specimen dish placed on a warming tray for 8±24 h and heterobathmic (Takhtajan 1997: 563). While there are removed by inspection. many similarities between the alismatids and aroids, the Photomicrographic imageswere recorded on 35 mm 1 most de®ning feature in this comparison is the presence Kodak Technical Pan ®lm. and elaboration of raphide crystals in the aroids. Fig. 1. Leaf sections. (A) Rhaphidophora decursiva, TS dorsiventral leaf showing marked differentiation of mesophyll into columnar palisade cells and type 1 spongy cells. Stoma and substomatal cavity also present on adaxial surface; (B) Gymnostachys anceps,TS leaf showing mesophyll with no differentiation; (C) Pothoidium lobbianum, TS isobilateral leaf with short palisade cells beneath each epidermal layer; (D) Culcasia rotundifolia, TS dorsiventral leaf showing weak mesophyll differentiation and papillate adaxial epi- dermis; (E) Anthurium veitchii, TS midrib showing broadly convex adaxial pro®le with single well-de®ned narrow angular ridge, and broadly and deeply rounded-convex abaxial surface; (F) Amorphophallus hildebrandtii, leaf segment TS midrib, narrowly grooved adaxially and irregularly semicircular abaxially. Scale bars: A,B,F ˆ 200 mm; C ˆ 50 mm; D ˆ 150 mm; E ˆ 300 mm. Fig. 11. Acorus leaf, rhizome and root sections. (A,B,D,E,G,H) A. calamus; (C,F,I) A. gramineus; (A) Lamina TS distal; (B) Lamina TS basal, above fusion of basal groove; (C) Lamina TS basal groove. Most chlorenchyma restricted to darkened peripheral portions of leaf; (D) Lamina PS showing epidermis and brachyparacytic stomata; (E) Midrib TS with largest vascular bundle; (F) Lamina TS lateral margin and vascular bundle; (G) Lamina showing oil cell (arrow o), stoma (arrow), and transition from peripheral chlor- enchyma to aerenchyma; (H) Rhizome TS showing amphivasal bundles of central cylinder and endodermis (arrow); (I) Root TS showing stele. Scale bars: A,B,F ˆ 150 mm; C ˆ 300 mm; D,G±I ˆ 50 mm; E ˆ 200 mm.