BOOK-ARACEAE

NH2 HOOC O Gluc C OOH HOOC C HO 1 CN I II C O2 OH O Gluc NH2 C OOH C OOH HO + 1_ O2 OH OH 2 VII IV V O 2 CH3 COOH N VI O NH2 HO NH2 O CO2 C OOH HO HO III VIII HO IX RO 3 1 CHO RO 4 OH 1 XII HO 3 5 8’ 14’ 1’ 17’ Y COOH XIV 11’ Me HO B- ring Y XI OH A-r in g PA L R O7 8 1B A O Gluc 6 O OH XV CH3 COOH s everal ty pes NH2 or of amid es C OOH C H2 X HOOC COOH XIII Figure 8. Some general lines of secondary metabolism of Araceae and some of their key chemical characters I-IX = Tyrosine and some of its metabolites: I = Tyrosine; II = cyanogenic glucoside Triglochinin; III = oxoaporphine Liriodenine; IV and V = Homogentisic acid and its 2-glucoside which are responsible for the so-called egumi-taste according to Hasegawa et al. (1959); VI = Acetic acid generated by total catabolism of tyrosine; VII = Tyramine; VIII = DOPA; IX = Dopamine, one of the possible causes of melanogenesis in dying parts of araceous plants; X-XII = presumably metabolites of phenylalanine:– X = Phenylalanine; XI = p-Coumaric (Y = H), Caffeic (Y = H, OH), Ferulic (Y = H, OMe) and Sinapic (Y = OMe) acids; XII = 3,4-Dihydroxybenzaldehyde (R=H) and its diglucoside, 3,4-Diglucosyloxybenzaldehyde (R = Glucosyl), which are responsible, at least partially, for “Hange” and “Taro” acridity according to Suzuki et al. (1975); XIII and XIV = pure Acetogenins:– XIII = Acetic or Malonic acids (activated forms); XIV = main allergen of Philodendron, 5-Heptadec-8’,11’,14’-trienylresorcinol; XV = the 6-C-glucoflavones Isovitexin (= Saponaretin: R = H) and Saponarin (R = Glucosyl); flavonoids are of mixed biogenetic origin. 1 and 2 = enzymes of tyrosine catabolism; 2 = “Homogentisicase” of Hasegawa et al. (1959). PAL = Phenylalanine lyase ® = alternative pathways indicated. P H Y T O C H E M I S T R Y A N D C H E M O T A X O N O M Y 39

four tested samples of an Anthurium species and four pounds still await definitive identification. It is rather out of six samples of Philodendron corcovadense were unexpected that no new AGIs were detected in cyanogenic) and Gibbs (1974: cyanogenesis was Araceae; their three main polyhydroxy alkaloids had demonstrated for Anthurium aemulum, A. scandens, formerly been isolated from other taxa, e.g. Calla palustris). I could not confirm the latter for a Leguminosae and Euphorbiaceae. Obviously the syn- Dutch sample of this species, Schismatoglottis sp. (“S. thesis of such sugar- and/or pipecolic acid-related ruttenii”), Xanthosoma lindenii (= Caladium lindenii) compounds is a common feature of a large number of or Zantedeschia rehmannii. In the case of Gibbs plants and microorganisms, but their accumulation is of (1974), only personal observations are taken into con- much more restricted occurrence. sideration because of his rather uncritical evaluation of results published by other investigators; he accepted Irritants in Araceae without comment many highly doubtful cases of cyanogenesis reported in the literature. Peckolt (1893; see Hegnauer 1963) reported long ago that several types of irritants must occur in Brazilian Triglochinin, a seco-derivative of dhurrin or taxi- Araceae. A comprehensive summary of irritant aroids phyllin, is the cyanogenic glucoside occurring in all and their irritating properties is given by Mitchell & Araceae so far investigated for cyanogenesis. It is Rook (1979); see also Hegnauer (1963, 1986), and Bown accompanied by a substrate-specific enzyme (“triglo- (1988). Brazilian scientists have paid much attention to chininase”) which rapidly splits the glucoside after the irritating and toxic properties of Araceae. Lethal injury. Cyanogenesis is therefore often extremely rapid intoxications of children from eating the spathe and in Araceae and in many instances the HCN is totally spadix of cultivated Zantedeschia aethiopica have been lost during the drying of plant parts. Triglochinin and reported (Ladeira et al. 1975). Working with juices the corresponding enzyme were shown to be the expressed from different parts of Dieffenbachia picta (= cause of cyanogenesis in Alocasia macrorrhizos (tribe D. maculata) it could be demonstrated that stems and Colocasieae), Arum maculatum (tribe Areae), Pinellia petioles are much more aggressive than leaf blades and tripartita (tribe Arisaemateae), Lasia spinosa (sub- the centrifugation of the stem and petiole juices resulted family Lasioideae) and Dieffenbachia picta (= D. in an inoffensive supernatant and a highly offensive maculata, tribe Dieffenbachieae) by Nahrstedt and his precipitate. Moreover, it was shown that the toxic prin- group (see Hegnauer 1986). ciple(s) of this species is (are) not of proteinaceous nature and neither stable to prolonged heating nor to Polyhydroxy alkaloids or alkaloidal vacuum drying (Ladeira et al. 1975; see also Carneiro glycosidase inhibitors (AGIs) 1985). Later these observations were confirmed and raphides present in the precipitates were investigated; This class of biologically active secondary metabo- they lost their irritating and toxic properties on washing lites has been shown in recent years to comprise the with ether, but not on washing with water (Jesus Neves toxic principles of Locoweeds (Astragalus, Oxytropis), et al. 1988). Dieffenbachia maculata contains an Darling Peas (Swainsona) and Moreton Bay Chestnut inhibitor of human salivary amylase and several other or Black Bean (Castanospermum australe), all belong- types of amylases (Padmanabhan & Shastri 1990). It is, ing to the Leguminosae. The same and similar however, improbable that such inhibitors are involved compounds occur also in some fungi, a fern, several in Dieffenbachia toxicity. Euphorbiaceae, Moraceae and Polygonaceae, but had not been reported previously from monocotyledonous Contact dermatitis plants. A research group at Kew (Sharp et al. 1993) has Contact dermatitis is caused by some species of been screening Araceae for this type of metabolite. So far about 70 species from 47 genera have been inves- Philodendron. Their main allergens have been iso- tigated, and appreciable amounts of AGIs were lated and identified as alkenylresorcinols with one, detected in leaves of several genera of subfamily two or three double bonds; they are accompanied by Aroideae: Anchomanes, Nephthytis and corresponding tridecyl-, pentadecyl- and heptade- Pseudohydrosme (tribe Nephthytideae) and Aglaonema cylresorcinols. Philodendron scandens subsp. and Aglaodorum (tribe Aglaonemateae). Trace amounts oxycardium is the most noxious taxon investigated of AGIs were present in species of Amorphophallus hitherto and may contain much 5-pentadecatrienylre- (tribe Thomsonieae). Accumulation of AGIs thus sorcinol. Philodendron angustisectum, P. erubescens appears to support the present circumscription of and P. radiatum are also suspected of being able to Nephthytideae and Aglaonemateae. The major araceous cause contact dermatitis because mono- and di-unsat- AGIs were identified as DMDP, a 2,5-dihydroxymethyl- urated alkenylresorcinols have been detected in them. 3,4-dihydroxypyrrolidine, an N-analogon of On the other hand, no alkyl- nor alkenylresorcinols fructofuranose and two piperidine derivatives, HNJ were observed in P. bipennifolium, P. fenzlii, P. sagit- (2,6-dihydroxymethyl-3,4,5-trihydroxypiperidine) and tifolium, P. squamiferum or P. tuxtlanum (Reffstrup et DMJ (= deoxymannojirimycine). Several minor com- al. 1982, Reffstrup & Boll 1985). 40 T H E G E N E R A O F A R A C E A E

Non-allergenic skin irritations metabolite of tyrosine, and its 2-glucoside were shown Non-allergenic skin irritations can occur in persons to be responsible for this taste sensation. It should not be forgotten, however, that large amounts of soluble handling large quantities of horticultural Araceae. They oxalates as well as certain proanthocyanidins can also may be caused by the combination of raphides and produce taste sensations of a tart to slightly bitter, papain-like proteolytic enzymes like the so-called astringent or acrid nature. Moreover, egumi-taste grad- dumbcain described for taxa of Dieffenbachia. This ually passes into taro acridity (Hasegawa et al. 1959). type of skin lesion is best known from workers in the pineapple industry (Ananas comosus, Bromeliaceae). Irritation of mucous membranes Biogenic amines and alkaloids Painful irritations of mucous membranes are caused Biogenic amines and alkaloids occur in many by many araceous plants. Long ago, Chauliaguet (1897; Araceae. True alkaloids seem to be rare but perhaps see Hegnauer 1963) showed that raphides alone are rather significant from the taxonomic standpoint harmless. Notwithstanding this fact, raphides are still because they indicate biochemical relationships with believed to be the main irritating factor of Araceae by Magnoliidae. The aporphine-type alkaloids liriodenine many modern authors who are not familiar with the and lysicamine were isolated from roots of Lysichiton pertinent literature. Raphides, however, are only the camtschatcensis. Tubers of Pinellia ternata yielded vehicles of irritating substances. The construction and ephedrine. A tertiary base, C20H35O2N, was isolated contents of araceous raphide idioblasts (Wiley 1903; from tubers of Eminium spiculatum (Ahmed et al. Safford 1905; Middendorf 1983; Tang & Sakai 1983) 1968). Traces of volatile biogenic amines, but no coni- and the structure of the individual needles (Sakai et al. ine, were shown to be present in tubers and leaves of 1972; Tang & Sakai 1983) suggest that they are adapted Arisarum vulgare, Arum maculatum (here also traces to transport acrid, pain-producing and otherwise irri- of nicotine were present) and Eminium spiculatum. tating substances. According to Suzuki (1969, 1975), 3,4-dihydroxybenzaldehyde (protocatechualdehyde) Just before and during flowering, spadices and and its 3,4-bisglucoside are acrid constituents of tubers spathes of a number of Araceae produce large amounts of Pinellia ternata and Colocasia antiquorum (= C. of volatile amines and indoles, giving them a putrescent esculenta). Presumably the sulphated and acidic deriv- odour attractive to their pollinators. The composition of atives of caffeic acid mentioned earlier also take part these “perfumes”, which may even contain skatol, is in araceous acridity. taxon-dependent (Hegnauer 1986; Bown 1988). The temperature of the upper part of the spadix rises It is still uncertain whether specific pain-producing considerably just before anthesis; this increases volatiliza- substances are also involved in painful irritations of tion of amines. The substance triggering heat production mucous membranes caused by Araceae. Pain-producing in the spadix is known as “calorigen” and was shown oligopeptides like moroidin of Laportea moroides in the recently to be salicylic acid in Sauromatum guttatum (= Urticaceae (Leung et al. 1986) may be part of the “dart” S. venosum) by Raskin et al. (1987). Spathes and male poison of some Araceae. Saponins may also be involved and female flowers of many Araceae also produce large in the acridity of aroid species which produce and store amounts of non-volatile amides of p-coumaric and fer- these substances; it is known that several saponins (for- ulic acids with putrescine, spermidine, tyramine and, in merly called sapotoxins) are highly irritating to mucous Zantedeschia aethiopica, spermine. Large amounts of membranes. According to Tang & Sakai (1983), Suzuki’s free tyramine and (or) dopamine were present in these (1969, 1975) diglucosyloxybenzaldehyde was a misiden- parts of Monstera deliciosa, Philodendron andreanum tification; they assumed that in fact the isolated (= P. melanochrysum), P. erubescens, P. martianum, P. compound was 5-(hydroxymethyl)-furfural, a decom- scandens, P. selloum (= P. bipinnatifidum), P. triparti- position product of some hexoses under acidic tum, Remusatia vivipara, Rhaphidophora decursiva and conditions. Notwithstanding the fact that the reports of Zantedeschia aethiopica, but not in Arisarum vulgare, Suzuki (1969) and Suzuki et al. (1975) contain inaccu- Arum maculatum, A. italicum, or Dracunculus vulgaris. racies, it is clear that Tang & Sakai failed to read carefully Up to 4 mg dopamine/g fr. wt were observed in one Suzuki’s (1969) paper and it is highly improbable that month-old ovaries of Monstera deliciosa (Ponchet et al. their acrid bisglucoside C19H26O13, was a derivative of 5- 1982). The tendency for spathes and other plant parts to (hydroxymethyl)-furfural. turn black (melanogenesis) in a number of Araceae may be due to the simultaneous presence of dopamine and Egumi taste phenol oxidases. Another effect of Araceae on mucous membranes The starch-rich tubers of North African and was described by Hasegawa et al. (1959) for dried Mediterranean Arisarum vulgare are known in tubers of Pinellia ternata (“Hange”). They have a Morocco as “Irni” or “Erni”. This name is used, how- harsh, somewhat bitter and astringent taste similar to ever, for several taxa of subfamily Aroideae with the so-called egumi-taste of edible shoots of some cul- starch-rich edible tubers which also contain toxins, and tivars of Phyllostachys edulis (Gramineae subfamily are suitable as food only after adequate treatments Bambusoideae). In both cases homogentisic acid, a P H Y T O C H E M I S T R Y A N D C H E M O T A X O N O M Y 41

such as heating and drying or repeated cooking etc. odoriferous compounds, which are synthesized by (Bellakhdar 1978). A pyrrolidine alkaloid called irniine, spathes and spadices of Peltandra and which func- C20H33N, was isolated from tubers of Arisarum vulgare tion as a lure for the pollinator (Patt et al. 1992). and shown to be one of the toxic principles of this species (Melhaoui et al. 1992). Notwithstanding the toxicity of asarone to many insects, it seems to have become, together with other Miscellaneous compounds volatile phenylpropanoids, asarylaldehyde and the sesquiterpenoid acarogermacrone, a trigger for certain Essential oils behavioural activities of fruit flies of the genera Ceratitis and Dacus (Jacobson et al. 1976). Larvae of fruit flies Acorus (Acoraceae), with two species, A. calamus are often serious orchard pests and substances attract- and A. gramineus, is a highly aromatic taxon. The roots, ing males and (or) females may become useful as lures rhizomes and leaves produce large amounts of essen- for capturing adults of pest-causing insects. An exam- tial oils, which are stored in idioblasts similar to the oil ple of such a use is Spathiphyllum cannifolium, which cells of woody polycarps (Magnolianae sensu Takhtajan is planted in Northern Thailand around orchards. Its 1959), aromatic Gramineae and Zingiberaceae. Both flowering inflorescences do not emanate foetid amines, species of Acorus comprise several chemodemes with but a mixture of benzyl acetate, methyleugenol, respect to the composition of essential oils. Best known methylchavicol, p-methoxybenzaldehyde, propyltet- are the taxa which produce the biologically active radecanate and yet other compounds, which are highly phenylpropanoid cis-asarone (= ß-asarone), which has attractive to several species of fruit flies; Lewis et al. insecticidal and antifeedant properties (Koul et al. 1990). (1988) report the same attraction for the fruit fly Dacus This compound occurs together with the less toxic musae in Northern Queensland. trans-asarone, mainly in triploid European A. calamus and in much larger quantities in Indian tetraploids Another fascinating observation was published by (Hegnauer 1963, 1986; Röst 1979b). Rhizomes of Indian Seidel et al. (1990) as follows. Anthurium gracile, A. A. calamus also contain the dimethylbutanoid lignan ernestii and Philodendron megalophyllum belong to acoradin (a dimer of cis-asarone, see Saxena & the so-called ant garden plants found in lowland Mukherjee 1985), as well as asar(yl)aldehyde, acora- Amazonian Peru. Workers of the ant Camponotus mone, and galangin; for mono- and sesquiterpenoid femoratus collect seeds of these and some other, tax- constituents of Acorus oils, the reader is referred to onomically unrelated plants and store them in brood Röst (1979b), Röst & Bos (1979) and Hegnauer (1986). chambers where they later germinate. Volatile aromatic Stereoisomers of acoradin are andamanicin from Piper compounds may be cues which initiate collection of sumatranum var. andamanicum, heterotropan from the seeds of these epiphytic plants by ants. In the case Asarum taxa and magnosalin from Magnolia taxa of the aroid species, 6-methyl methyl salicylate, ben- (Malhotra et al. 1990). A chemotype of Acorus zothiazole and vanillin were detected in seed coats gramineus cultivated in Italy accumulates a number of and adhering fruit parts. phenylpropanoids, including ß-asarone and its epoxide, which suppress growth of several microalgae (Della Phytosterols and triterpenes Greca et al. 1989). Sitosterol-type phytosterins are ubiquitous in vas- Another aromatic genus of Araceae is Homalomena cular plants. ß-sitosterol palmitate was isolated from (Wealth of India 1959; Bown 1988: 223). H. aromatica tubers of Amorphophallus campanulatus (= A. yielded 1.2% essential oil containing mainly monoter- paeoniifolius) together with other phytosterols, tria- penoids (Hegnauer 1986). Rhizomes of Homalomena contane, lupeol and betulinic acid by Chawla & aromatica contain an essential oil with up to 80% of Chibber (1976). New di- and trioxigenated sterols linalool and several sesquiterpenoids of which the have been isolated from tubers of Colocasia escu- homalomenols C and D are new compounds with the lenta (Ali 1991) and from Pistia stratiotes (Aliotta et rather rare carbon skeleton of mintsulphide and the al. 1991; Monaco & Previtera 1991). aphanomols (Sung et al. 1992). Rhizomes of Homalomena occulta, a Chinese medicinal crude drug, As far as I am aware, there exist only a few other yielded 0.79% of essential oil with much linalool and reports concerning the occurrence of triterpenes in lesser amounts of other monoterpenoids and sesquiter- Araceae. Taraxerol acetate was isolated together with penoids (Zhou et al. 1991). Pistia stratiotes produces phytosterins, lignoceric acid, hentriacontane, hentria- similar allelochemicals, one of which was shown to be contanol and hentriacontanone from rhizomatous asarone (Aliotta et al. 1991). Peltandra virginica is pol- stems of Alocasia fornicata (Sharma et al. 1972). linated by the chloropid fly Elachiptera formosa, which also breeds in the inflorescences of this plant. This Air-dried aerial parts of Xanthosoma robustum col- peculiar symbiosis between a plant species and its pol- lected in Oaxaca, Mexico, yielded four antibacterial linating insect seems to depend, at least in part, on hydroperoxy derivatives of 4,14-dimethyl-cholesta-8- en-3ß-ol and of cycloartenol (Kato et al. 1996). 42 T H E G E N E R A O F A R A C E A E

Epicuticular waxes family are metabolites derived from tyrosine (triglo- C chinin and tyramine, dopamine and their amides; Behnke & Barthlott (1983), Fröhlich & Barthlott egumi-taste principles (Hasegawa et al. 1959) and the (1988) and Barthlott (1990) showed that the pres- presumably related, acrid constituents). However, the ence and architecture of epicuticular waxes is a participation of tyrosine in the production of sec- character useful for the classification of mono- ondary metabolites such as the cyanogenic glucosides cotyledons. Their so-called Strelitzia-type and dhurrin, taxiphyllin and triglochinin (Hegnauer 1973, Convallaria-type seem to be restricted to monocots 1977, 1986: 356–357) and the amaryllidaceous and and to characterize major taxa, i.e. either parts of the colchicinoid types of alkaloids (Hegnauer 1986: 316, Zingiberiflorae, Commeliniflorae, Areciflorae and 576), is widely scattered in monocotyledons. Bromeliiflorae, or parts of the Liliiflorae. The Ariflorae, like the Alismatiflorae, Triuridiflorae and Ultimately, one point cannot be overemphasized. Dioscoreales (Liliiflorae), lack these characteristic Use of chemical characters for taxonomic purposes types of epicuticular waxes (compare Dahlgren et can only be meaningful if all available facts are al. 1985: 65, 96, 98–99). checked and evaluated carefully. Authors who neglect to exploit information made readily available in hand- 2. Chemotaxonomy books (Wehmer 1929, 1931, 1935; Hegnauer 1963, 1986 (for monocotyledons)) do not do a good job as far as It is advisable not to put too much weight on the chemotaxonomic part of their work is concerned. chemical characters with regard to estimating phylo- Furthermore, the need to consult as many original genetic relationships of the Araceae as long as the papers as possible can be illustrated by two curious chemical structures of araceous saponins remain examples. First, the cause of the painful acridity of unknown. As stated here and in an earlier publication most araceous plants is still ascribed solely to the pres- (Hegnauer 1986), there are similarities between the ence of raphides by many authors. Secondly, the primary and secondary metabolism of the Araceae summary in Hegnauer (1986), based on the original and a rather large number of other monocotyledo- paper, of the results of the phenolic research carried nous taxa. The most striking chemical attributes of the out by Williams et al. (1981) is actually more accurate than a summary later given by the authors themselves (Williams & Harborne 1988). P H Y T O C H E M I S T R Y A N D C H E M O T A X O N O M Y 43

C 12 E C O L O G Y A N D L I F E F O R M S Croat (1990, 1992a) has published a more detailed this habit. Many other genera, however, seem to rep- review of araceous life forms and ecology. resent relatively small adaptive shifts from a mesophytic norm. Within subfamily Lasioideae for example, Lasia, The growth of Araceae is dependent on abundant Podolasia and many species of Cyrtosperma have available water and prevailing atmospheric humidity. habits which are intermediate between the helo- Structurally and physiologically they are not well phytic/mesophytic and geophytic/mesophytic adapted for growth in arid or cold conditions, and hence categories. Nephthytis, in which the rhizomes normally do not occur in the most extreme environments. grow superficially, has a considerably more meso- phytic habit than the strongly tuberous stems of the Araceae are most diverse and abundant in the other two genera of tribe Nephthytideae. Culcasia has humid tropics and it is there that the richest variety of many terrestrial species, spanning the hemiepi- their life forms is found. Relatively few genera inhabit phytic/mesophytic categories, and Philodendron is temperate regions of the world and these are either similar. Anubias is predominantly helophytic but geophytes (e.g. Arisaema, Arum, Pinellia) or helo- Dieffenbachia and Spathiphyllum, while typical of phytes (Calla, Lysichiton, Orontium, Peltandra, wetter habitats, also occur on drier ground within a Symplocarpus, ). humid tropical habitat. In the predominantly geophytic tribes Caladieae and Colocasieae, the genera Alocasia, The very few genera found at high altitudes exist Colocasia and Xanthosoma have mesophytic species in a warm temperate climatic regime and are also with decumbent to erect, arborescent stems, while geophytic. Gorgonidium has been found at around Steudnera and Chlorospatha are exclusively of this 3000m in the Andes, while Arisaema occurs at up to type. The most primitive Araceae, subfamilies 3000m in Africa (A. ruwenzoricum in the Ruwenzori) Gymnostachydoideae and Orontioideae, are geo- and 4400–4500m in the Himalaya (A. flavum, A. phytes, rhizomatous helophytes or aquatics, and jacquemontii, A. lobatum). largely extratropical. While their habits are doubtless a prerequisite for survival in a more demanding cli- Taxonomic considerations mate, and therefore could have evolved from a mesophytic common ancestor, it is nevetherless From the taxonomic and evolutionary viewpoints equally possible that the mesophytic habit has evolved there are some generalizations that can be made con- various times within the more advanced subfamilies cerning life forms. Hemiepiphytes are commonest in from geophytic or helophytic ancestors. the more primitive tribes and subfamilies. Most genera of subfamilies Pothoideae and Monsteroideae are The geophytic habit is strongly represented in the hemiepiphytes and among more advanced genera this relatively primitive subfamily Lasioideae and particularly life form occurs only in tribe Culcasieae, Philodendron common in the most advanced subfamily Aroideae. and Syngonium, all belonging to subfamily Aroideae. These latter genera show marked structural adapta- The rheophytic habit is characteristic of tribe tions in their habit and in these features must be Schismatoglottideae, the genera being almost exclu- considered derived. Aquatic, subaquatic and helophytic sively rheophytic except for Schismatoglottis, which genera are scattered throughout the family from very consists mainly of terrestrial mesophytic herbs. primitive groups such as subfamily Orontioideae to very advanced ones such as tribe Cryptocoryneae and Hemiepiphytes Pistia in subfamily Aroideae. Humid tropical forests are the characteristic habi- The least specialized mesophytic habit is shown by tat of hemiepiphytic genera. The species vary some rainforest terrestrial herbs. In these the stem is considerably in size, from shortly climbing plants aerial and erect or decumbent, with short but distinct found on the major branches or trunks of trees (some green internodes. This habit type has been judged Anthurium species) to huge plants with attached primitive in the family by some previous authors (e.g. stems growing high into the forest canopy and pro- Grayum 1990), but in fact is found predominantly in ducing enormously long, pendent flowering stems more advanced genera. Typical examples are (e.g. Philodendron scandens). Hemiepiphytes can be Aglaonema (tribe Aglaonemateae), Dieffenbachia (tribe divided into primary and secondary hemiepiphytes. Dieffenbachieae), Homalomena (tribe Homalomeneae) Primary hemiepiphytes begin growth above ground and Schismatoglottis (tribe Schismatoglottideae), all of level but produce feeder roots which eventually grow subfamily Aroideae. Among primitive groups only tribe Spathiphylleae and terrestrial Anthurium species have 44 T H E G E N E R A O F A R A C E A E

down to the forest floor. Secondary hemiepiphytes Philodendron species (e.g. P. insigne) are litter basket germinate on the forest floor, grow up tree boles, epiphytes. The large leaves form an inverted cone in become detached from the ground by rotting of the which leaf litter and other debris accumulate and into juvenile stem but then become reconnected later by which the roots grow and ramify in a dense mass. feeder roots which grow down from the upper inter- Remusatia vivipara, which has a tuberous stem, is a nodes. Hemiepiphytic aroids typically have anchor widespread epiphyte, due to the dispersal of hooked roots as well, and are thus often called “root climbers”. bulbils which are probably transported by birds and primates high in the forest canopy. Arophyton buchetii Flagelliform shoots, heteroblastic leaf development appears to occur only as an epiphyte in leaf litter accu- and shingle plants (see chapter 2) are characteristic fea- mulated within large Pandanus crowns. tures of hemiepiphytic Araceae, though not present in all species of each genus. Highly developed heteroblasty Lithophytes coupled with skototropism, a specific growth strategy for seeking host tree boles, has been described in Many hemiepiphytes, epiphytes and geophytes are Monstera (Madison 1977a, Strong & Ray 1975). In certain also found as lithophytes in suitable conditions. Certain species the seedling is a very slender, non-photosyn- groups, such as the Anthurium coriaceum complex in thetic plant with long internodes and minute scale leaves. eastern Brazil, are characteristically lithophytic. Different Having germinated on the forest floor it seeks the defined species of this complex grow in humid coastal forests area of shadow represented by the nearest tree bole. (e.g. A. coriaceum) where they are common on exposed Once the tree has been reached the plant transforms areas of outcropping rocks, and in the semi-arid interior itself into the shingle form and later, higher up, into a (A. erskinei), where they survive exposure during the mature flowering plant. Vegetative reproduction may prolonged dry season. Vining hemiepiphytes frequently then take place by the production of flagelliform shoots. grow on rocks in forest regions wherever shade and Seed size is almost certainly an important element in the humidity are sufficient, the rock surface providing much growth strategies adopted by hemiepiphytes. In Monstera the same conditions for attachment as tree boles. seeds are relatively large and lack endosperm, which probably increases the efficiency and duration of the A number of geophytes are characteristically found seedling’s nutrient supply until it has reached a suitable growing in the eroded, litter- or humus-filled cavities habitat for photosynthesis. Other hemiepiphytes which of limestone outcrops; examples are Amorphophallus produce flagelliform shoots have numerous very small albispathus, Colocasia gigantea, and Typhonium albis- seeds and endosperm (e.g. Philodendron fragrantissi- pathum in S.E. Asia and Amorphophallus hildebrandtii mum, Rhodospatha latifolia) and probably have a Carlephyton and Colletogyne in Madagascar. different kind of seedling development. More observa- Rheophytes are also typically lithophytic. tions are needed, especially in tropical Asia. To date the most important ecological observations of hemiepiphytic Geophytes aroids have been made in tropical America (Blanc 1977a, b, 1978, 1980; Madison 1977a, Ray 1986, 1987a–c, 1988, This category includes all genera with tuberous, 1990) and tropical Africa (Knecht 1983). rhizomatous, subterranean or partly subterranean stems. Geophytic aroids characteristically have peri- Cercestis (e.g. C. mirabilis) and Philodendron (e.g. P. odic dormant periods when no leaves are present linnaei), among other genera, have species with another and these normally correspond to the dry season (or growth strategy aptly termed “rhythmic growth” by Blanc winter) of their habitat. However, rainforest geophytes (e.g. 1977a). The mature flowering region of the stem is exhibit growth periodicity and dormancy even in non- short with abbreviated internodes and more-or-less rosu- seasonal climates, e.g. Amorphophallus maculatus, late foliage leaves. The continuation shoot climbs A. titanum, Asterostigma riedelianum, Dracontium upwards and is slender and flagelliform with cataphylls prancei, Zomicarpella amazonica. instead of foliage leaves. After an interval it produces another rosulate-leaved mature zone. The repetition of Several genera occur in more than one kind of cli- this pattern produces a series of connected rosulate matic regime. In Stylochaeton, the rainforest species S. plants one above the other on a single tree trunk. zenkeri is evergreen with unthickened roots and the inflorescence appears with the leaves. Other species, Epiphytes such as S. natalensis, grow in areas with a strongly marked dry season during which they are dormant. True epiphytes, which never become connected to This species has thick, fleshy roots and usually flow- the ground by feeder roots, are found in Anthurium, ers before the leaves or just as they emerge. The Arophyton, Philodendron, Remusatia, Scindapsus and genera Amorphophallus and Dracontium are similarly Stenospermation. The seeds presumably germinate diverse ecologically, with species in rainforest or in directly on the host tree after dispersal by birds or seasonal evergreen forest, deciduous forest, savannas other animals. Many species of Anthurium sect. or grasslands (A. abyssinicus, D. margaretae). Pachyneurium (e.g. A. hookeri) and some E C O L O G Y A N D L I F E F O R M S 45

Geophytes from deciduous forests, savannas or species generally look very different in shape, size, colour C strongly seasonal grasslands flower without the leaves and structure. Submerged leaves are softer and emergent at the end of the dry season, mostly after the first rains ones more coriaceous. Many species occur in the fresh- fall. Leaf and fruit development take place during the water tidal zone where there is a daily cycle of exposure rainy season. Variations occur in this basic phenologi- and submersion. Some species are found only in fresh- cal pattern. Biarum davisii (Crete and Turkey) flowers water, like C. ferruginea, C. lingua and C. pontederiifolia, in the autumn (± November) after the rainy season while others can grow both in fresh and brackish water has started, whereas B. ditschianum (Turkey) flowers (C. ciliata). A few species are helophytes, preferring at the end of the rainy season (± May) and the fruit swampy conditions and growing during the dry season development then takes place over a year (Bogner & completely emergent in normal soil, like C. spiralis, a Boyce 1989). The Mediterranean species of Arum (e.g. weed of rice fields in India. Usually Cryptocoryne species A. dioscoridis, A. italicum) grow during the relatively flower at low water level when the plants become emer- warm winter rainy season, whereas the more northerly gent. Cryptocoryne consobrina occupies a somewhat A. maculatum grows from spring to summer and is more specialized niche. The leaves are present only dur- dormant during the cold winter (Boyce 1993a). ing the monsoon season when the streams are in flood. Flowering occurs after the monsoon rains when the No Araceae occur in true deserts except Eminium streams have dried up and the leaves are shrivelled. spiculatum subsp. negevense, from the Negev desert During the dry season the plants are completely dormant (Koach 1988). Some species, however, grow in very dry with the rhizomes buried in the soil. C. nevillii flowers areas, e.g. Arum and Eminium in central Asia, Arum and at the beginning of the monsoon season before the Biarum in North Africa and Asia Minor, Arisaema and leaves appear or as they emerge. Sauromatum venosum in the Arabian Peninsula and East Africa and a few Stylochaeton species of the Sahel Helophytes zone of Africa. All these regions normally have some rain each year during which the plants grow vegeta- About 38 genera are helophytic or have at least tively, or they may occur in places with a ground water some helophytic species, i.e. plants which grow in supply. Zamioculcas zamiifolia is a succulent plant swampy habitats or along river and stream margins. which stores water in its thick petioles and is some- Four of these genera (Gearum, Mangonia, times found in very dry habitats, but it is more common Scaphispatha, Spathicarpa) are geophytes which habit- in evergreen seasonal forests and savannas. ually or frequently grow in seasonally flooded sites, Nineteen genera are strictly helophytes or are aquatics Rheophytes with some helophytic species (e.g. Cryptocoryne). Rheophytes are flood-resistant plants, usually of trop- This life form is thus widespread throughout the ical rainforests, growing in or along swift-running streams family in many different taxonomic groups, and in both or rivers up to the flood level. They are characterized by temperate and tropical genera. There is little constancy narrow, leathery leaves and a firmly attached, usually in habit type. The stem may by tuberous (e.g. Caladium, epilithic stem. In addition to tribe Schismatoglottideae, in Typhonium), rhizomatous (Typhonodorum), rhizoma- which the majority of genera have this habit, rheo- tous and arborescent (Montrichardia), semi-prostrate phytes are also found in Homalomena, Anubias and to aerial (Lasia), erect and arborescent (Philodendron) Holochlamys, and rarely in Anthurium. or merely shortly erect and aerial (Homalomena). Submerged or periodically submerged aquatics The helophytic life form may be considered relatively unspecialized in the majority of genera which exhibit it. Jasarum steyermarkii and many Cryptocoryne species Tuberous or rhizomatous stems may be associated with are permanently submerged plants (hydrophytes, sensu seasonally flooded habitats and a marked dry season. Cook 1990). Either the inflorescence as a whole Rhizomes may, on the other hand, be adaptations for col- (Jasarum) or its upper portion (Cryptocoryne) is held onizing muddy riverine margins as in the case of the above the water surface while all other parts are com- strict helophytes Typhonodorum and Montrichardia. pletely submerged. Cryptocoryne is the largest genus of Genera such as Dieffenbachia, Homalomena and aquatic aroids and merits more detailed consideration. Spathiphyllum exhibit no special adaptations in their There are a number of species which are usually sub- helophytic species, which appear to take advantage of merged but which are emergent at times of exceptionally wetter habitats for more vigorous growth rather than low water (C. affinis, C. aponogetifolia, C. purpurea). because of a strict requirement for a flooded substrate. The submerged leaves of such species are relatively large, whereas the emergent leaves are quite small, indi- Free-floating aquatics cating that such conditions are unfavourable to their growth. The submerged and emergent leaves of the same The only free-floating species of Araceae is the pantropical Pistia stratiotes. 46 T H E G E N E R A O F A R A C E A E

C 13 P O L L I N AT I O N B I O L O G Y Recent reviews of pollination biology and flowering known about pollination mechanisms in genera with phenomena have been given by Grayum (1984, 1986b, bisexual flowers and simpler or spreading spathes 1990) and Meeuse & Raskin (1988). Few thorough (e.g. Anthurium, tribe Monstereae, many Lasioideae). studies have yet been made of araceous pollination While “trap” seems an accurate description in Arum, biology, though this is clearly a field of the utmost Arisaema and Cryptocoryne, it is less clear in other interest; most of the unusual and important taxonomic genera (e.g. Amorphophallus, Philodendron) whether characters of aroid inflorescences and flowers are the pollinators are unable or merely unwilling to leave probably linked in one way or another to floral bio- the inflorescence, once they have entered it, kept logical adaptations. Detailed studies of tropical genera there perhaps by possible attractants such as stigma have mostly been published only in recent years: secretions, food bodies or sites for reproduction. Williams & Dressler (1976) for Spathiphyllum, Ramirez & Gomez (1978) for Monstera, Silva (1981) for Pistia, Odour is evidently a prime factor in attracting Shaw & Cantrell (1983) for Alocasia macrorrhizos, pollinators and while Araceae are famous for foul Gottsberger & Amaral (1984) for Philodendron and inflorescence odours, which have been compared to Young (1986) for Dieffenbachia. For temperate genera dog faeces, horse dung, rotten fish, old socks, sul- there are the classic studies of Arum by Knoll (1926) phurous gas, dead cow, mushrooms, cheese, etc., and Prime (1960), Vogel’s observations of fungus gnat many others are not foul-smelling. Floral odours in pollination in Arisaema and Arisarum (Vogel 1978) Philodendron, Spathiphyllum and Xanthosoma are and various studies of Arisaema (e.g. Barnes 1934, heavy and spicy and in Anthurium range from spicy Bierzychudek 1982). Only Young (1986) and to the smell of decaying fruit. At least two species of Bierzychudek (1982) have studied pollination biology Amorphophallus are known to have a pleasant floral in relation to entire populations of plants of the same odour (A. galbra, A. manta). The wide range of species. There is a large literature which reports less odours must be correlated with different kinds of critical studies or observations of insect visits to pollinator but though some studies of odour chem- Araceae inflorescences. The phenology and behav- istry have been made (e.g. Meeuse 1966b, 1978, iour of the spathe, spadix and flowers during anthesis 1985), this fascinating field is largely unworked from are also subjects which in general have not been stud- a comparative standpoint. ied critically. Notable exceptions are the studies of Knoll (1926, Arum), Croat (1980, Anthurium), Gotts- The colour of the spathe, and to a lesser extent of berger & Amaral (1984, Philodendron) and Meeuse & the spadix, varies considerably within the family, Raskin (1988, Sauromatum). ranging from inconspicuous greens (e.g. Anthurium, Nephthytis) to elaborate patterns (e.g. Colletogyne, Araceae inflorescences are almost always insect Sauromatum) or striking “flags” (e.g. Anthurium pollinated, although “wind tunnel” pollination has been andraeanum). In fly pollinated species the spathe proposed for Pinellia (Uhlarz 1985). The pollinators of colours and patterns are known to be important in Acorus (Acoraceae) are still unknown. Pollinators so far attracting pollinators (myriophile colours). reported for Araceae (see Table 2) include trigonid Differentiated colour zones are frequent: in bees, euglossine bees, beetles, flies, possibly thrips Philodendron many species have purple zones inside and very doubtfully mites. the spathe tube, while the blade is white or pale green. In Arisaema and Arisarum, the reverse situa- Some species of drosophilid flies are known to tion is found, with white spathe tubes and dark breed on the inflorescences of Alocasia, Colocasia purple blades which are often striped. The foul and Homalomena (Okada 1986). They also exhibit odours of such species as Amorphophallus konjac specialization in their behaviour even on the spadix of are very often associated with flesh-coloured or livid a single species: stamenicolous species lay their eggs spathes, resembling carrion. By contrast, in the male zone while pistilicolous species lay their Zantedeschia aethiopica and many Spathiphyllum eggs in the female. Furthermore, several different pairs species have pure white spathes and perfumed of fly species, one stamenicolous and the other pistil- inflorescences from which euglossine bees collect icolous, are known to breed in association on one fragrances. This also occurs with naturalized aroid species (synhospitality). Zantedeschia in tropical America (G. Gerlach pers. comm.). Trigonid bees collect pollen which also The trap mechanisms of genera with unisexual attracts certain beetles, e.g. Nitidulidae, that eat the flowers and relatively complex spathes (e.g. Arum, fertile male flowers of Aridarum nicolsonii (Bogner, Arisaema, Arisarum, Philodendron) have attracted pers. obs.) most attention in pollination studies. Much less is P O L L I N A T I O N B I O L O G Y 47

Inflorescence heating (thermogenesis) in connec- and, more speculatively, thus to give precision both to tion with flowering has been studied more, particularly the duration of the period of pollinator attraction in Arum (Rees et al. 1976, 1977), Sauromatum (length of heating period), type of odour compound (Meeuse 1966a–b, 1975, 1978, 1985, Meeuse & Raskin utilized (molecular weight) and distance over which 1988) and Philodendron (review by Mayo 1991). pollinators can be attracted (maximum temperature). Thermogenesis in other genera has also been studied In other genera, however, odours are produced appar- (e.g. Symplocarpus), but the only general comparative ently without heating. surveys are by Leick (1914, 1916, 1921 and, briefly, by Engler (1920b). While thermogenesis is very common Odour production and thermogenesis (when pre- in Araceae it is by no means universal. Its function is sent) occur mainly in terminal appendices or in the generally agreed to be to volatilize odour compounds, male zones of the spadix in those genera that lack appendices (e.g. Philodendron, Xanthosoma). Table 2. Pollinators of Araceae. Data mainly from Grayum (1984, 1990). euglossine bees Anthurium, Spathiphyllum, Xanthosoma trigonid bees Monstera, Spathiphyllum, Amorphophallus beetles: Asilidae Amorphophallus beetles: Cetoniidae Amorphophallus beetles: Curculionidae Anthurium, ?Pistia beetles: Dermestidae Dracunculus beetles: Nitidulidae Alocasia, Amorphophallus, Anchomanes, Anubias, Aridarum, Cercestis, Culcasia, Cyrtosperma, Nephthytis, Typhonium, Urospatha, Xanthosoma beetles: Ptiliidae Typhonium beetles: Scaphidiidae Pseudohydrosme beetles: Scarabaeidae Alocasia, Amorphophallus, Anubias, Caladium, Dieffenbachia, neotropical Homalomena, Philodendron, Syngonium, Xanthosoma beetles: Scydmaenidae Typhonium, Zantedeschia beetles: Silphidae Amorphophallus beetles: Staphylinidae Alocasia, Amorphophallus, Anthurium, Chlorospatha, Dracunculus, Lysichiton, Piptospatha, Pseudohydrosme, Typhonium flies: Anthomyiidae Alocasia flies: Calliphoridae Amorphophallus, Dracunculus, Helicodiceros flies: Centropogonidae Arum, Cryptocoryne flies: Chloropidae Peltandra flies: Chorideae Pseudohydrosme flies: Drosophilidae Alocasia, Colocasia, Culcasia, Homalomena, Nephthytis, Schismatoglottis flies: Ephydridae Cryptocoryne flies: Neurochaetidae Alocasia flies: Phoridae Cryptocoryne flies: Psychodidae Arum flies: Sciaridae Arisaema, Arum flies: Simuliidae Arum flies: Sphaeroceridae Arum, Pseudohydrosme flies: Syrphidae Peltandra flies: Mycetophilidae Arisaema, Arisarum mites Ambrosina (reported in literature but very doubtful as pollinators: fruit-set is extremely rare and possibly the pollinators are extinct). thrips (Heterothrips arisaemae) Arisaema (reported in literature but doubtful) 48 T H E G E N E R A O F A R A C E A E

Unpleasant odours from the spathe have been crossing seems to be the general rule. Some cases of C observed in Dracontium and in the spathe blade of self-pollination or apomixis are known or suspected Lagenandra (Bogner pers. obs.) and Cryptocoryne (Amorphophallus bulbifer, Anthurium bakeri, Pinellia, (Vogel 1963, 1990, Bogner pers. obs.). In the spathe some Arum). tube “kettle” of Lagenandra a different, fruity odour is produced which emanates mainly from the anthers Manipulation of pollinator behaviour within the and occasionally also from the so-called “olfactory inflorescence may be the basis for many of the spe- bodies” above the female flowers (Buzgó pers. obs.). cialized features of the spadix, particularly in These structures are present only in Lagenandra and unisexual-flowered genera. Spathe constrictions may Cryptocoryne and their function is still unclear. The act as “skid zones” (Arum) or as “brooms” to eject thickened connectives of stamens or synandria com- pollinators from the female chamber after pollina- monly found in several tribes of subfamily Aroideae tion (Philodendron). The various types of hairs, (e.g. Philodendreae, Homalomeneae, Anubiadeae, scales and warts found on the inner base surface of Caladieae, Colocasieae) probably represent adapta- the spathe (e.g. Amorphophallus) or the wide range tions for osmophore function (see Vogel 1963, 1990 of staminodial or pistillodial structures on the spadix for an important survey of osmophores). In taxa with (e.g. tribe Areae, Bucephalandra) have less obvious well-developed terminal appendices (e.g. tribes functions. In Arum the filamentous pistillodes and Thomsonieae, Areae, Arisaemateae) thickened con- staminodes are thought to exclude inappropriately nectives are usually absent. Odour production in large insect visitors while in Dieffenbachia the pro- genera with bisexual flowers is much less well under- tein-rich staminodes have been shown to be food stood. In Spathiphyllum the stigma plays this role bodies for scarab beetles (Young 1986). It is con- (Vogel 1963, 1990, Williams & Dressler 1976) and in ceivable that at least some of these structures are Anthurium it is probable that the thickened tepal also osmophores which create odour gradients within apices are involved. the inflorescence itself. Certain structures, like the filamentous projections on the spadix appendix and Araceae are always protogynous and the female spathe of Helicodiceros, mimic characteristics of ani- (stigma receptivity) and male (anther dehiscence) mal corpses normally used by the pollinating insects phases usually do not overlap, so that obligate out- for egg deposition or breeding. P O L L I N A T I O N B I O L O G Y 49

C 14 D I S P E R S A L The seeds of most Araceae do not remain viable for Sabah. Shaw et al. (1985) reported Lewin’s honeyeater long. Those lacking endosperm and with large birds (Meliphaga lewinii) and regant bowerbird embryos cannot withstand dessication and the genera (Sericulus chrysocephalus) eating the ripe berries of with fleshy testas are similarly vulnerable. In certain Alocasia brisbanensis (as “A. macrorrhiza”) in east- species, however, the seeds can withstand dessication ern Australia. It is not known if the seeds are for longer periods, e.g. Philodendron bipinnatifidum, regurgitated, destroyed in the gizzard or stomach, or which can remain viable for over 12 months (Mayo, voided intact in the faeces. Circumstantial evidence for pers.obs.). Seeds with a leathery testa, such as in Arum other genera points to birds (Anthurium) and mam- and Arisaema, also have longer viability. One conse- mals, including primates (Anchomanes, Philodendron) quence of short viability is that Araceae seeds are and bats (?Philodendron), as the commonest vectors. unlikely to survive long distance dispersal by natural The tawny-capped euphonia (Euphonia anneae) was vectors. This makes it unlikely that generic disjunctions reported to feed heavily on fruits of Anthurium over major ocean basins have resulted from long dis- (Loiselle & Blake 1990). Wheelright et al. (1983) tance dispersal. Islands located far from continental observed three different birds (resplendent quetzal: areas tend to be very poor in native Araceae or lack Pharomachrus mocinno, long-tailed manakin: them altogether, e.g. most islands of the Pacific. In Chiroxiphia linearis, common bush tanager: Sauromatum venosum and Remusatia vivipara, how- Chlorospingus ophthalmicus) feeding on three uniden- ever, long distance dispersal may well be the cause of tified species of Anthurium. A fecal sample of the their widely scattered Old World distribution patterns. wood thrush (Hylocichla mustelina) was observed con- Little is known of the viability of Sauromatum seeds, taining seeds of an unidentified species of but its purple berries are very probably dispersed by Dieffenbachia by Blake & Loiselle (1992). birds, since in upland forests it often occurs as epi- phyte. Remusatia vivipara, also epiphytic in upland The common palm civet (Paradoxurus hermaph- forests, is almost certainly distributed mainly by means roditus) is reported to disperse the seeds of Colocasia of its peculiar bulbils, which are burr-like with esculenta in Indonesia (Hambali 1980). Marks were recurved scales. left on the peduncle by the claws and teeth of this mammal and germinating seeds of two cultivars of C. Another important generalization that can be made esculenta were seen in its excreta. This is the only is that animal dispersal, and more specifically, ornitho- report seen that records the passage of viable aroid chory (bird dispersal), must be the dominant mode, seeds through the digestive system of an animal. due to the universality of berried fruits in Araceae. Hambali (1980) confirmed an earlier report by Reliable data on dispersal is very scarce, a recent Leeuwen (1932) that seeds of Colocasia gigantea were exception being that of Barbara and David Snow dispersed by the common palm civet, and Hambali (1988) for bird dispersal by blackbirds (Turdus merula) (1980) also stated that ripe fruits of Homalomena pen- and robins (Erithacus rubecula) of Arum maculatum dula are usually eaten by the same animal. The fruits in England. Peckover (1985) observed in Papua New of all three aroid species are odoriferous, which may Guinea that captured birds of paradise (magnificent attract the civet (Hambali 1980). riflebird: Ptiloris magnificus) fed on fruits of Amorphophallus paeoniifolius and regurgitated seeds Presentation of the fruits for dispersal is normally about four hours later. Glossy-mantled manucode bird a rather sudden event. In Philodendron and of paradise (Manucodia atra) was also observed on an Dieffenbachia the spathe falls off or splits at maturity infructescence of Amorphophallus paeoniifolius and to reveal the infructescence, whereas in genera with was presumed to be feeding rather than merely perch- non-persistent spathes the berries remain inconspic- ing. If a similar delay between feeding and uous during maturation and take on their bright regurgitation also occurs in the wild a wider dispersal colours in a final rapid flush. In Anthurium the berries could thereby be achieved. Indigenous people (the are extruded from the perigone at maturity and in local Batak population) have observed hornbills most species dangle by tiny threads of tepal epider- (Bucerotidae) eating berries of Amorphophallus mis. This mode of presentation has also been titanum, and the bulbul bird (Pycnonotus zeylanicus) observed in Cyrtosperma by Hay (1988). The bright feeding on the berries of A. brooksii in Sumatra colours, sticky gelatinous mesocarp and mode of pre- (Hetterscheid, pers. comm.). T. Lamb (pers. obs.) has sentation in such species strongly suggests bird added support to this evidence by his own observa- dispersal. In other, terrestrial or rupicolous species, tions of bulbul birds eating berries of A. lambii in e.g. Anthurium erskinei, the inconspicuous greenish berries merely fall into a heap onto the ground, sug- 50 T H E G E N E R A O F A R A C E A E

gesting some other kind of vector. Genera of sub- dense infructescences of Caladium, Colocasia, C family Monsteroideae display the seeds by Philodendron, Syngonium, may disappear very soon simultaneous abscission of the stylar region of each after exposure. flower. The resulting compound mature fruit has the seeds embedded in sticky, often sweet and mucilagi- Dispersal by ants (myrmecochory) has been nous material. T. Croat (pers. comm.) has suggested observed in Biarum, in which the strophiole of the that these fruits may be dispersed by monkeys. seed is probably implicated. In the tropics it is likely Philodendron subgen. Meconostigma has fruits with that ant dispersal is involved in the occurrence of cer- a similarly sweet, pineapple to mango flavour and tain aroids in Amazonian root gardens (e.g. some of its Brazilian vernacular names suggest Philodendron megalophyllum (syn. P. myrmecophilum), mammal dispersal agents (“monkey’s banana” or Anthurium ernestii, A. gracile; see Ule 1905). “bat’s banana”). Black spider monkeys (Ateles panis- Anthurium gracile is a characteristic plant of ant-gar- cus) have been observed eating the infructescences dens (Benzing in Huxley & Cutler 1991). T. Croat (pers. of Philododendron goeldii (Bogner, pers. obs.). In comm.) has observed ants dispersing seeds of Anthurium, Dieffenbachia, Nephthytis and other Philodendron megalophyllum. genera, the berries are often sufficiently distant from one another within the infructescence to be dis- Dispersal by water (hydrochory) is very probable in persed individually. Mature berries in Arum and the helophytic genera Montrichardia and Nephthytis have been observed to remain for a rela- Typhonodorum, which have very large berries and tively long time awaiting dispersal. In contrast the floating seeds. Lagenandra, Cryptocoryne and Pistia are certainly water-dispersed and have much smaller fruits and floating seeds. D I S P E R S A L 51

C 15 G E O G R A P H Y Reviews of araceous geography have been published Lysichiton reaching subarctic latitudes in Alaska. An recently by Croat (1979), Grayum (1990), Hay (1992b) unusual pattern is shown by the close taxonomic and Mayo (1993). Individual generic distributions are relationship of Peltandra (eastern North America) given in the maps, generic treatments and country and Typhonodorum (Madagascar). lists (Appendix). Certainly the most interesting feature of araceous The genera of Araceae are concentrated in the geography is the occurrence of three genera tropics of America, Southeast Asia and the Malay (Homalomena, Schismatoglottis and Spathiphyllum) Archipelago (we include under this designation with ranges disjunct between the Malay Archipelago or Malaysia, Indonesia, the Philippines, Papua New Melanesia and tropical America. These are all rainfor- Guinea, Singapore and Brunei). Continental tropical est herbs for which long-distance dispersal by water or Africa is the next richest region, followed in order of suitably far-ranging animal vectors is probably impos- decreasing diversity by temperate Eurasia, southern sible. It is also difficult to imagine these genera being Africa, Madagascar and the Seychelles, and North rafted by the southern Gondwanic route (Chile, America (including northern Mexico). Australia has Antarctica, Australia, New Zealand) or via the Bering two endemic genera (Gymnostachys, Lazarum). The Straits because of their intolerance of even subtropical other native Australasian genera (in northern conditions, much less temperate climate. Their pat- Australia) are essentially extensions of the tropical terns of diversity are dissimilar. While Homalomena Asian and Malay Archipelago flora. and Schismatoglottis are most diverse in southeast Asia and Malesia, Spathiphyllum is richest in tropical The vast majority of genera are endemic to the America. No really plausible historical explanation has major regions (as given above), but some extend fur- yet been proposed for these disjunctions, nor for those ther. Pistia is pantropical and Calla is circumboreal of other taxa with this kind of range, such as Sloanea in the northern hemisphere, reaching as far as the (Elaeocarpaceae) and Heliconia (Heliconiaceae). subarctic zone in northern Scandinavia. Arisaema is Nevertheless, it is tempting to speculate that these pat- the most widespread genus of more than one species. terns are the relicts of a once-continuous distribution It is most diverse in south-west China, and extends in Gondwanaland during the Cretaceous period. west as far as Tanzania and Burundi, and east to eastern North America and northern Mexico. A number of species which are important as food Amorphophallus, Remusatia, Rhaphidophora and (Colocasia esculenta, Cyrtosperma merkusii, Sauromatum are shared between tropical and sub- Xanthosoma sagittifolium complex) or ornamental tropical Africa and Asia and the Malay Archipelago. plants (Alocasia macrorrhizos, Epipremnum pinnatum Amorphophallus and Remusatia also occur in ‘Aureum’, Monstera spp., Philodendron spp., Syngonium Madagascar and northern Australia. Pothos is found spp., Zantedeschia aethiopica) have been widely dis- in Madagascar as well as in tropical Asia, Malay persed throughout the tropics by man and have become Archipelago, northern Australia and the western naturalized as well. Some Typhonium species (Nicolson Pacific. Lysichiton and Symplocarpus occur in north- & Sivadasan 1981) have become widely dispersed and eastern Asia as well as in temperate North America, weedy in many parts of the tropics. 52 T H E G E N E R A O F A R A C E A E

C 16 U S E S Bown (1988) gives an excellent general account of tified as Xanthosoma, showing that Xanthosoma must useful aroids to which the reader is referred for have been cultivated as a food plant in precolumbian greater detail. times (Costantin & Bois 1910, Towle 1961). Xanthosoma sagittifolium is most widely known as Food plants cocoyam. There are many other names associated with the other species used for food, X. violaceum, X. The most important food aroids are from tribes atrovirens, X. mafaffa (used especially in Nigeria, Colocasieae and Caladieae, i.e. Colocasia and Okeke 1992), X. brasiliense, X. caracu and X. robus- Xanthosoma. The great majority of Araceae are poi- tum. These are all more-or-less closely related to X. sonous when fresh and in almost all cases, edible sagittifolium, but much taxonomic confusion cur- species must be cooked or processed in some way rently reigns in this group of taxa and needs to be before they can be used as food. clarified urgently (S. Thompson, pers. comm.). Recent technical reviews of edible aroids, especially Alocasia macrorrhizos, the giant taro, was widely taro, are given by Wang (1983) and Chandra (1984). used for animal fodder in the nineteenth century and Colocasia esculenta, the taro plant, originated in trop- more rarely for human food. The stems have relatively ical Asia and has been cultivated there for more than little mucilaginous material and are eaten after roasting. 2000 years. Its original geographical range is obscure, They are said to be tasty when warm but irritant and but Assam and Burma are likely possibilities. Today it unpleasant when cold. The stems may grow to as long is an important root crop in most humid tropical coun- as 5m and contain abundant latex. tries, especially in the Caribbean, Africa, Madagascar, Asia and the Pacific Islands. The tuberous stem is a rich The starch-rich tubers of the elephant yam, starch source (13% to 29% by weight, depending on the Amorphophallus paeoniifolius (syn. A. campanulatus), cultivar), and the leaves of certain cultivars are widely are commonly used as food in tropical Asia, especially eaten as a spinach. Taro tubers are also richer in pro- in India, where the species is widely cultivated. The tein than most other major starch crops and provide tubers may reach 10 kg in weight and are eaten after very nutritious food (Chandra 1984). When fresh, all roasting or boiling, like potatoes. In the Malay Peninsula parts of the plant are poisonous and must be cooked, the tubers of Amorphophallus prainii are also eaten roasted or heated in some way to become edible. The after cooking, but must be sliced and soaked in water uncooked tissues contain an irritant toxin which can beforehand. burn the skin (see chapter 11). Amorphophallus konjac (syn. A. rivieri), the konjac Colocasia esculenta is known by a plethora of plant, is widely cultivated in Japan. The mannan-rich local vernacular names which correspond to the many tubers (see chapter 11) are harvested after 1 to 3 years. different land races and cultivated varieties that have They are boiled in water and then treated with lime evolved by human selection. The species is most milk to make a flour from which noodles (chira take) or widely known today as taro, its name in the Pacific cakes (chiroko) are prepared. A jelly (nama konjaku) is Islands. In some varieties the flesh of the tuberous also made from the lime milk preparation and the gum- stem is white or yellow while in others it is violet; the like juice can also be used to make glue. Konjac flour latter are sometimes preferred for their stronger is traditionally prepared by slicing the raw tubers into flavour. Some cultivars are used as animal food, espe- 5mm thick pieces and then drying them in the sun for cially for pigs. Cultivars with tubers rich in a week or so until the water content is reduced from mucilaginous material tend to be used for animal feed about 90% to about 15%. The dried material is then while those with a lower mucilage content are used pounded into a flour. Modern methods to extract kon- for human consumption. jac mannan involve washing the flour with 70% ethanol in a nylon filter or by acetylation. The quality of the flour The neotropical genus Xanthosoma also contains differs according to the plant variety, the area in which species which are very important food plants, partic- it is grown and the method of preparation. The rigid gels ularly Xanthosoma sagittifolium. This and other prepared from the flour also vary according to the species are widely cultivated, not only in tropical source of the raw material. The three main varieties of America but also in tropical Africa and Asia. As in konjac cultivated in Japan are: ‘zairai-shu’ (traditional Colocasia esculenta, both leaves and starch-rich tubers cultivar), ‘shina-shu’ (Chinese cultivar) and ‘bicchu-shu’. are used and can only be eaten after cooking. Plant remains from ancient Peruvian graves have been iden- The thick, starch-rich rhizomes of Cyrtosperma merkusii (swamp taro) are used for food in Southeast Asia and the Pacific Islands where the plant is grown in U S E S 53

nutrient-poor sites unsuitable for the main starch staple, family as a whole. The various medicinal uses of aroids Colocasia esculenta. The tubers of various species of described by Bown include external healing of stings, Dracontium have been recorded as a food source for wounds, skin complaints and arthritis, expectorants and Amerindians; preparation is by roasting. decongestants, contraceptives, parasite insecticides, anti- cancer agents, sedatives and hallucinogens. Acorus The fresh, ripe infructescences of Monstera deli- (Acoraceae) has a long history of use as a medicinal ciosa are eaten and used to flavour ice cream in plant for problems of the digestive system and the sci- Mexico; the taste recalls that of pineapple. The tri- entific literature that exists on the subject is extensive chosclereids in the rind of the infructescence are (see Röst 1978, 1979a, b, Röst & Bos 1979) troublesome and the wider use of this delicious fruit is hampered by lack of better cultivars. In Brazil, the Toxic effects in Araceae are widely known and infructescences of Philodendron bipinnatifidum are have received some attention from chemists (see chap- occasionally eaten by man (Crisci & Gancedo 1971); ter 11). The effects of the highly toxic Dieffenbachia they have a mango-like flavour and a slightly irritant are the best known and most completely studied. after-taste. Araceae are used in arrow poisons and fish toxins (Bown 1988). The seeds of Typhonodorum lindleyanum and Montrichardia linifera are recorded as having been Magical and ritual uses of aroids are known, but lit- eaten by indigenous peoples of Madagascar and trop- tle studied. The use of Dieffenbachia and Caladium to ical South America respectively, after cooking or ward off the “evil eye” is widespread in Brazil. roasting. The seeds of Orontium aquaticum were eaten after drying and boiling by North American Fibres indigenous people. The roots of Heteropsis and Philodendron have Theophrastos recorded the use of tubers of Arum been used for fibres in tropical South America. italicum as a source of food starch in ancient Greece. Heteropsis spruceana is still today an important source In the Middle Ages, tubers of Arum maculatum were of twine and basket-weaving material in Brazilian used as famine food, especially in England. North Amazonia. The stems of Montrichardia linifera provide American Amerindians are known to have made a flour a fibre which is suitable for paper-making. Bown from the tubers of Arisaema triphyllum, but neither (1988) reports the use of fibres from species of Arum nor Arisaema appear to be used for these pur- Anthurium, Cercestis, and Gymnostachys anceps. poses today. Medicinal, toxic and magical uses Ornamental uses C The ethnobotany of the Araceae appears to be Araceae are best known as ornamental plants and diverse and fascinating, judging from the many circum- commercially are among the most important foliage stantial reports which are scattered through the scientific ornamentals cultivated and sold for display. Their beau- literature. No comprehensive modern review exists and tiful and unusually diverse leaf forms and textures form none is attempted here. However, valuable contribu- an essential part of any tropical plant display. In the tions have been made by Plowman (1969) and Croat tropics they are universally seen in private and public (1994) for the New World and particularly by Bown gardens. Further details are given in chapter 17. (1988), who gave an excellent modern account for the 54 T H E G E N E R A O F A R A C E A E

C 17 C U L T I VAT I O N Crops Orontium, Lysichiton, Peltandra and Symplocarpus. They do well in a nutrient-rich, loamy soil in a sunny The major crop species are Colocasia esculenta (taro), or half shaded position. Calla prefers a more acid habi- Xanthosoma sagittifolium (cocoyam), Cyrtosperma tat in shaded or partially shaded sites. Orontium does merkusii (swamp taro), Alocasia macrorrhizos (giant best in full sun. Propagation is mostly by seed in taro), Amorphophallus paeoniifolius (elephant yam) and Lysichiton, Orontium and Symplocarpus. Acorus and Amorphophallus konjac (konjac). Calla are easily propagated by dividing the rhizomes. In Europe, naturally occurring Acorus calamus is The cultivation methods of major aroid crop species triploid and cannot set seed. Division of Lysichiton, are dealt with in some detail by Wang (1983) and Orontium and Symplocarpus is best not attempted Chandra (1984), which also give extensive bibliogra- because their rhizomes are situated rather deeply and phies of the subject; Bown (1988) gives a very readable should not be disturbed. and informative account. Colocasia esculenta, Xanthosoma sagittifolium and Cyrtosperma merkusii Tropical genera are generally grown in humid conditions, although In tropical countries Araceae have an important role some varieties of Colocasia esculenta are also suitable for drier sites. Their preference for moisture makes as ornamental plants. They are cultivated in public these species especially suitable for humid tropical parks as well as in private gardens or as house plants. subsistence farming on sites unsuitable for other kinds Under these favourable conditions they are easily of food crop. Amorphophallus paeoniifolius and A. grown and need little care beyond regular watering in konjac are grown in shade and on well-drained soils. regions with a drier climate. Their varied life forms pro- vide magnificent climbers, bedding and large, sculptural Ornamental Araceae terrestrial plants. Commonly cultivated aroids in tropi- cal countries include species of the genera Aglaonema, Temperate genera Alocasia, Anthurium, Caladium, Colocasia, Many hardy Araceae do best in shady and half Dieffenbachia, Epipremnum, Homalomena, Monstera, Philodendron, Rhaphidophora, Schismatoglottis, shady sites with humus-rich soil and are therefore very Scindapsus, Spathiphyllum, Syngonium, Xanthosoma suitable for temperate woodland gardens. This group and Zantedeschia (especially at higher altitudes). Lasia includes species of Arisaema, Arisarum (with winter spinosa, Typhonodorum lindleyanum, Montrichardia protection by leaf litter), Arum, Dracunculus and linifera and M. arborescens are often cultivated in or Pinellia. In cultivation they prefer leaf litter on the soil beside ponds where because of their large size they surface and this also protects the tubers or rhizomes form impressive and unusual-looking stands. Pistia stra- from frost in winter. Growth starts in the spring or wet tiotes is often grown in ponds and lakes. The geophytic season and dormancy sets in during the autumn or (tuberous and rhizomatous) tropical Araceae are only dry season. They are suitable for growing with ferns seldom cultivated in gardens, perhaps because they and other shade-loving plants. Smaller species can also require more care. be grown in shady parts of rock gardens. Half hardy species can be grown in pots and overwintered in In temperate regions, tropical genera are very frost-free conditions. Pinellia ternata easily becomes important as house plants and are raised and sold on weedy by vegetative propagation from the bulbils an industrial scale. Nearly every apartment or house in which form on the petioles. Europe and the Americas has at least one aroid as an ornamental plant. These, the best known and most Biarum, Eminium and some Arum species thrive widely grown aroids, tolerate low light levels (espe- best in the rock garden, with stony, drier soils and cially important for surviving the winter), dry air from sunny sites on walls or between rocks, but they require central heating, irregular watering and general neglect, some protection in winter. Helicodiceros, a all of which makes them suitable for cultivation as Mediterranean genus, is not quite hardy and the tubers house plants. A variety of Araceae is also grown in need to be well protected during a temperate winter. hydroculture in large offices and public buildings. One way to do this is to remove them from the soil and Particularly well known house plant aroids belong to keep them frost-free and dry in sand. the genera Monstera (M. deliciosa), Philodendron (P. scandens, P. erubescens, P. bipinnatifidum and many Other genera are best grown in wet places, like other species and hybrids), Dieffenbachia (many cul- bog gardens, along streams, in pools or on the margins tivars of D. seguine and D. maculata), Aglaonema of a pond. These include Acorus (Acoraceae), Calla, C U L T I V A T I O N 55

(many cultivars of A. commutatum, A. nitidum and in the old soil and left undisturbed, but they may also be C other species), Syngonium (cultivars of S. podophyllum, removed from the soil and stored in a shady, humid often with variegated leaves), Zantedeschia (especially place. Repotting should be carried out as the new roots Z. aethiopica and hybrids), Spathiphyllum (especially begin to develop, i.e. near the end of the dormant period S. floribundum, S. cannifolium and other species and before the shoots appear. For example, in Europe the hybrids), Epipremnum (E. pinnatum ‘ Aureum’, also tubers of Caladium bicolor are repotted in the early part known as Pothos aureus or Scindapsus aureus), of the year (February, March) and grow during the sum- Anthurium (particularly A. andraeanum and A. mer until the end of August or September when they scherzerianum and cultivars) and Caladium (C. bicolor re-enter dormancy. They must then be kept dry in the old and cultivars). Many species of Cryptocoryne are prized compost or removed entirely from the pot until the fol- by aquarists and species of Anubias are also widely lowing year. Amorphophallus konjac, Colocasia esculenta grown as aquarium plants. and Sauromatum venosum can be grown outside from spring to autumn in central Europe. Propagation of these Commercial nurseries that raise Araceae as house genera is by seed or daughter tubers. plants are common in many parts of the world, per- haps the most important being the USA and the Cultivated tropical aquatics and helophytes include Netherlands. Anthurium and Zantedeschia are the Cryptocoryne, Cyrtosperma, Dracontioides, Lagenandra major sources of aroid cut flowers and are cultivated and Urospatha. Most of these are grown only in a few on a large scale. Anthurium andraeanum has a very botanic gardens. The most difficult species to grow, large number of cultivars differing in spathe colour such as Jasarum and some species of Cryptocoryne, are with various shades of red, pink, white and green and from nutrient-poor black water rivers. These species in spathe texture and the attitude of the spadix. require an acid soil and very soft water. Compost made Zantedeschia aethiopica and other species and their from siliceous rock (granite or gneiss) or acid peat or hybrids have a range of spathe colours which includes leaf mould (especially from Fagus sylvatica) can be white, yellow, and various shades of red, and the used. For less demanding species a mixture of loam, spathes vary widely in size and shape. Plants for the sand and peat or just sand and loam is sufficient. cut flower market are cultivated in beds. Some are grown in gravel for hydroculture (hydroponic) or in Some Araceae which do not become completely ordinary, standardized or peat soils. Propagation is dormant need a drier period during which growth usually vegetative by means of cuttings but for certain more-or-less ceases. At this time, however, they cultivars tissue culture is also employed. should not be allowed to dry out completely. Zantedeschia aethiopica, for example, needs a resting Tropical aroids are mostly evergreen and are forest period in Europe between spring and summer (July) herbs, climbing hemiepiphytes or epiphytes in their nat- when it should be kept in rather dry and sunny con- ural habitats. Anthurium species grow best in conditions ditions. Following this the plant should be repotted of high humidity and a minimum temperature of and watering gradually increased as the plant comes 18–20°C, with a soil mixture of sphagnum moss and into more active growth. Other species (e.g. Z. albo- fibrous peat. Dieffenbachia, Philodendron, Monstera maculata) have a completely dormant period without and Spathiphyllum can be grown in standardized soil leaves. Some Alocasia species, especially those with mixtures or specially made up composts composed of beautifully coloured foliage (e.g. A. lowii, A. sande- sterilized soil, treated leaf mould or bark chips, peat riana), can be difficult to grow and are only suitable and sand with some balanced fertilizer. As a base, sandy for “stove” conditions (high temperature, high humid- loam with added peat and sand can also be used. ity and shade). They also have a dormant period Zantedeschia prefers a heavier, more loamy mixture. during which they should be watered only very spar- Climbers like Philodendron, Monstera and Epipremnum ingly, while never being allowed to dry out need a support on which to develop properly, such as completely. Special attention must be given to the a bamboo stick, moss-covered stick, roughened wooden plants during this critical period. When the shoot billet or a suitable wall. starts into growth the plant should be repotted or replanted in a mixture of sphagnum moss and peat as Much more care is required to successfully grow sea- used for Anthurium; chopped and partly decomposed sonally dormant genera with tubers or rhizomes and wood or bark chips can also be used in the mixture. these are consequently less commonly cultivated, e.g. Alocasia macrorrhizos, A. odora and A. portei, on the Anchomanes, Amorphophallus, Caladium, Gonatopus, other hand, are evergreens with no resting period. Dracontium, Remusatia, Synandrospadix, Taccarum, Propagation of Alocasia is by division of the stem, Theriophonum, Typhonium and Xanthosoma sect. from the tuber-like stolons or by seed. Acontias. These plants must be kept dry during their dormant period and they are best grown in pots with a True rheophytes like Aridarum, Bucephalandra, rather sandy compost in order to control soil humidity Hottarum and Piptospatha are rather difficult to maintain better. They require regular and abundant water during in cultivation because of their special requirements. the growing period but watering must be reduced and They are best grown in a mixture of sphagnum moss very carefully controlled at the beginning and end of and fibrous peat, if possible on rocks with a little humus; dormancy. While dormant the tubers are best kept dry they require high temperature and humidity. 56 T H E G E N E R A O F A R A C E A E

C 18 C O N S E R VAT I O N The main threat to the long term survival of many results in faster flowing streams, which in turn gouge Araceae is the loss and reduction in quality of their nat- out the stream bed and deposit sediment on the plants ural habitats, especially in the rainforest regions of Asia, along and near the banks. In such a situation the orig- the Malay Archipelago, Africa and tropical America. inal plants soon die out. Fire is also very destructive, both in its immediate effects and in favouring the inva- Some Araceae are highly adapted to specific habi- sion of weedy plants which prevent the regeneration tats and cannot survive in changed conditions; e.g. of the great majority of forest Araceae. Only a small Chlorospatha and rheophytic species of tribe minority of species can establish themselves in sec- Schismatoglottideae. These taxa have proved difficult ondary regrowth or in open areas, e.g. some of the to cultivate and are unlikely to survive in the long term large species of Xanthosoma sect. Xanthosoma, in botanic gardens. The same is also true of forest Alocasia, Typhonium blumei, Zantedeschia aethiopica aquatics such as Jasarum (from blackwater rivers) and and certain races of Colocasia esculenta with a vigor- several species of Cryptocoryne. ous stoloniferous habit, which have become naturalized widely in the tropics. Some species of Arum have Currently, not one species of Araceae is listed in established themselves in parks and plantations, and CITES (Convention on International Trade in Cryptocoryne spiralis is a rice field weed. Arisaema Endangered Species). Formerly Alocasia zebrina and A. mooneyanum appears to have spread vigorously in sanderiana from the Philippines were in Appendix 2. open highland fields in Ethiopia, although it was prob- ably originally a forest edge species. These, however, The overcollection of Araceae for commercial trade are exceptions; the great majority of Araceae disappear is unlikely to be a major cause of extinction, although along with the forest habitat. some Cryptocoryne species have been collected in large quantities for the aquarium trade, much reducing Restricted endemic species are at particular risk. their natural populations. Removal of tropical forest, The determining ecological or historical factors for however, eliminates most terrestrial, climbing and epi- such narrow ranges remain completely unknown and phytic species, many of which are shade-dependent. this makes the future prospects for such species bleak. Other habitats are also affected. In the case of the aquatic genus Cryptocoryne, reduction in tree cover C O N S E R V A T I O N 57

C 19 F O S S I L R E C O R D Gregor & Bogner (1984, 1989) have reviewed the fossil Miocene of Europe. Limnobiophyllum may be consid- record of Araceae and their papers should be consulted ered as standing between Pistia (Araceae) and for further details, particularly for the specialist literature. Spirodela (Lemnaceae). No fossils are known earlier than the Eocene which Leaf fossils can definitely be ascribed to the Araceae, although Spathiphyllum-like pollen has been described from the Leaf fossils from North America have been assigned Palaeocene and Limnobiophyllum scutatum from the to the genera Peltandra, Philodendron and Pistia late Cretaceous (see below). while South America leaf fossils have been placed in the genera Stenospermation (S. columbiense) and Fossils consisting of different parts of the plant Caladiosoma. Caladiosoma miocenica from the Miocene of Trinidad belongs either to the genus The genus Acorus (Acoraceae) is well documented Caladium or to Xanthosoma (Berry 1925). from the Miocene of Spitzbergen by the fossil A. brachystachys, in which at least the leaf, inflorescence A North American Eocene leaf fossil with leaf vena- axis and spadix are connected (Heer 1870). However, tion typical of Peltandra was described by Hickey (1977) the fossils from North America treated under this name as P. primaeva. A remarkably large fossil leaf, also from by Lesquereux (1878) belong to the Coniferae. Acorites the Eocene of North America, was described in great heeri (syn. Acorus heeri), from the Eocene of North detail by Dilcher & Daghlian (1977) as Philodendron America, is also close to Acorus. The fossils of this limnestis. However, Mayo (1991) suggested that the leaf species consist of one single spadix and one spadix venation indicates that this fossil would be better placed with a piece of its inflorescence axis (Crepet 1978). in the tribe Peltandreae, near Typhonodorum. Kvaček (1995) has recently drawn attention to Two leaf remains from Sumatra were described by Limnobiophyllum Krassilov, a fossil of extraordinary R. Kräusel (1929): Araceophyllum engleri is probably interest in connection with the Araceae/Lemnaceae rela- from the Pliocene and resembles Pothos, while tionship. This fossil genus consists of free-floating, Araceophyllum tobleri is probably from the Upper stoloniferous plants with one or two suborbicular to Miocene and belongs to the tribe Monstereae. reniform leaves of different size, and simple as well as branched roots on a reduced main axis. There are Two other species of Araceophyllum and the genus (9)10–14 curved primary veins in the leaf blade, and Anthuriophyllum are leaf fossils which only doubtfully these join the margin or run into the apex. The second belong to the Araceae. Weyland (1957) described order veins are reticulate between the primaries (higher Anthuriophyllum spectabile based on cuticular analy- order venation is absent in Spirodela and Lemna). There sis from Tertiary brown coal deposits along the lower is no sign of the lateral pouches which occur in Rhine. More material showing the venation type is Spirodela and Lemna but aerenchyma and pigment required to confirm that it belongs to the Araceae. The cells are present. Two species are recognized: descriptions of Araceophyllum striatum (Weyland 1957) Limnobiophyllum scutatum (Dawson) Krassilov and L. and A. tarnocense (Rásky 1964) are too incomplete to expansum (Heer) Kvacek. Turion-like bodies are asso- be accepted confidently as Araceae fossils. ciated with L. scutatum. No seeds were found directly connected to these fossils, but numerous, isolated, Several leaf impressions of the genus Pistia (P. ribbed seeds were associated with the same strata. claibornensis, P. corrugata, P. mazelii, P. nordenskiöldii, These seeds resemble those of Spirodela, Lemna and P. wilcoxensis) have been described, but some are similar fossil remains were described by Nikitin (1976) not aroids at all, while others are only doubtfully so. as Lemnospermum pistiforme Nikitin from Siberia. P. wilcoxensis Berry (1916), for instance, from the Generally, Limnobiophyllum most resembles Spirodela Eocene of North America, has a venation type not but is larger and the branched roots are more like those found in the Araceae and should be excluded alto- of Pistia. The fossils are less similar to Limnobium Rich. gether. Chondrophyllum nordenskiöldii was (Hydrocharitaceae), a relationship suggested by some transferred by Berry (1910) to Pistia but it represents authors. Limnobiophyllum scutatum is known from the dicotyledonous leaves. Depape & Gauthier (1953) latest Cretaceous to the Oligocene of western North described fossil leaves from the Eocene of Morocco America and the Palaeocene of east Asia. and suggested them to be close to extant Pistia stra- Limnobiophyllum expansum is known from the tiotes. They also gave a survey of reputed Pistia leaf fossils, but none of these can be convincingly assigned to this genus or to any other aroid genus. 58 T H E G E N E R A O F A R A C E A E

Recently, three different leaf fossils have been Fossil fruits and seeds described from the Eocene of the Grube Messel (Germany), although not formally named (Wilde 1989). Nikitin described Acorus procalamus based on fruits One belongs to tribe Colocasieae and another to tribe and seeds from Quaternary deposits in the former Monstereae. The third resembles Aglaonema or Soviet Union and this has been well illustrated by Katz Homalomena (i.e. subfamily Philodendroideae in the et al. (1965). Fossil seeds of Calla palustris and Acorus sense of Engler 1920b) in having a coriaceous leaf calamus are also described by Katz et al. (1965), while blade with parallel-pinnate venation and laticifers. seeds typical of Pistia were described by Dorofeev (1963) as P. sibirica from the Oligocene of Siberia. Leaf-like fossils were described as Arisaema cretacea (Lesquereux 1892) and as Arisaema dubia Among the best-known aroid fossils are seeds of (Hollick 1897) which were interpreted as portions of a Monstereae and Lasioideae described from European spathe, but both are very doubtfully araceous and are brown coal deposits of Oligocene, Miocene and best excluded from the family. Pliocene age. These were formerly considered to belong to Epipremnum, but are today included within Fossil spadices and infructescences three genera: Epipremnites (type: E. ornatus) and Scindapsites (type: S. crassus) of tribe Monstereae, and Acoropsis eximia (syn. Carex eximia, Acoropsis Urospathites (type: U. dalgasii) of subfamily Lasioideae minor), from the Eocene Baltic Amber, represents a (Gregor & Bogner 1984, 1989). Keratosperma, with the well-preserved infructescence of the tribe Monstereae single species K. allenbyense, is an Eocene fossil seed (Bogner 1976c). This infructescence lacks tepals and from Canada and belongs to the Lasioideae (Cevallos- is therefore not related to Acorus as suggested by Ferriz & Stockey 1988). Conwentz (1886). Fructus polyspermus from the European Miocene could belong to the Araceae but Scindapsites has reniform seeds with a smooth testa cannot be assigned with certainty to the family of variable thickness covered with scattered foveolae. (Engelhardt 1877). Crepet (1978) considered the There are distinctive outgrowths on each side of the Eocene spadix Araceaeites fritelii from North America raphe which are sometimes partly coherent. to be too incomplete to be included in the Araceae Epipremnites has curved seeds with the foveolae and the same must be said for the impressions of arranged in rows on an otherwise smooth testa; no Lysichiton washingtonensis from the Miocene of west- outgrowths are present on the raphe. Urospathites has ern North America (Berry 1931). Lysichiton nevadensis somewhat curved seeds and the testa is warty, spinose is another impression from the Miocene of western or tuberculate. North America and could be araceous (MacGinitie 1933). An infructescence from the Cretaceous of east- Fossil pollen ern North America was described as Arisaema mattewanense, but no internal structure was preserved Some fossil monosulcate pollen grains have been and so this fossil remains very doubtfully ascribed compared with Araceae, but this pollen type also (Hollick 1897). A spadix from the Miocene of North occurs in many other families of monocots and dicots America was described as Orontium fossile (Cockerell and these fossils cannot be accepted as araceous in the 1926). Araceaeites parisiense, from the Palaeocene of absence of associated floral remains. France, was described as a spadix by Fritel (1910) and was compared by him to Spathiphyllum, but it is very Mtchedlishvili and Shakhmundes (1973) described incomplete and thus doubtful. Knowlton (1926) the pollen of four species in a new genus, Jugella (type: described a very questionable infructescence from the Jugella sibirica) from the Lower Cretaceous of the for- Miocene of North America as Arisaema hesperia. mer Soviet Union. The authors compared them to Spathiphyllum pollen because of their striate exine, but Apart from Acoropsis, Acorites and Acorus brachy- Spathiphyllum has inaperturate pollen grains whereas stachys, all these fossil spadices or infructescences, in Jugella they are monocolpate which suggests that this described under Araceaeites or an extant genus, are too genus should be excluded from the Araceae. incomplete to be ascribed with confidence to the Araceae and their taxonomic assignment must there- Wodehouse (1933) described North American fore remain highly dubious. Better preserved material pollen from the Eocene as Peltandripites davisii and of these fossil taxa is needed to clarify their position. compared it with Peltandra, describing it as “...without germinal furrows or pores...”. However, his rough Infructescences from the latest Eocene to earliest drawing shows something like a furrow and since the Oligocene from Egypt described as Teichosperma pollen of Peltandra is inaperturate this argument seems spadiciflorum (Renner 1907, Kräusel & Stromer 1924) doubtful. Peltandripites dubius probably belongs to in the Pandanaceae and new material studied by the Asteraceae (Sah & Dutta 1966, Thanikaimoni 1969). Tiffney & Wing (unpubl.) show that these fossils, with their smooth reniform seeds, belong to the genus Biswas (1962) described fossil pollen from the Epipremnum of the Araceae (E. spadiciflorum). Eocene of Assam as Colocasioideaepites, but these probably represent palm pollen grains (Nypa type). Monsteroideaepites eospathiphyllum is inadequately F O S S I L R E C O R D 59

described and figured and is therefore only doubtfully Perhaps Viracarpon represents a genus of an extinct C araceous (Muller 1981). family. However, it would be helpful if its as yet unknown vegetative parts were found. The fruits of V. Inaperturate pollen grains with a striate exine were hexaspermum are sessile on an unbranched axis and reported by Graham (1976) from the Miocene of are bractless. The fruit is hexangular with six locules Veracruz, Mexico and assigned to Spathiphyllum. This around a central core which is slightly longer than is probably correct because other genera of Araceae the locules. Each locule contains a single seed. The with the same pollen type do not occur in Mexico. outer wall of the ovary extends upwards in six peri- anth-like lobes, forming a cup-like structure Pollen similar to that of Spathiphyllum was described (interpretation by Bande & Awasthi, but questionable) from the Palaeocene of Colombia by Hammen & Garcia free at the apex and connate below with a vertical de Mutis (1966) as Ephedripites vanegensis. ridge running the entire length of the middle of the inner surface of each lobe. The inside of the lobes are Pollenites tranquillus from the Eocene was com- densely covered with long hairs. The peduncle has pared with Acorus (Potonie 1934) and was later monocotyledonous bundles. transferred to the form-genus Monocolpopollenites as M. tranquillus by Thomson & Pflug (1953); this pollen Porosia verrucosa (Hickey 1977) of the North probably belongs to the Arecaceae (Nichols, Ames & American Late Cretaceous and Palaeocene has thick, Traverse 1973). reniform leaves and has been compared to Araceae, Lemnaceae and Hydrocharitaceae ; more complete Fossils excluded from the Araceae material is necessary to resolve its relationships (Gregor & Bogner 1989, Crane et al. 1990, Kvaček 1995). Aracispermum Nikitin (type: A. canaliculatum), a seed fossil, may confidently be excluded from the Fossil flowers, probably of Eocene age, have been Araceae, because it is characterized by a very large described from the Indian Deccan Intertrappean series micropylar aperture, probably the site of an aril. All as Sahnipushpam shuklai (syn. Sahnipushpam glan- aroid seeds known have only a very small micropyle. dulosum) and were assigned to the Araceae by Prakash Gregor & Bogner (unpubl.) and Mai (1995) consider & Jain (1964). However, following further examination that the seeds of Aracispermum canaliculatum and of the original material, Gregor & Bogner (1989) have related species belong to the Zingiberaceae. excluded this fossil from the Araceae. The flowers were Aracispermum johnstrupii belongs to the dicot genus always single and abscissed individually (unknown in Myrica (Myricaceae, Kirchheimer 1957). Aracispermum Araceae). The pollen and the structure of the ovary jugatum was transferred by Mai to Caricoidea locules and style are different from those found in (Cyperaceae) as C. jugata (Mai & Walther 1978, see also Araceae. Although the true taxonomic position of Friis 1985). Aracispermum hippuriformis was consid- Sahnipushpam is still unclear, it may nevertheless be ered by Mai (in Mai & Walther 1978) as closely related excluded from the Araceae without reservation. to the genus Alpinia (Zingiberaceae). Ettinghausen (1870) described a fossil aerial root The fossil “strobilus” of Aracistrobus dravertii is in from the Miocene of Croatia as Aronium extinctum fact an old infructescence of Platanus (Platanaceae), and compared it with Anthurium, but this fossil is best showing the scars where the fruits were previously sit- regarded as “incertae sedis”. uated (Gregor & Bogner 1989). The “strobilus” remains of Araceites hungaricus are impressions of the shoot Cyclanthodendron sahnii from the Deccan tips of a species of Pinus in which the needles are Intertrappean beds, considered to be of early Eocene missing (Gregor & Bogner 1989). age, was first described as Palmoxylon sahnii. Later a separate genus, Cyclanthodendron, was established, The fruits and seeds of Campylospermum hord- mainly on the basis of the compound vascular bundles wellensis (syn. Cyrtospermites hordwellensis) belong to (absent in Arecaceae) and assigned to the the genus Visnea (Theaceae), according to Mai (1971). Cyclanthaceae. This fossil has also been compared with the Pandanaceae and Araceae, in which com- The monocotyledonous infructescence Viracarpon pound bundles were observed by French & Tomlinson hexaspermum from the early Eocene Deccan (1986). New material from the same locality showed Intertrappean beds of India has been critically rein- that this fossil is very close to the Strelitziaceae (Biradar vestigated by Bande & Awasthi (1986), based on new & Bonde 1990). well preserved material, and a new reconstruction made. Four species of the genus Viracarpon (includ- Some other fossils which have been thought to ing Shuklanthus) have been described but only two belong to the Araceae have been transferred in previ- (V. hexaspermum and V. elongatum) should be rec- ous palaeobotanical literature to other families: Arecites ognized (after Bande & Awasthi 1986). A relationship trabucci (Arecaceae), Aronites dubius (Coniferae, partly with Araceae, Cyclanthaceae and Pandanaceae has indeterminable), Aroites tallyanus (Coniferae), Pistites been suggested by various authors. New studies show loriformis (Cycadaceae), Aroides stutterdii (Stichoporella that these fossils do not belong to the Araceae or in the Dasycladaceae), Pothocites grantonii (Calamites Cyclanthaceae and also that they cannot be included s.l. in the Calamariaceae). Aroides crassispatha is a in the Pandanaceae (B.C. Stone pers. comm.). very doubtfully assigned fossil from the Permian. 60 T H E G E N E R A O F A R A C E A E

C 20 P H Y L O G E N E T I C R E L AT I O N S H I P S W I T H I N THE MONOCOTYLEDONS The phylogenetic relationships of Araceae to other represents the “type” or “Bauplan” of the monocots, or monocots have never been clear. The fossil record pro- to put it another way, that the Liliiflorae has a basal vides no clues and we have to rely on the comparative phylogenetic position within the monocots as a whole. study of modern taxa. Engler (1920b) recognized clear evidence of the “monocot type” in the floral structure of some genera Some influential earlier authorities, such as Engler & of Araceae (e.g. tribe Potheae) and consequently he Gilg (1919) and Wettstein (1935), grouped the Araceae considered these to be primitive aroids. He even stated together with the Arecaceae and Cyclanthaceae. This that his subfamily Pothoideae differed from the view was based on the common tendency in these fam- Liliaceae essentially only in having a fleshy outer seed ilies to floral reduction, “spathe” development, the integument. In his classification however, Engler (e.g. condensation of the inflorescence into one or several Engler & Gilg 1919) did not associate Araceae and spadices and an associated tendency to pseudanthial Liliaceae closely and it is reasonable to suppose that he inflorescences. For similar reasons, the Araceae have saw the similarities between them in terms of shared also been associated in the past with Pandanaceae and primitive characters (plesiomorphies); i.e. Araceae must Typhales. The link with Typhales was also suggested have arisen by independent evolution from primitive because it had been observed that Sparganium monocot forms. (Sparganiaceae) and Acorus (Acoraceae) are infected by the same rust, Uromyces sparganii. Whether there is a Characters which may be considered primitive in close phylogenetic relationship between Acorus and Araceae by outgroup comparison to Liliiflorae are: Sparganium remains unclear, but since Grayum’s (1987) absence of laticifers, inconspicuous spathe, bisexual widely accepted removal of Acorus from the Araceae, flowers with a perigone of two whorls of 3 free tepals the case for an Araceae-Typhales relationship has been and two whorls of 3 free stamens, basifixed anthers on seriously, if not fatally, weakened. As regards the more-or-less elongated filaments, thecae dehiscing by Arecaceae, Cyclanthaceae and Pandanaceae, their flo- longitudinal slit, monosulcate pollen, syncarpous, 3-loc- ral and vegetative structure differ so strikingly from the ular, superior ovary, several anatropous ovules per Araceae that any resemblances must surely be due to ovary locule, axile placentation, and presence of parallel evolution rather than sister group relationship. endosperm in the mature seed. However, many of these characters can be regarded as primitive in The Dioscoreales (in the sense of Dahlgren et al. Araceae by comparison with the monocots as a whole, 1985, consisting of Dioscoreaceae, Petermanniaceae, which weakens the concept of Liliiflorae as a sister Smilacaceae, Stemonaceae, Taccaceae, Trichopodaceae, group for the Araceae. Trilliaceae) is a group sometimes linked to the Araceae because of a superficial similarity in leaf form and A further problem is that the Liliiflorae are almost venation. The differences, however, are also impres- certainly paraphyletic or polyphyletic. Recent molecu- sive. The Dioscoreales differ from the Araceae in the lar studies (e.g. Duvall et al. 1993, Chase et al. 1995) usually twining habit, different type of reticulate vena- have placed the genus Pleea (Melanthiaceae) near the tion (with numerous densely arranged, scalariform base of the monocot clade, together with the Araceae, cross veins connecting the major veins), inferior Alismatiflorae and Acorus. These studies also indicated ovaries, usually winged, capsular fruit and usually that Lemnaceae belong within the Araceae, a result winged seeds. Furthermore, the Dioscoreales seem recently further supported by a comprehensive analy- unlikely to be monophyletic, which further compli- sis of cpDNA characters of Araceae and Lemnaceae cates an assessment of their relationship to the Araceae. (French et al. 1995). The concept of an origin from a common ancestor In the following discussion, therefore, the possible of the Liliiflorae (sensu Dahlgren et al. 1985), or even sister group relationships of the Araceae are considered from within this taxon has been widely held (e.g. Hallier only in relation to the Acoraceae, the Alismatiflorae 1912, Bessey 1915, Hutchinson 1934, 1959, Novak 1954, and the Lemnaceae, together with an overview of the Kimura 1956, Takhtajan 1959). Hutchinson (1934) went proposed primitive character states of Araceae. so far as to suggest that the Araceae could be derived directly “... from the stock of the tribe Aspidistreae of Relationships with the Acoraceae Liliaceae in which the flowers are arranged in dense spikes (Tupistra, Rohdea, Gonioscypha).” This latter Acorus has long been considered a member of similarity must surely be, however, the result of paral- Araceae, but recently Grayum (1987, 1990) has pre- lel evolution. Lying behind the Liliiflorean origin sented a convincing case for its removal to a separate hypothesis was the very general view that this group P H Y L O G E N E T I C R E L A T I O N S H I P S W I T H I N T H E M O N O C O T Y L E D O N S 61

family, a view which has been widely accepted. The Table 4. Proposed synapomorphies and long list of significant characters by which Acoraceae plesiomorphies of the Araceae. and Araceae are distinguished (Table 3) also strongly suggests that they are not even sister taxa. POSSIBLE SYNAPOMORPHIES • Lack of vessels in stems and leaves (stem vessels Table 3. Characters separating Acoraceae from Araceae. of climbing Araceae derived?) • Presence of tannins • Ethereal oil cells • P2cs sieve tube plastids • Lack of raphides • Distichous leaves • Unifacial ensiform leaves • Presence of spathe • Unique pattern of bud trace insertion • Presence of spadix • Separate vascular systems in peduncle • Flowers lacking floral bracts • Introrse anthers (in Araceae known only in • Protogyny • Anthers extrorse Zamioculcas) • Pollen 2-nucleate when shed • Stellate endothecial thickenings • Tapetum periplasmodial • Tapetal cells with 2–4 nuclei • Tapetal cells 1-nucleate • Secretory anther tapetum • Placentation basal (equivocal) • Exclusively axile vascular supply to placentae • Helobial endosperm development with cellular • Location and structure of ovular trichomes • Presence of perisperm development in micropylar chamber • Dicot-type cellular endosperm development • Embryogeny caryophyllad or solanad • Berries (dehiscent in Lagenandra) • Lack of endosperm (equivocal) The presence of ethereal oil cells, absence of POSSIBLE PLESIOMORPHIES raphides, presence of secretory anther tapetum, pres- • Presence of raphides ence of perisperm and dicot-like cellular endosperm • Absence of silica development are characters which could suggest a • Absence of tricin link to monocot-like dicot families such as Piperaceae • Absence of ethereal oils and Aristolochiaceae. The molecular studies of Duvall • Absence of laticifers et al. (1993) and Chase et al. (1995) now indicate that • Leaves bifacial with petiole and lamina Acorus could be considered the most primitive living • Flowers actinomorphic monocot taxon. • Tepals 4–6 in two whorls of 2 or 3 • Anthers basifixed Primitive characters of the Araceae • Stamen filaments ± elongate • Pollen mother cell cytokinesis successive The removal of Acorus from Araceae has made it • Pollen monosulcate (or at least aperturate) easier to reassess the hypothetical primitive character set • Ovary syncarpous for the family. The basic evolutionary trend within • Ovary superior Araceae has always been viewed as the evolution of uni- • Ovules crassinucellate sexual flowers in monoecious pseudanthoid • Perisperm absent inflorescences from simpler inflorescences with bisexual flowers and a lesser degree of pseudanthial develop- Relationships with Alismatiflorae ment. This hypothesis is supported by our cladistic study (Mayo et al., in prep.). Once this trend is accepted, the Dahlgren and co-workers (Dahlgren & Clifford 1982, evolutionary polarity of certain other important charac- Dahlgren & Rasmussen 1983, Dahlgren, Clifford & Yeo ters can be defined by correlation. Thus most bisexual 1985) argued strongly for a sister relationship of genera lack laticifers and have aperturate pollen. Based Ariflorae (Araceae, Lemnaceae) to Alismatiflorae, point- on our cladistic analyses and a reassessment of some ing out many important differences between Araceae other characters not used in our analysis but discussed and Arecaceae. Grayum (1984, 1990, 1991b, 1992a) by Dahlgren and colleagues (Dahlgren & Clifford 1982, Dahlgren et al. 1985), the primitive character states shown in Table 4 are proposed (see Grayum 1990 for a similar but not identical list). 62 T H E G E N E R A O F A R A C E A E

agreed with Dahlgren’s views; his re-interpretation of Certain other characters adduced by Dahlgren et al. araceous endosperm development as helobial further (1985) to support the Alismatiflorean link must be ruled emphasized a relationship with Alismatiflorae. out in the light of our cladistic study, since they emerge as derived within Araceae – presence of intravaginal The characters that link Alismatiflorae to Araceae, squamules, presence of laticifers and presence of 3- based on our assessments of primitive character states nucleate pollen grains. in Araceae, are given in Table 5. The sister group relationship with Alismatiflorae Among the clearest of these are periplasmodial remains equivocal on this evidence but is a hypothe- tapetum, 1-nucleate tapetal cells and caryophyllad sis which continues to merit serious consideration, and embryogeny, but data for these are based on a very appears stronger than any other yet proposed. The patchy coverage of the two taxa. Many of the other molecular studies of Duvall et al. (1993) and Chase et characters could be interpreted as plesiomorphies for al. (1995) are not completely consistent in grouping the monocots as a whole. Araceae and Alismatiflorae together, but as with the morphological and anatomical data, it also remains Table 5. Proposed synapomorphies (or plesio- one of the strongest available hypotheses based on morphies) of the Alismatiflorae and Araceae. molecular evidence. • Vessels lacking in stems and leaves (stem vessels Relationships with Lemnaceae probably derived in climbing Araceae) The most widely accepted sister group relationship • Leaves distichous (primitive in Araceae, present in of Araceae is that with the Lemnaceae, but this is by no Scheuchzeriaceae, Potamogetonaceae) means universally accepted and has been strongly chal- lenged recently by Grayum (1984, 1990, 1991b, 1992a). • Petiolate and laminar leaves with sheathing base (also in Liliiflorae) The Lemnaceae are considered by most authors to be at least closely related to Araceae, and to have • Presence of spadix (Aponogetonaceae, evolved from them by neotenous reduction of leaf and Juncaginaceae, Potamogetonaceae, Zosteraceae) inflorescence. This view is based, among other charac- ters, on the similarity between seedlings of Pistia and • Stamen filaments distinct (plesiomorphy in Spirodela, embryological characters (Maheshwari 1956, monocots) 1958, Maheshwari & Khanna 1956) and the putative homology between the aroid spathe and the “spathe” • Anthers terminal (plesiomorphy in monocots) of Spirodela and Lemna. Other similarities adduced are • Anthers extrorse (also in Liliiflorae, doubtful in the presence of grooved raphides and operculate seeds in Pistia and Lemnaceae. An important difference, how- Lemnaceae) ever, is that the pollen of Lemnaceae is ulcerate and • Tapetum periplasmodial (unusual in monocots) spinose while in Pistia it is inaperturate and plicate (or • Tapetal cells 1-nucleate (unusual in monocots) ± striate); ulcerate pollen is unknown in Araceae (for • Ovules several per locule (plesiomorphy in mono- Limnobiophyllum see chapter 19). cots?) The three hypotheses considered in more detail here • Ovules anatropous (plesiomorphy in monocots) are shown in Figure 9. This illustrates the need to assess • Placentation basal (equivocal as primitive state in possible synapomorphies (shown as thicker lines) for the following sister group pairs: Lemnaceae-Araceae, Araceae) Lemnaceae-Pistia, and Lemnaceae-monoecious Araceae. • Ovule with nucellar cap (occurs elsewhere in A fourth possibility is a trichotomous monophyletic group composed of Lemnaceae-Araceae- Alismatiflorae. In the monocots) following discussion, character states in Spirodela are • Embryo sac of Polygonum type (plesiomorphy in regarded as representing the primitive condition in Lemnaceae. monocots) • Endosperm development with helobial type of cha- Lemnaceae-Araceae as sister groups lazal chamber (helobial endosperm development Possible synapomorphies are listed in Table 6. The occurs elsewhere in monocots, e.g. Liliiflorae) strongest of these characters is that of endosperm • Embryogeny of caryophyllad type (onagrad and development, which typically combines cellular devel- asterad types commonest in monocots) opment in the micropylar chamber with a haustorial • ?Embryo chlorophyllous (data scarce) chalazal chamber containing a single hypertrophied • ?Endosperm absent (equivocal as primitive state in nucleus (Grayum 1991b). Araceae) • ?Unique type of seedling development (linked to previous character; a similarity between Scheuchzeriaceae, Aponogetonaceae and Araceae with endospermless seeds; equivocal as primitive state in Araceae) P H Y L O G E N E T I C R E L A T I O N S H I P S W I T H I N T H E M O N O C O T Y L E D O N S 63

Lemnaceae Table 7. Possible synapomorphies of the Araceae Lemnaceae and monoecious Araceae. Alismatiflorae • Flowers unisexual (controversial interpretation in Lemnaceae Lemnaceae) monoecious Araceae other Araceae • Pollen 3-nucleate Alismatiflorae • Exine spinose (but pollen aperturate and ulcerate in Lemnaceae) • Asterad embryogeny (primitively caryophyllad in Araceae and Alismatiflorae) • Endosperm starchy (no endosperm in Alismatiflorae) • Base chromosome number x=10 (this number occurs in monoecious Araceae but not in more primitive bisexual genera) Lemnaceae advanced subfamily Aroideae. Unisexual flowers, 3- nucleate and spinose pollen and asterad embryogeny Pistia are all more characteristic of advanced rather than primitive Araceae. A base chromosome number of other Araceae x=10 occurs in monoecious Araceae but not in prim- itive ones. However the aperturate pollen of Figure 9. Alternative sister groups of the Lemnaceae Lemnaceae is ulcerate and this type is unknown in Araceae. Almost all monoecious Araceae have inaper- Table 6. Possible synapomorphies of the turate pollen and so this character is problematic were Lemnaceae and Araceae. Lemnaceae to be inserted well within the monoecious aroid clade (subfamily Aroideae). In a sister group • Flavonols present (rare in Alismatiflorae, absent relationship, however, there is less difficulty. The tribe in Pistia) Zamioculcadeae, which is basal in subfamily Aroideae, has aperturate pollen (and so, occasionally, does tribe • Raphides present (widespread in monocots, Stylochaetoneae – also near basal), and so a position absent in Alismatiflorae and monocot-like dicots) for Lemnaceae near the base of the Aroideae clade is at least conceivable. The molecular data, on the other • Spathe present (controversial interpretation in hand, suggests that Lemnaceae are embedded well Lemnaceae, bract subtending inflorescence within the Aroideae clade (French et al. 1995). On this widespread in monocots) hypothesis, some further explanation will be required for the evolution of the pollen characters. • Outer integument overtopping inner (?distribu- tion in other monocots) Lemnaceae-Pistia as sister groups • Ovary syncarpous (controversial interpretation The major character conflicts and the possible in Lemnaceae, possible symplesiomorphy with synapomorphies of these two taxa are shown in Table 8. Liliiflorae) The six character differences are a major obstacle to • Helobial endosperm development with cellular linking Lemnaceae and Pistia in a sister group rela- development in micropylar chamber (unique in tionship, particularly the pollen and nucellar characters. monocots) The latter is also problematic in grouping Lemnaceae within Araceae since although few genera of Araceae • Endosperm present in ripe seeds (possibly have been studied, none are yet known certainly to derived in Araceae, general in monocots) have crassinucellate ovules. Lemnaceae-monoecious Araceae as sister groups On the other hand, the joint possession of opercu- late seeds and practically identical seedling Possible synapomorphies, additional to those link- development appear to be strong evidence for a close ing Lemnaceae to Araceae as a whole, are listed in relationship. It is nevertheless tempting to regard these Table 7. as convergences resulting from their highly special- ized niche as floating aquatics. The molecular studies A number of lemnaceous characters which do not fit with primitive Araceae agree better with the 64 T H E G E N E R A O F A R A C E A E

of French et al. (1995) place the two taxa in widely sep- Table 8. Character conflicts and possible synapo- arate positions within subfamily Aroideae (as defined morphies of the Lemnaceae and Pistia. in our classification). This hypothesis of sister group relationship thus remains very doubtful. CHARACTER CONFLICTS The Lemnaceae are best considered for the pre- Lemnaceae Pistia sent as a derived offshoot of Araceae with a sister group relationship to the major advanced clade repre- • Stamens free • Stamens connate sented by subfamily Aroideae. There is a great need for more comparative embryological studies of Araceae • Pollen ulcerate • Pollen inaperturate which could help to shed further light on this problem. • Ovules crassinucellate • Ovules tenuinucellate Are the Araceae paraphyletic? • Exine spinose • Exine plicate The weight of evidence, especially the new data from molecular studies, now strongly favours the • Apiose present • Apiose absent inclusion of Lemnaceae within Araceae as merely one subclade of an overall monophyletic group. We • Tannins absent • Tannins present have not, however, included Lemnaceae in our clas- sification nor in our taxonomic treatment. Our POSSIBLE SYNAPOMORPHIES C reasons are pragmatic. The molecular evidence • Grooved raphides emerged at a very late stage in the preparation of this • Floating aquatic (rare in monocots) book. In addition, Landolt (1986) and Landolt & • Stolons arising in lateral pouch Kandeler (1987) have provided a recent comprehen- • Operculate seeds (also in Commelinales and sive taxonomic treatment of Lemnaceae which we certainly could not improve upon. Zingiberales) • Seedling development: median cotyledonar It would also be desirable to carry out a combined analysis of molecular and morphological data to lobe, no primary root, operculum persistent establish the position of Lemnaceae more precisely over root pole, first adventitious roots arising within Araceae. With these qualifications in mind it is around root pole. nonetheless important to emphasize that the family • Fruit not a berry Araceae is almost certainly paraphyletic, while the • Laticifers absent order Arales (Araceae including Lemnaceae) is very • Root hairs absent probably monophyletic. P H Y L O G E N E T I C R E L A T I O N S H I P S W I T H I N T H E M O N O C O T Y L E D O N S 65

C 21 P H Y L O G E N E T I C R E L AT I O N S H I P S W I T H I N A R A C E A E Previous work on the classification of Araceae has anastomosing laticifers, female zone of spadix fused reached a reasonable consensus on the circumscrip- to spathe, etc.) and especially because this taxon tion of the tribes and subtribes. The major difficulties emerged as a monophyletic group in the molecular revolve around the subfamily concepts. These were analysis of French et al. (1995). In contrast, no sup- introduced by Engler (1876b) and have been found port was found for the concept of subfamily very useful by subsequent authors. It is much easier Philodendroideae in the sense of Engler (1920b), to think taxonomically in terms of 8 subfamilies Grayum (1990) or Bogner & Nicolson (1991). rather than 30 tribes. The reluctance of modern authors to abandon the subfamily concept, despite Most currently accepted tribal and subtribal groups the obviously superior taxonomic quality of the tribal stood up well to analysis, thus confirming the prevail- groups, is shown by their constant use in aroid liter- ing view of aroid taxonomists that they are mostly ature. A stage has now been reached in which monophyletic (or natural) groups. The tribal groups confusion abounds. Current classifications (Grayum held together even when chromosome base number, 1990, Bogner & Nicolson 1991, Hay & Mabberley generally considered an important tribal character, was 1991) differ radically in the composition of several excluded from the analysis. subfamilies and it is no longer possible to speak, for example, of subfamilies Aroideae, Lasioideae or The results of a comprehensive cpDNA study by Philodendroideae without citing the author of the French et al. (1995) give a strikingly similar result to system being used. ours in the basic structure of the cladogram. We have therefore incorporated into our classification We therefore carried out a series of cladistic analy- the strongest features of these two independent ses using all genera as terminal taxa, and without analyses and we have not hesitated to modify our assuming any higher groupings at the outset. Our major taxon concepts in the light of their results. classification (chapter 23) is based on the results. We hope this will help to achieve long term stability The details of these analyses will be given in a sep- in aroid classification. arate publication (Mayo et al., in prep.). The purpose of this chapter is to explain the phylogenetic basis of With few exceptions, the suprageneric groups rec- the classification we have adopted. The cladistic ognized and named in our classification represent study used 63 morphological and anatomical char- monophyletic taxon concepts resulting from the cladis- acters and 109 taxa (including 3 outgroups). The tic analyses. We have deliberately chosen not to character data was gleaned from the literature, our overemphasize the internal topologies of these clades, own studies of herbarium, spirit and living material preferring to concentrate on the “main skeleton”. New and from unpublished observations generously sup- studies and analyses will be needed to give a more reli- plied by various colleagues (see acknowledgements). able picture of the internal phylogeny of the various The main outgroups used were Tofieldia tribes and subfamilies recognized here. (Melanthiaceae-Liliiflorae) and Scheuchzeria (Scheuchzeriaceae-Alismatiflorae), representing Figure 11 shows the cladistic relationships of the Liliiflorean versus Alismatiflorean sister group rela- seven subfamilies we recognize, based on a consensus tionships respectively. Figure 2 shows one of the 100 tree of 100 equally parsimonious trees, one of which equally parsimonious trees found in an analysis using is shown in Figure 10. Tofieldia as the outgroup. A. Major Group Proto-Araceae In producing a classification from the cladistic results we have taken the view that groups which This clade consists of subfamilies Gymnostachy- emerged consistently from the analysis should be doideae and Orontioideae and is defined by the considered seriously as named taxa in the classifica- following characters – medium sized pollen, condensed, tion. In cases where traditionally recognized groups non corm-like thickened stem, subterranean stem, usu- failed to emerge in the cladogram it was necessary to ally unilocular ovaries and locules with 1–2 ovules. decide whether this was a sufficient reason to reject them in the classification. In the case of tribe These are rather weak and highly homoplasious Zomicarpeae, for example, we have recognized the synapomorphies, which suggests that the group may tribe despite the fact that it usually failed to emerge be paraphyletic rather than monophyletic. It is note- as a monophyletic group. We justify this because worthy, however, that French et al. (1995) they possess some “good” characters in common (e.g. independently and consistently found the same group using cpDNA data. 66 T H E G E N E R A O F A R A C E A E

1. Subfamily Gymnostachydoideae clade. The tribe Spathiphylleae failed to group consis- tently with these three tribes but does so in the cpDNA Among other peculiar characters, Gymnostachys analysis of French et al. (1995). Our subfamily has linear leaves with parallel venation and a flower- Monsteroideae thus differs from Engler’s only by the ing shoot of unique structure. We therefore prefer to addition of Anadendrum and Heteropsis. keep it in its own monospecific subfamily (following Bogner & Nicolson 1991). 5. Subfamily Lasioideae 2. Subfamily Orontioideae Our subfamily Lasioideae corresponds to tribe Lasieae of earlier systems and is a very stable and con- This group corresponds to tribe Orontieae of pre- sistent clade. vious classifications. The synapomorphies are:– leaf blade expanded not linear, anatropous or hemiana- The synapomorphies are:– monosulcate pollen tropous ovules, endosperm sparse to absent, base (derived by reversal from the inaperturate state), chromosome number x=13. absence of pollen starch, basal ribs of primary veins very well developed, dracontioid leaf margin devel- B. Major Group True Araceae opment, spadix with basipetal flowering sequence, anthers dehiscing by oblique pore-like slits and very This is a previously unrecognized group and is sup- often unilocular ovaries. ported by the following synapomorphies:– conspicuous or flag-like spathe, major internode of Monosulcate pollen is normally regarded as primi- inflorescence between spathe and next leaf below, tive in the family and it is possible that its occurrence continuation shoot in axil of penultimate leaf before as a reversal here may be an artefact contingent on the spathe, leaf blade expanded not linear, basal or near- topology of this particular cladogram (see discussion basal placentation. These characters are strong and under subfamily Aroideae). less homoplasious which suggests that the group is indeed very probably monophyletic. 6. Subfamily Calloideae 3. Subfamily Pothoideae The genus Calla consistently emerges as a single clade and was usually among the basal branches in our Tribe Potheae is a consistent group defined by the analysis. French et al. (1995) also found that Calla following synapomorphies:– monopodial shoot archi- emerged consistently as an independent clade, but in tecture, lack of endosperm, chromosome base number their analysis it occurred further up the tree. The x=12. The genus Anthurium failed to group consis- autapomorphies are:– perigone absent, pollen diaper- tently either with tribe Potheae or any other group in turate, pollen globose, laticifers simple, petiole sheath our analysis. French et al. (1995), however, found that long-ligulate, ovary unilocular, chromosome base num- Anthurium consistently grouped with tribe Potheae ber x=18. Calla seems to be highly autapomorphic and we have adopted this to form the subfamily and its sister relationships remain obscure. Pothoideae. Figure 10 is an example of one family of cladograms which show the two taxa as sister groups, 7. Subfamily Aroideae and in this case they share a single synapomorphy – fine leaf venation with secondary and tertiary veins The most striking feature of the analysis is the large forming mostly cross veins to primaries; the ple- clade which contains all the monoecious genera. This siomorphic condition in Anthurium was assumed to group corresponds to Schott’s “Diclines” (Schott 1860) be that shown in Anthurium sect. Digitinervium. and is not recognized in the classifications of Engler (1876b, 1920b), Grayum (1990), Bogner & Nicolson 4. Subfamily Monsteroideae (1991) and Hay & Mabberley (1991), which all embody the idea that monoeicy and associated advanced spathe In our analysis the genera of the tribes Monstereae, and spadix characters must have evolved several times Heteropsideae and Anadendreae form a single consis- from bisexual-flowered ancestors. tent clade. The synapomorphies are:– spathe undifferentiated into tube and lamina and soon decid- Strong support for our concept comes from the uous or marcescent with distinct basal abscission, DNA work of French et al. (1995) which also pro- perigone connate. The latter character occurs only in duces a single clade for all monoecious taxa. On the Anadendrum, the perigone being lost further up the basis of both studies, we therefore feel confident in advocating the taxonomic recognition of this group as subfamily Aroideae since it represents a major advance and simplification in our understanding of aroid phylogeny. P H Y L O G E N E T I C R E L A T I O N S H I P S W I T H I N A R A C E A E 67

Tofieldia ACORACEAE GYMNOSTACHYDOIDEAE ORONTIOIDEAE CALLOIDEAE ANTHURIEAE POTHEAE SPATHIPHYLLEAE ANADENDREAE HETEROPSIDEAE MONSTEREAE LASIOIDEAE ZAMIOCULCADEAE STYLOCHAETONEAE SPATHICARPEAE Dieffenbachia AGLAONEMATEAE ZANTEDESCHIEAE PHILODENDREAE HOMALOMENEAE ANUBIADEAE Bognera Figure 10. Cladogram of the genera of Araceae. One of 100 equally parsimonious trees (Mayo et al., in prep.). Terminal taxa are shown as subfamilies or tribes when monogeneric or when all genera emerged consistently as a single clade. Generic names are given where the genera of recognised tribes failed to form monophyletic groups in the analysis (see chapter 23 for a full synopsis of the classification). 68 T H E G E N E R A O F A R A C E A E

MONTRICHARDIEAE CULCASIEAE THOMSONIEAE NEPHTHYTIDEAE CALLOPSIDEAE Zomicarpella Ulearum AROPHYTEAE ARISAREAE AMBROSINEAE PISTIEAE Filarum Pinellia Zomicarpa Arisaema AREAE CRYPTOCORYNEAE SCHISMATOGLOTTIDEAE PELTANDREAE Protarum Alocasia Steudnera Colocasia Remusatia Gonatanthus Scaphispatha Ariopsis Caladium Hapaline Jasarum Syngonium Xanthosoma Chlorospatha P H Y L O G E N E T I C R E L A T I O N S H I P S W I T H I N A R A C E A E 69

Major Groups Subfamilies PROTO-ARACEAE GYMNOSTACHYDOIDEAE TRUE ARACEAE 1 genus ORONTIOIDEAE 3 genera POTHOIDEAE 4 genera MONSTEROIDEAE Flowers 12 genera bisexual LASIOIDEAE 10 genera CALLOIDEAE Flowers 1 genus unisexual AROIDEAE 74 genera Figure 11. Cladogram of the subfamilies of the Araceae, based on a consensus tree (Mayo et al., in prep.), showing the Major Groups and subfamilies recognised in the classification, and the distribution of floral sexuality amongst these taxa. There is a question as to precisely where to draw the re-evolution of monosulcate pollen from inaperturate boundary of the subfamily. Our cladogram offers two in subfamily Lasioideae. This seems highly implausible. possibilities which are supported by strong characters. A much more likely arrangement would have inaper- A subfamily Aroideae which excluded the tribes turate pollen evolving between Stylochaeton and tribe Zamioculcadeae and Stylochaetoneae would be defined Spathicarpeae, in association with loss of perigone. by absence of perigone, presence of simple laticifers, Inaperturate pollen in tribe Spathiphylleae and thick stamen connectives and porose anther dehiscence. Anadendrum would then be homoplasic. A further By contrast, subfamily Aroideae including these tribes point here is that V. Tarasevich (pers. comm.) has sug- is defined by unisexual flowers, clear differentiation of gested on the basis of TEM studies that Spathiphyllum the spathe into a tube and blade, and spadix differen- pollen is in fact multi-aperturate. tiated into male and female zones. The latter, more inclusive concept is a better fit with the DNA cladogram The internal topology of our subfamily Aroideae of French et al. (1995). A further consideration is that concept remains largely unresolved above the tribal these characters are of more practical use for distin- level and this is a problem to which future phylogenetic guishing subfamily Aroideae, since unisexuality and studies should devoted. We have, for convenience, gross morphology of the inflorescence is a much more recognized an informal paraphyletic “Perigoniate obvious combination of features than absence of a Aroideae” and a monophyletic “Aperigoniate Aroideae”, perigone, presence of laticifers or small floral characters. the latter supported by strong characters as noted ear- This is therefore the concept we have opted for. lier. Our analysis has thus not made very much further progress in clarifying the relationships of the tribes of Despite our strong advocacy of this taxon, it should the Englerian subfamilies Aroideae, Philodendroideae be pointed out that the distribution of certain charac- and Colocasioideae. ters on the cladogram, especially inaperturate pollen, is unsatisfactory. Inaperturate pollen arises between Under the influence of the molecular studies of tribe Spathiphylleae and the other monsteroid clade French et al. (1995), we have adopted Grayum’s low down on the stem of the cladogram, requiring the (1984,1990) device of grouping certain tribes into infor- mal Alliances (see chapter 23). The Dieffenbachia 70 T H E G E N E R A O F A R A C E A E

Alliance is taken directly from their results. Our analy- ribs, presence of a sympodial leaf submarginal vein, C sis never associated tribes Spathicarpeae and thickened stamen connectives, connate stamens, pres- Dieffenbachieae, but we think this is a very interesting ence of staminodes in the female spadix zone, and possibility and the molecular results give strong sup- base chromosome number of x=14. port to this clade. Recognition of the Philodendron Alliance reflects the fact that both morphological and The other tribes are not arranged into Alliances and molecular analyses gave a similar result. The their sequence reflects only a generalized affinity as we Schismatoglottis Alliance is a strong clade in the mol- prefer to remain non-committal. In our analysis a con- ecular analysis and although not present in our analysis sistent group is formed by Arisarum, Ambrosina and with Tofieldia as outgroup, it was commonly found in Pistia. With Tofieldia as outgroup, tribe Arophyteae analyses with other outgroups. also associated consistently with these three genera, although with other outgroups it emerged in a differ- The Caladium Alliance is a novel group which ent position. French et al. (1995) found a somewhat emerged strongly in the molecular analysis of French similar result, except that tribe Pistieae grouped with et al. (1995). In our morphological analysis tribe the tribes Areae and Colocasieae. We have followed Zomicarpeae was problematic, failing to emerge as a their results in keeping Pistieae separate. monophyletic group and on the whole being associ- ated with the clade including tribe Areae. The results Clearly there are many possibilities for alternative of French and colleagues reconcile the known pres- topologies within subfamily Aroideae. The tribes of ence of anastomosing laticifers in tribe Zomicarpeae subfamily Aroideae are likely to remain, more-or-less with their occurrence in tribe Caladieae, the neotrop- as circumscribed here, as reliable taxonomic units, ical distribution of both tribes and the possibly but it may be expected that in the future they will intermediate status of Scaphispatha between them. On undergo much rearrangement. The cost of our this basis we have adopted their result for our classifi- approach is measured in redundancy in the classifi- cation. One of the consequences of this is that the old cation since there are still many monotypic tribes. A Englerian subfamily Colocasioideae is no longer rec- desirable future objective would be to find partners ognized in the classification, despite the fact that it for these solitary genera since in our opinion a clas- emerges consistently in our morphological analysis. sification is more useful if it emphasizes sister group This clade is associated in our analysis with the tribes relationships rather than degree of anagenesis. Peltandreae and Ariopsideae and this larger clade Monotypic tribes are a way of roughly indicating that (Peltandreae–Chlorospatha in Figure 10) is defined by a genus has no obvious sister group relationships. the following synapomorphies:– short distinct leaf basal These should therefore be priority targets for improv- ing the classification in the future. P H Y L O G E N E T I C R E L A T I O N S H I P S W I T H I N A R A C E A E 71

C 22 P R E V I O U S C L A S S I F I C AT I O N S Excellent surveys have been published by Nicolson cally important characters in common. For Schott, it may (1960a, 1983,1984a, 1988b) and Croat (1992c). We have be supposed that the most significant taxonomic char- discussed only those systems which have, in our opin- acters were the floral ones he had so painstakingly ion, exercised most influence on the progress of elucidated and magnificently illustrated in the Genera araceous systematics. Aroidearum (Schott 1858), where the plates presented each genus as an analysis of its floral structure. Schott, the founding father of the systematics of Araceae, published his first classification in the Engler, however, considered Schott’s classifica- Meletemata botanica (Schott 1832). He grouped 35 tion to be artificial because it was based primarily on genera into the Araceae and placed Gymnostachys and floral characters. His new approach was strongly Acorus in a separate family “Ordo Acoroideae”. Within influenced by the earlier morphological and anatom- his Araceae he divided the genera with unisexual, naked ical studies of Irmisch (1858, 1874) and Tieghem (i.e. perigone absent) flowers into subfamily (“Subordo”) (1867, 1872) and by his own original research on the Androgynanthae, and those with bisexual, perigoniate anatomy and shoot organization of the family (Engler flowers into subfamily Hermaphroditanthae. The order 1877, 1878). In his first complete system Engler of genera progressed essentially from those which, (1876b) presented the first classification based on according to our modern paradigm, had the most phylogenetic principles (see Appendix, Table 11). derived inflorescence and floral types, to those with the The order of the genera was reversed so that those least, beginning with Cryptocoryne and Ambrosina and with the least derived flowers and inflorescences ending with Symplocarpus and Orontium. came first and those with the most derived came last. Ambrosina and Cryptocoryne, for example, were now Although twenty eight years elapsed before Schott placed almost at the end of the system. He recog- (1860) published his final classification in the nized 10 subfamilies, one of which was the Prodromus systematis Aroidearum, now greatly Lemnoideae, corresponding to the modern expanded to include 107 genera, the fundamentals of Lemnaceae. Subfamily Pothoideae was the group his classification remained the same (see Appendix, which represented the most primitive forms, i.e. Table 10). The “Araceae” have now become the those with the most obvious connections to other “Aroideae”, Gymnostachys and Acorus are included as monocot families through the joint possession of a subtribe at the end of the system, and the two sub- such common characters as 3-merous, perigoniate, families are renamed as rankless taxa, the Diclines, bisexual flowers. The subfamilies were defined by a containing those with unisexual, naked flowers and the combination of vegetative and floral characters, with Monoclines, containing those with bisexual, perigoniate a strong emphasis on the presence or absence of flowers. A notable feature of Schott’s Prodromus treat- laticifers and trichosclereids, life form, shoot organi- ment is the extremely detailed family description which zation, leaf venation and phyllotaxis. His subfamily shows that Schott was fully aware of the great variety classification embodied the idea of the evolution of of vegetative characters of the Araceae, including latex unisexual-flowered genera from bisexual-flowered (“succo decolori l. lacteo”) and, probably, trichosclereids ones in several independent phylogenetic lines, or (“rhaphidibus”). His suprageneric taxa are, neverthe- clades as we would say today. less, based almost entirely on floral characters. Only the tribes and subtribes are ranked. The higher, unranked Engler’s attitude to Schott’s system and the phylo- taxa have names which describe their most important genetic method he applied to his own classification of diagnostic character, e.g. “Efilamentatae”: filaments the Araceae are summarized explicitly in his 1876 absent, “Pachyzeugmaticae”: large anther connectives, paper, which includes detailed dendrograms showing “Stenozeugmaticae”: slender anther connectives, presumed phylogenetic relationships (orig. “Orthotropooae”: orthotropous ovules, “Anatropooae”: “Verwandtschaft”) between suprageneric taxa. The anatropous ovules, “Gymnogoneae”: gynoecium naked, introductory paragraph is worth quoting in full (freely and “Peristatogoneae”: gynoecium with associated sta- translated from German) for the light it sheds on minodes. Schott (1860) used vegetative characters Engler’s thinking:– occasionally in the diagnoses of his tribes and subtribes and much more extensively in the generic descriptions. “Natural system of the Araceae. In the following overview of the Araceae, which There is no evidence that Schott used evolutionary is systematic rather than analytical, the [suprageneric] principles in his work. His philosophy was clearly that of groups and genera are arranged so as to convey an classification by “natural affinities”, i.e. grouping together idea of the gradual reduction in floral parts [which is those taxa which have the largest number of taxonomi- seen in the family]. This system makes it possible to 72 T H E G E N E R A O F A R A C E A E

perceive with particular clarity that reduction in the hemisphere distribution and preference for swampy floral parts must have occurred in various [supra- habitats, subfamily Lasioideae by their frequent pos- generic] groups and thus that Schott’s classification, in session of deeply sagittate or dracontioid leaves, which the primary subgroups are based on the floral subfamily Philodendroideae by their unisexual flow- structure, is unnatural. My system of the Araceae ers and parallel-pinnate leaf venation, subfamily would have been quite different had I been aiming to Colocasioideae by their unisexual flowers, anasto- provide an aid to identification for botanists less famil- mosing laticifers and special type of leaf venation iar with the family. My intention is otherwise: to make (“colocasioid”; see glossary), and subfamily Aroideae as clear as possible all the phylogenetic relationships by their unisexual flowers, mostly geophytic habit [orig. “verwandtschaftlichen Beziehungen”] existing and frequent possession of a smooth terminal spadix between the individual [suprageneric] groups. The appendix. This classification has been the basis for number of suprageneric groups is consequently larger most subsequent taxonomic work on the family and than might at first sight seem necessary. The Araceae it is only in recent years that it has undergone sub- is a poorly represented family in herbaria. Our know- stantial alteration. ledge of its forms continues to be incomplete, as is shown by the almost annual discovery of new genera, To begin with, only minor modifications were and it may be expected that some of the suprageneric made. Bogner (1979a) published a synopsis in which groups comprising only a few genera will later be new genera were inserted into Engler’s framework and enriched by the inclusion of one or several [new] new generic synonymy accounted for. Shortly after- ones. All [literature] citations are here omitted since wards, however, French and French & Tomlinson these will be found in my monographic systematic began to publish a series of anatomical studies which treatment of the Araceae [he must have been referring suggested that some parts of Engler’s classification to his forthcoming monograph in de Candolle’s were unnatural, particularly his subfamilies Pothoideae Monographiae Phanerogamarum – see Engler 1879] and Lasioideae. Engler’s concept of araceous phy- and are here of little interest. In those cases where the logeny allowed for the evolution of unisexual-flowered origin of one genus from another is [considered] genera from bisexual-flowered ones within these two highly probable the names are arranged one above subfamilies and thus he saw no difficulty in including the other. Where only a common origin [for the gen- unisexual-flowered genera such as Culcasia and tribe era] is assumed, their names are placed side-by-side. Zamioculcadeae within subfamily Pothoideae, and Generic names in parentheses refer to taxa considered likewise Amorphophallus, Anchomanes and others as genera by Schott but which most probably should within subfamily Lasioideae. The new anatomical evi- be considered only as subgenera.” dence, however, began to reveal that such placements did not reflect real relationships. This earliest classification represents the most cre- ative phase of Engler’s Araceae work. It was produced Two earlier classifications should be mentioned during the period when he was most competely which essentially followed the principles of Schott’s immersed in his studies of the family and was not yet system, those of Hooker (1883) and Hutchinson (1934, burdened by the immensely ambitious projects of his 1959, 1973). Hooker (1883) presented a classification later career. The dendrograms, published again in de which was essentially an updated version of Schott’s Candolle’s Monographiae Phanerogamarum (Engler (1860) system in the light of Engler’s new studies. Like 1879), were by far the most detailed cladistic state- Schott’s (1860), the classification starts with the uni- ment that he ever published. Later on, following the sexual-flowered genera and ends with the discovery and description of many new genera and bisexual-flowered ones. Hooker recognized 11 tribes innumerable new species, he modified his system but no subfamilies, and placed all the most primitive somewhat (Engler 1887-1889, 1920b), in particular by genera (Orontium, Lysichiton, Symplocarpus, Lasia, the exclusion of Lemnaceae, but the essential struc- Podolasia, Urospatha, Anaphyllum, Ophione (= ture continued unaltered. Dracontium), Cyrtosperma, Spathiphyllum, Anthurium, Pothos, Pothoidium, Acorus and Gymnostachys) in the In his final classification, Engler (1920b) reduced final tribe, Orontieae, divided into seven subtribes. The the number of subfamilies to eight (see Appendix, 98 genera recognized have similar circumscriptions to Table 12). These are based on a rather broad range those of Schott and Engler. of taxonomic characters but for the most part are not very sharply defined. As Engler had made clear from Hutchinson (1934, 1959, 1973) based his system on the outset, they were concepts intended to convey Hooker’s but reversed the order of the genera to reflect phylogenetic meaning rather than ease the path of a phylogenetic sequence, starting with the bisexual- identification. In a rough and ready manner subse- flowered genera and ending with the unisexual-flowered quent specialists of the family learned to recognize ones. The first tribes are Acoreae, Orontieae and subfamily Pothoideae by their complete lack of lati- Spathiphylleae and the last is Areae. He recognized 18 cifers, subfamily Monsteroideae by their tribes and 126 genera but no subfamilies. The major trichosclereids and mostly aperigoniate bisexual flow- contribution of his classification was the publication of ers, subfamily Calloideae by their temperate Northern a key to all the genera in English. Hutchinson’s system was used, notably, by Thanikaimoni (1969) and P R E V I O U S C L A S S I F I C A T I O N S 73

Marchant (1970, 1971a, b, 1972, 1973), but in general it different clades. His subfamilies Pothoideae and made little impact on the prevailing use of the Engler Lasioideae both include unisexual-flowered genera system. There are a number of unnatural tribal circum- (tribe Zamioculcadeae and Stylochaeton respectively). scriptions, for example, the separation of Symplocarpus His subfamilies Calloideae (Philodendroideae), from Lysichiton and Orontium, of Holochlamys from Colocasioideae and Aroideae consist almost entirely of Spathiphyllum and of Pycnospatha from Dracontium unisexual-flowered genera as in Engler’s system. and related genera. The treatments of Hooker and Hutchinson were parts of general classifications of the Grayum’s work has been crucial to the re-evaluation Flowering Plants rather than specialist studies as in the of the family classification. Not only was it based on a case of most other systems mentioned here. thorough review of the available taxonomic data from many character fields, both old and new, but he Hotta (1970) published a classification of Araceae employed modern analysis to produce his classification of east Asia and the Malaysian region which was based and was bold in proposing major changes backed up on Engler’s system. Among other original ideas, it is by cogent arguments. Grayum presented extensive dis- notable for the recognition of the subfamily Acoroideae cussion and a masterful revision of the literature but did (Acorus and Gymnostachys) and the sinking of sub- not summarize his conclusions in the form of a detailed family Monsteroideae into subfamily Pothoideae. He taxonomic presentation with diagnoses and keys to transferred Lysichiton and Symplocarpus (as tribe his generic and suprageneric taxa. Symplocarpeae) to subfamily Lasioideae and Calla to subfamily Philodendroideae. All these concepts were Bogner & Nicolson (1991) presented a classification adopted, at least in part, by later authors, such as which adhered more closely to Engler’s subfamily con- Grayum (1984, 1990) and Bogner & Nicolson (1991). cepts but which likewise took account of new data, especially that of French (see Appendix, Table 14). Grayum (1984,1990) published a comprehensive Their primary aim was to provide a detailed synopti- survey of aroid taxonomic characters, an original com- cal key to the genera, in which the diagnostic parative survey of pollen morphology and a cladistic characters were clearly presented. The mass of new analysis of the entire family. He presented a very dif- data on which their changes were based was not ferent-looking picture of the family’s taxonomy to that detailed in the form of literature citations. They fol- of Engler (see Appendix, Table 13). In the first place lowed Grayum in eliminating Acorus, but in addition he argued forcefully for the exclusion of Acorus from they elevated Gymnostachys to subfamilial rank. the family altogether, which strengthened consider- Subfamily Pothoideae was greatly altered. All unisex- ably the homogeneity of the Araceae as a result. Major ual-flowered genera were ejected to other subfamilies alterations of all the subfamilies were proposed, and Anadendrum and Heteropsis were transferred to including those suggested by Hotta (1970). Engler’s subfamily Monsteroideae, leaving only three closely subfamily Pothoideae and subfamily Monsteroideae related genera (Pothos, Pedicellarum and Pothoidium). were merged. Subfamily Calloideae was split, with Subfamily Monsteroideae acquired the circumscrip- Calla going to join the genera of the old subfamily tion accepted in our treatment by the addition of Philodendroideae and the three genera of Engler’s Anadendrum and Heteropsis. Subfamily Calloideae tribe Symplocarpeae being transferred to the subfam- was reduced to include only Calla, while subfamily ily Lasioideae. Engler’s tribe Amorphophalleae (= tribe Lasioideae now included tribes Orontieae (= Engler’s Thomsonieae) was shifted to subfamily Aroideae, and tribe Symplocarpeae), Anthurium, Zamioculcadeae, his tribes Nephthytideae and Montrichardieae were Callopsideae, Nephthytideae, Culcasieae and moved to the subfamily Philodendroideae (= sub- Montrichardieae, as well as the “core” tribe Lasieae. family Calloideae). The Lasioideae also acquired Subfamilies Philodendroideae, Colocasioideae and Stylochaeton from tribe Aroideae. Subfamily Aroideae, however, remained very much as Engler had Philodendroideae was renamed Calloideae, because of left them, apart from the inclusion of more recently the inclusion of Calla, and enlarged yet further with described genera and the transfer of Protarum to sub- the inclusion of tribes Spathicarpeae (Engler’s family Colocasioideae, and tribe Thomsonieae to Asterostigmateae), Callopsideae, Arophyteae and subfamily Aroideae. Pistia remained in its own sub- Culcasieae. Subfamily Colocasioideae changed only family as proposed by Engler. slightly with the addition of Zomicarpa. Subfamily Aroideae was modified by the inclusion of Pistia as a Bogner & Nicolson’s paper was the first detailed tribe, tribe Thomsonieae and Ariopsis. taxonomic synopsis down to the level of genera pub- lished by specialists of the family since Engler’s time. These changes eliminated certain characters which It was the result of many years of study and a profound had helped to preserve the integrity of Engler’s sub- and first-hand knowledge of the taxonomic characters families. In particular, parallel-pinnate leaf venation of the genera. Although these authors proposed no ceased to be a useful character for defining subfamily explicit phylogenetic scheme, their simplification of Philodendroideae (= Grayum’s subfamily Calloideae). subfamilies Pothoideae and Calloideae reduced the However, Grayum preserved the essentially Englerian number of subfamilies which included both unisex- concept of independent evolution of unisexual-flow- ual-flowered and bisexual-flowered genera to one, ered groups from bisexual-flowered ancestors in various their much-expanded subfamily Lasioideae. 74 T H E G E N E R A O F A R A C E A E

Hay & Mabberley (1991) took a completely differ- presented a classification based largely on Grayum C ent approach to the question of araceous phylogeny, (1990) but differing in a number of significant starting out from the precepts of the durian theory of respects. The Lasioideae are placed first, and the Corner (1949, 1952, 1953, 1954a, 1954b). In a long Pothoideae and Monsteroideae are recognized as sep- paper bursting with fascinating if unorthodox ideas, arate subfamilies, the former including Anthurium, they drew a picture of an ancestral araceous type and Anadendrum, Heteropsis and the Zamioculcadeae and recognized many features of the Lasieae (our the latter with only the Monstereae and the Lasioideae) as primitive, partly through their corre- Spathiphylleae. The Aroideae retains the Spathicarpeae, spondence with Corner’s hypotheses of primitive as in Engler’s system, but also includes Ariopsis, angiosperm characters. One point they bring out is Thomsonieae and Pistia. that an argument for adaptation in a particular char- acter state does not necessarily support a hypothesis The classification presented here (chapter 23), that it is derived in the taxon under consideration. together with the results of the molecular studies of Thus, there is no reason in principle why the ances- French et al. (1995), differs from other modern systems tral aroid should not have had a tuberous stem and and that of Engler in grouping all unisexual-flowered unilocular ovaries with a single basally inserted genera together, as Schott did. Unlike Schott, however, ovule. They proposed that the araceous inflorescence our proposal is explicitly phylogenetic. In essence, it may have been the result of homoeotic saltation from appears that the configuration of the available taxo- Nymphaealean ancestral forms and therefore that the nomic data favours a simpler hypothesis than Engler typically monocotyledonous features observed in some proposed, and indeed modern results flatly contradict genera (i.e. those that we consider plesiomorphic Engler’s view of araceous phylogeny. It is not necessary in Araceae) are really derived convergences. They to hypothesize the multiple evolution of unisexual flow- ers to arrive at a natural classification of Araceae. P R E V I O U S C L A S S I F I C A T I O N S 75

BTAXONOMY

C 23 S Y N O P S I S O F T H E C L A S S I F I C AT I O N O F A R A C E A E Family Araceae Jussieu V. Subfamily Lasioideae Engler 21. Dracontium L. MAJOR GROUP PROTO-ARACEAE 22. Dracontioides Engler 23. Anaphyllopsis A. Hay a. Flowers bisexual 24. Pycnospatha Gagnepain 25. Anaphyllum Schott I. Subfamily Gymnostachydoideae Bogner & 26. Cyrtosperma Griffith Nicolson 27. Lasimorpha Schott 28. Podolasia N.E. Brown 1. Gymnostachys R. Brown 29. Lasia Loureiro 30. Urospatha Schott II. Subfamily Orontioideae Mayo, Bogner & P.C. Boyce VI. Subfamily Calloideae Endlicher 31. Calla L. 2. Orontium L. 3. Lysichiton Schott b. Flowers unisexual 4. Symplocarpus Nuttall VII. Subfamily Aroideae MAJOR GROUP TRUE ARACEAE PARAPHYLETIC GROUP: PERIGONIATE a. Flowers bisexual AROIDEAE (perigone present) III. Subfamily Pothoideae Engler Tribe Zamioculcadeae Engler 32. Zamioculcas Schott Tribe Potheae Engler 33. Gonatopus Engler 5. Pothos L. 6. Pedicellarum M. Hotta Tribe Stylochaetoneae Schott 7. Pothoidium Schott 34. Stylochaeton Leprieur Tribe Anthurieae Engler MONOPHYLETIC GROUP: APERIGONIATE 8. Anthurium Schott AROIDEAE (perigone absent) IV. Subfamily Monsteroideae Engler Dieffenbachia Alliance Tribe Spathiphylleae Engler Tribe Dieffenbachieae Engler 9. Spathiphyllum Schott 35. Dieffenbachia Schott 10. Holochlamys Engler 36. Bognera Mayo & Nicolson Tribe Anadendreae Bogner & French Tribe Spathicarpeae Schott 11. Anadendrum Schott 37. Mangonia Schott 38. Taccarum Schott Tribe Heteropsideae Engler 39. Asterostigma F.E.L. Fischer & C.A. Meyer 12. Heteropsis Kunth 40. Gorgonidium Schott 41. Synandrospadix Engler Tribe Monstereae Engler 42. Gearum N.E. Brown 13. Amydrium Schott 43. Spathantheum Schott 14. Rhaphidophora Hasskarl 44. Spathicarpa W.J. Hooker 15. Epipremnum Schott 16. Scindapsus Schott 17. Monstera Adanson 18. Alloschemone Schott 19. Rhodospatha Poeppig 20. Stenospermation Schott 78 T H E G E N E R A O F A R A C E A E

Philodendron Alliance Tribe Culcasieae Engler C 74. Culcasia Palisot de Beauvois Tribe Philodendreae Schott 75. Cercestis Schott 45. Philodendron Schott Tribe Montrichardieae Engler Tribe Homalomeneae M. Hotta 76. Montrichardia H. Crüger 46. Furtadoa M. Hotta 47. Homalomena Schott Tribe Zantedeschieae Engler 77. Zantedeschia K. Sprengel Tribe Anubiadeae Engler 48. Anubias Schott Tribe Callopsideae Engler 78. Callopsis Engler Schismatoglottis Alliance Tribe Thomsonieae Blume Tribe Schismatoglottideae Nakai 79. Amorphophallus Decaisne 49. Schismatoglottis Zollinger & Moritzi 80. Pseudodracontium N.E. Brown 50. Piptospatha N.E. Brown 51. Hottarum Bogner & Nicolson Tribe Arophyteae Bogner 52. Bucephalandra Schott 81. Arophyton Jumelle 53. Phymatarum M. Hotta 82. Carlephyton Jumelle 54. Aridarum Ridley 83. Colletogyne Buchet 55. Heteroaridarum M. Hotta Tribe Peltandreae Engler Tribe Cryptocoryneae Blume 84. Peltandra Rafinesque 56. Lagenandra Dalzell 85. Typhonodorum Schott 57. Cryptocoryne Wydler Tribe Arisareae Dumortier Caladium Alliance 86. Arisarum P. Miller Tribe Zomicarpeae Schott Tribe Ambrosineae Schott 58. Zomicarpa Schott 87. Ambrosina Bassi 59. Zomicarpella N.E. Brown 60. Ulearum Engler Tribe Areae 61. Filarum Nicolson 88. Arum L. 89. Eminium (Blume) Schott Tribe Caladieae Schott 90. Dracunculus P. Miller 62. Scaphispatha Schott 91. Helicodiceros K. Koch 63. Caladium Ventenat 92. Theriophonum Blume 64. Jasarum Bunting 93. Typhonium Schott 65. Xanthosoma Schott 94. Sauromatum Schott 66. Chlorospatha Engler 95. Lazarum A. Hay 67. Syngonium Schott 96. Biarum Schott 68. Hapaline Schott Tribe Arisaemateae Nakai No Alliance 97. Pinellia Tenore 98. Arisaema Martius Tribe Nephthytideae Engler 69. Nephthytis Schott Tribe Colocasieae Engler 70. Anchomanes Schott 99. Ariopsis Nimmo 71. Pseudohydrosme Engler 100. Protarum Engler Tribe Aglaonemateae Engler 101. Steudnera K. Koch 72. Aglaonema Schott 102. Remusatia Schott 73. Aglaodorum Schott 103. Colocasia Schott 104. Alocasia G. Don Tribe Pistieae Blume 105. Pistia L. S Y N O P S I S O F T H E C L A S S I F I C A T I O N 79

C 24 FA M I L Y D E S C R I P T I O N O F A R A C E A E Family Araceae shoots, which are stolon-like branches with leaves reduced to small cataphylls and very elongated intern- Araceae Juss., Gen. Pl. 23 (1789, “Aroideae”), nom. odes, adapted to seek out new host trees; other cons. adaptative forms are subterranean stolons (e.g. Spathiphyllum, Lasimorpha) and tubercules (e.g. ANATOMY: Calcium oxalate raphides and druses abun- Dracontium), bulbils with recurved scales borne on dant, raphides always present, laticifers commonly specialized shoots (Remusatia), bulbils on petioles present, either simple and articulated or more rarely (Pinellia ternata) or leaf blade (Amorphophallus bulb- anastomosing, trichosclereids present (Monstereae, ifer), or formation of new plants from abscissed leaflets Spathiphylleae, rarely in Potheae), tannin cells com- (Zamioculcadeae). ROOTS: always adventitious, pri- mon, resin canals sometimes present (Culcasieae, mary root withering soon after germination, sometimes Homalomeneae, Philodendreae). HABIT: evergreen to dimorphic (climbing hemiepiphytes) with anchor roots seasonally dormant herbs, perennial, sometimes gigan- and larger feeder roots, sometime contractile roots pre- tic, climbing or subshrubby hemiepiphytes, epiphytes, sent (geophytes), rarely roots very fleshy, water-storing lithophytes, terrestrial, geophytes, helophytes, some- (Stylochaeton). LEAVES: usually spirally arranged, times rheophytes, true aquatics, rarely free-floating sometimes distichous; normally differentiated into peti- (Pistia). STEM: aerial and erect to climbing or creeping ole and expanded blade (except e.g. Gymnostachys, with very short (plant rosulate) to very long (plant some Biarum spp.), usually glabrous, rarely pubes- scandent) internodes, or subterranean and consisting of cent, tomentose, villous or with small to large and a subglobose to depressed-globose tuber (sometimes complex trichomes or papillae (e.g. Philodendron turnip- or carrot-like or irregular in shape) or horizon- squamiferum) on the petiole; ptyxis usually convo- tal to erect rhizome; terrestrial plants and helophytes lute, rarely involute (e.g. Anthurium sect. sometimes arborescent with massive stem and terminal Pachyneurium, Lagenandra); blade and petiole often rosette of leaves (Xanthosoma, Alocasia, Montrich- variegated or mottled with spots, bands, blotches or ardia, Philodendron) or arborescent with a pseudostem irregularly shaped patches and zones of various of petiole sheaths (large in Typhonodorum, small in colours, usually shades and mixtures of green, yellow many Arisaema); geophytes often with solitary leaf. and silver. CATAPHYLLS: caducous, marcescent, decid- SHOOT ORGANIZATION: the mature, flowering stem is uous or persistent, sometimes beautifully mottled and almost always a sympodium composed of a series of patterned (e.g. Arisaema, Asterostigma), when persis- articles, rarely the stem is monopodial (Potheae, tent sometimes a conspicuous feature of plant and Heteropsideae); each article begins with a 2-keeled either membranous or forming fibrous mass (e.g. many (except Orontioideae) prophyll followed by a series of spp. of Anthurium, Philodendron). PETIOLE: often as leaves and terminates with an inflorescence; leaf num- long as or longer than blade, usually smooth, some- ber per article may be determinate or indeterminate, times hairy, papillose, warty, prickly or aculeate (e.g. and from one or very few to very many; the leaves of Lasioideae), occasionally covered with large multicel- each article normally consist of a mixture of foliage lular processes (e.g. Philodendron squamiferum), leaves with partially to fully developed blades and cat- rarely massively succulent and water-storing (e.g. aphylls; the sympodial leaf is that subtending the Zamioculcas, Philodendron martianum), often genic- inflorescence and may be a foliage leaf or a cataphyll; ulate (pulvinate) apically (e.g. Anthurium), basally or the prophyll is almost always a cataphyll (except rarely centrally (e.g. Gonatopus boivinii); sheath nor- Orontioideae); subsequent articles (continuation mally well-developed, often at least half as long as shoots) normally arise at the second node below the entire petiole, sometimes ligulate apically, often very spathe node (except Orontioideae); juvenile shoots reduced in sympodial leaves (especially in Anthurium, and flagelliform branches are usually monopodial; ter- most Philodendron spp.). LEAF BLADE: simple to com- minal inflorescences may be solitary or may form a pound, extremely variable in shape – rarely filiform floral sympodium of several inflorescences; the arti- (e.g. Cryptocoryne consobrina), linear (Jasarum), most cles of floral sympodia normally consist of a single commonly elliptic, ovate, oblong, sagittate, hastate, 2-keeled prophyll and an inflorescence; the subse- less commonly trifid to trisect, pedatifid to pedatisect, quent article of a floral sympodium normally arises in radiatisect, dracontioid (i.e. trisect with each primary the axil of the preceding prophyll, i.e. at the first node division further much divided), pinnatifid to pinnatisect below the spathe node. VEGETATIVE PROPAGATION: (Zamioculcas), bipinnatifid, tripinnatifid to quadripin- climbing hemiepiphytes frequently form flagelliform natifid (Gonatopus), fenestrate (Monstera) or laciniate 80 T H E G E N E R A O F A R A C E A E

(i.e. fenestrate with slit-like holes, Cercestis mirabilis); Philodendron, Homalomena) or whole spathe gradu- heteroblasty frequent, especially in climbing hemiepi- ally withering and rotting (most Areae). SPADIX: phytes, “shingle leaves” sometimes formed (e.g. some usually erect, often fleshy and relatively thick, sessile Potheae, some Monstereae); seedling leaves usually or shortly stipitate, rarely very long-stipitate (e.g. entire when epigeal, rarely first foliage leaf compound Lysichiton, Orontium, some Anthurium spp.), usually (Gonatopus, Amorphophallus). LEAF VENATION: free, sometimes adnate basally (e.g. Hapaline, midrib almost always differentiated, sometimes massive Dieffenbachia) or entirely (Spathicarpa) to spathe, and succulent (e.g. Philodendron crassinervium); pri- either ± uniform in appearance (flowers bisexual or mary veins usually arising pinnately from midrib (and monoclinous), or divided into distinct floral zones then called primary lateral veins), either running into (flowers unisexual or diclinous), fertile zones contigu- marginal vein (e.g. Philodendron, Dieffenbachia) or ous or separated by sterile zones, female (pistillate) joining distally to form a submarginal collective vein on zone always basal and male (staminate) zone either each side (e.g. Caladium, many Anthurium spp.), apical or intermediate in position (except Spathicarpa), sometimes primary veins all arising from petiole inser- rarely bisexual flowers occur between male and female tion and running arcuately into leaf apex (e.g. zones (e.g. Arophyteae); sterile zones may be basal, Orontium, Anthurium sect. Digitinervium), rarely intermediate or apical or any combination of these, api- strictly parallel (Gymnostachys) or subparallel (Pistia), cal sterile zone usually known as a terminal appendix; sometimes not differentiated at all (e.g. Philodendron rarely a single plant produces inflorescences bearing crassinervium); secondary and tertiary veins either male flowers only, followed in later years by inflores- reticulate (e.g. Areae), or parallel-pinnate, i.e. running cences bearing female flowers only, and vice versa parallel to primaries (e.g. Philodendron), or arising (paradioecy, known only in Arisaema). FLOWERS: 2- to from primaries at a wide angle and then arching 3-merous (-mery often hard to detect in unisexual flow- strongly towards leaf margin (e.g. Colocasia), some- ers), bisexual (monoclinous, hermaphrodite) or times forming sinuous or zig-zag interprimary veins unisexual (diclinous), very small, protogynous, lacking (e.g. Caladieae); higher order venation reticulated or floral bracts, usually numerous (except e.g. Pistia, forming cross connections between lower order veins. Ambrosina), sessile (except Pedicellarum), usually INFLORESCENCE: terminal, solitary, or 2 to many in a densely arranged, sometimes laxly so (e.g. some Pothos synflorescence, usually appearing to be axillary to sym- spp., male flowers of Arisarum, female flowers of podial leaf, consisting of a spadix (spike) of small Dieffenbachieae); bisexual flowers with or without flowers and subtended by a spathe (bract), usually (Calloideae, Monstereae) a perigone (perianth), uni- erect, sometimes pendent (e.g. Anthurium wendlin- sexual flowers usually without a perigone (present in geri, Stenospermation, Piptospatha), sometimes Zamioculcadeae, Stylochaeton) but sometimes includ- becoming pendent after anthesis (e.g. Typhonodorum). ing rudimentary organs representing modified sexual PEDUNCLE: very short to very long, usually similar to parts of the other sex (e.g. staminodes of female flow- petiole in appearance, coloration, pubescence or arma- ers in Dieffenbachia and Spathicarpeae, cup-like ture, normally longer than spadix stipe, sometimes ± synandrodium in Arophyteae, pistillodes present in suppressed and spadix stipe elongated (Orontium, male flowers of e.g. Stylochaeton, Furtadoa, some Lysichiton). SPATHE: nearly always conspicuous Spathicarpeae spp., central stigmatoid body of the (except Gymnostachys, Orontium), very variable in synandrium present in Taccarum and some shape and colour, simpler forms (e.g. many Anthurium Gorgonidium spp.). PERIGONE: composed of free (e.g. spp.) often green, reflexed or spreading, more complex Anthurium) or partially connate (some Pothos spp.) forms often showy and highly coloured, erect, usually tepals, or consisting of a single cup-like structure (e.g. either boat-shaped or constricted centrally to form a Spathiphyllum cannifolium); when free, tepals 4 to 6 basal tube and an apical blade; tube may enclose the (–8) and imbricate in 2 whorls, membranaceous (e.g. female zone of the spadix or both fertile zones or Anadendrum) or more commonly thickened at least rarely the entire spadix (e.g. Cryptocoryneae), very apically, truncate (Zamioculcadeae) to cucullate occasionally much longer than blade (e.g. many (Lasioideae). STAMENS (bisexual perigoniate, bisex- Cryptocoryne spp.), tube margins usually convolute, ual non-perigoniate, unisexual perigoniate flowers): sometimes connate (e.g. Sauromatum, Stylochaeton, usually free (filaments connate in Gonatopus and often Arisarum); blade usually erect and gaping, sometimes in Lasimorpha), equal in number and opposite to widely spreading, twisted, reflexed or rarely margins ± tepals (when present), rarely more (e.g. some closed forming slit-like opening (e.g. most Lagenandra Dracontium spp.); filaments distinct, often ± oblong spp.); spathe constriction may lie between or above and flattened (e.g. Anthurium), rarely filiform male and female zones or occur in two places (e.g. (Stylochaeton), usually rapidly elongating to push some Remusatia spp.); spathe entirely deciduous soon anthers above perigone or gynoecium at anthesis; after anthesis (e.g. most Monstereae), or tube persistent anthers usually terminal, basifixed, extrorse (introrse in to fruiting and blade marcescent to deciduous after Zamioculcas, latrorse in Pedicellarum), always com- anthesis (Caladieae, Colocasieae, Schismatoglottideae), posed of 2 thecae each with 2 microsporangia; or spathe entirely persistent until fruiting (e.g. connective usually slender, inconspicuous, often over- F A M I L Y D E S C R I P T I O N 81

topped by thecae, thecae dehiscing by single longitu- with a single clavate staminode (e.g. Homalomeneae, dinal slit or apical stomial pore, with all intermediate Schismatoglottis). GYNOECIUM (bisexual and uni- degrees occurring. MALE FLOWER (unisexual non- sexual flowers): ovary usually 1–3 locular, rarely more perigoniate flowers): 1–8 androus (rarely more, e.g. (e.g. Philodendron, most Spathicarpeae), 1-locular Alocasia brisbanensis), floral grouping of stamens ovaries probably always pseudomonomerous; ovules sometimes obvious in mature inflorescence (e.g. many 1–many per locule, orthotropous, hemiorthotropous, Philodendron and Homalomena spp.), often obscured campylotropous, amphitropous, hemianatropous or during ontogeny; stamens free or partially to com- anatropous; placenta 1–several, axile, parietal, apical, pletely connate to form a synandrium. FREE STAMENS basal, or basal and apical; stylar region (tissue lying (unisexual non-perigoniate flowers): usually sessile to between ovary and stigmatic epidermis) usually well- subsessile, filament sometimes distinct (e.g. developed, usually at least as broad as ovary, Schismatoglottis), connective sometimes ± slender (e.g. sometimes attenuate and elongate (e.g. many Areae) but often strongly thickened, apically broad, Amorphophallus, Arisaema, Biarum spp., Dracontium, fleshy and probably osmophoric (e.g. Philodendron), some Spathiphyllum spp.) or massive and truncate thecae lying opposite or adjacent on one side of sta- (most Monstereae spp.), rarely dilated and connate men, dehiscing by single longitudinal slit or apical with those of neighbouring gynoecia (Xanthosoma); stomial pore, with all intermediate degrees occurring, stigma hemispheric, capitate, discoid, umbonate, rarely both microsporangia dehiscing independently more-or-less strongly lobed (e.g. some by separate stomial pores (some Amorphophallus spp.), Amorphophallus, Dieffenbachia spp.), rarely stel- rarely theca prolonged apically into a horn dehiscing by lately lobed (Asterostigma), sometimes brightly single pore (Cryptocoryneae, some Schismatoglottideae). coloured (e.g. some Alocasia, Amorphophallus spp.), SYNANDRIUM (unisexual non-perigoniate flowers): always wet from copious stigmatic secretion during usually ± sessile, sometimes formed by fusion of fila- female anthesis, sometimes producing conspicuous ments only (e.g. Arisaema, Arisarum, Carlephyton sect. nectar droplet (e.g. Anthurium). FRUIT: normally a Pseudocolletogyne, Gorgonidium), more commonly juicy berry, rarely mesocarp leathery; berries nor- composed of completely connate stamens and then mally free, rarely connate (Syngonium) or connate usually apically truncate and ± prismatic in apical cross- and dehiscent as a syncarp (Cryptocoryne), usually section (e.g. Caladieae), sometimes mushroom-shaped red, orange or purplish red, sometimes white (e.g. (Asterostigma) or cylindric (Taccarum), very rarely the some Philodendron, Stenospermation), yellow synandria themselves connate (Ariopsis); common con- (Typhonodorum), green (Arophyton buchetii, nective usually broad, fleshy and probably osmophoric Lysichiton, Orontium, Peltandra virginica), very rarely (e.g. Caladieae, Alocasia); thecae either lateral, apical blue (Amorphophallus kerrii, Gymnostachys), or or marginal depending on the degree of elongation of brownish (Jasarum); infructescence densely packed, the thecae and the extent to which they are overtopped cylindric to globose, exposed by withering, basal by the common connective, dehiscing by single longi- abscission (e.g. Philodendron) or splitting (e.g. tudinal slit or apical stomial pore, with all intermediate Alocasia, Dieffenbachia) of spathe, rarely berries degrees occurring. POLLEN: shed in monads, rarely dehiscent, either basally (Lagenandra) or apically with shed in tetrads (Xanthosoma, Chlorospatha), aperturate the seeds exposed by ± simultaneous sloughing of in most bisexual-flowered genera, inaperturate in most stylar regions of all berries (e.g. Monstereae). SEED: unisexual-flowered genera, exine various (see chapter 1–many per berry; testa thick to thin, smooth, rough- 10). STERILE ORGANS (pistillodes, staminodes, ened, verrucose or striate-costate, papery in seeds synandrodes): often forming zones between fertile with highly developed embryos (Gonatopus), some- zones, sometimes present below female zone times decaying at maturity (Orontium), or lacking (Schismatoglottideae), or on base of terminal appendix, altogether (Gymnostachys, Nephthytis), sometimes aril- very variable in shape, most often ± truncate and pris- late with a conspicuous strophiole (e.g. most Areae, matic (e.g. Philodendron), more rarely filiform, Ambrosina), rarely operculate (e.g. Pistia); embryo subulate, bristle-like or elongate-clavate (Areae), usually straight, sometimes curved (e.g. Cyrtosperma, spathulate (Bucephalandra), cylindric (Aridarum) or Epipremnum), usually undifferentiated, rarely with enlarged and pearl-like (Amorphophallus margaritifer). highly developed plumule (e.g. Cryptocoryne ciliata, TERMINAL APPENDIX: present only in some genera Orontium, Typhonodorum) and then endosperm lack- (e.g. Thomsonieae, Areae), probably always ing and outer cell layers of embryo chlorophyllous; osmophoric, partly or completely (e.g. endosperm copious or absent, with all intermediate Pseudodracontium) covered with staminodes, rugose states occurring. or corrugated or entirely smooth (e.g. most Areae), in- termediate conditions also occur (e.g. Ulearum). 105 genera, over 3300 spp.; distribution subcosmopoli- FEMALE FLOWER (unisexual, non-perigoniate flow- tan, most abundant and diverse in tropical latitudes. ers): Gynoecium sometimes surrounded by a whorl of variously shaped staminodes (e.g. Dieffenbachia, C Spathicarpeae), or sometimes ± regularly associated 82 T H E G E N E R A O F A R A C E A E

C 25 K E Y TO T H E G E N E R A O F A R A C E A E A N D A C O R A C E A E 1. Plants free-floating aquatics; leaves rosulate, hairy; flowers unisexual, naked; inflorescence with a single female flower and a few male flowers · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 105. Pistia 1. Plants terrestrial or helophytes, climbing hemiepiphytes, epiphytes or lithophytes or other, but never floating 2. Leaves not differentiated into petiole and blade, primary venation strictly parallel; inflorescence borne on a culm-like axis 3. Leaves ensiform, unifacial; spadix solitary, pseudolateral and overtopped by a single, erect, leaf-like spathe; flowers 3-merous, tepals 6 · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · Acorus (Acoraceae) 3. Leaves dorsiventrally flattened, bifacial; flowering shoot with long culm-like axis, bearing numerous spadices distally, these borne in axillary clusters subtended by elongate bracts; flowers 2-merous, tepals 4 · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 1. Gymnostachys 2. Leaves with distinct petiole and expanded blade, primary venation never strictly parallel 4. Flowers with obvious perigone of free or fused tepals (except Pycnospatha which lacks perigone, but has dracontioid leaf, tuberous stem and boat-shaped, fornicate spathe – see lead 22) 5. Flowers bisexual, spadix uniform in appearance with flowers of only one type 6. Higher order leaf venation parallel-pinnate; tissues with abundant trichosclereids 7. Spathe persistent; tepals free or connate; ovary 2–4-locular; ovules 2–8 per locule, placenta axile · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 9. Spathiphyllum 7. Spathe deliquescent; tepals connate; ovary 1-locular; ovules several, placenta basal · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 10. Holochlamys 6. Higher order leaf venation clearly reticulated; tissues without trichosclereids or trichoscle- reids very few 8. Stem aerial, not tuberous or rhizomatous, never aculeate; plant usually a climbing hemiepi- phyte or epiphyte, less often lithophyte or terrestrial, only very rarely helophytic (some spp. of Anthurium) 9. Neotropical plants; seeds with copious endosperm; pollen usually forate, never monosul- cate · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 8. Anthurium 9. Palaeotropical plants; seeds without endosperm; pollen monosulcate or inaperturate 10. Stigma transversely oblong; stamens always 4 per flower; pollen inaperturate; perigone consisting of a single cup-like structure · · · · · · · · · · · · · · · · · · · 11. Anadendrum 10. Stigma hemispheric to discoid; stamens usually 6 per flower; pollen monosulcate; perigone usually consisting of free tepals or when connate and cup-like the flowers are borne on short pedicels 11. Ovary 3-locular; locules 1-ovulate; flowering shoot with inflorescences always axillary 12. Flowers sessile; tepals free, very rarely basally united · · · · · · · · · 5. Pothos 12. Flowers pedicellate; tepals connate · · · · · · · · · · · · · · · · · 6. Pedicellarum 11. Ovary 1-locular; flowering shoots terminating in a branching system of spadices · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 7. Pothoidium 8. Stem typically subterranean, tuberous or rhizomatous, sometimes aerial and creeping or scrambling but then aculeate; plant frequently a helophyte 13. Plants of temperate regions (N. America, NE. Asia); leaf blade always entire, ovate to elliptic 14. Ovary 2-locular; ovules 2 per locule, placenta axile · · · · · · · · · · · · · 3. Lysichiton 14. Ovary 1-locular; ovule 1, placenta apical or basal 15. Placenta basal; spathe inconspicuous; spadix cylindric, stipe very long · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 2. Orontium 15. Placenta apical; spathe thick, ventricose, enclosing spadix; spadix subglobose, stipe short · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 4. Symplocarpus 13. Plants of tropical and subtropical regions; leaf blade sagittate, pinnatifid, pinnatisect or dracontioid K E Y T O G E N E R A 83

16. Leaf deeply sagittate, anterior division not pinnatifid or pinnatisect 17. Ovary many- to 2-ovulate, rarely 1-ovulate; seeds with endosperm 18. Plants without stolons; petiole spines dispersed; stamen filaments free; tropical Asia to Oceania · · · · · · · · · · · · · · · · · · · · · · · · 26. Cyrtosperma 18. Plants with long stolons; petiole spines in ridges; stamen filaments free or connate; tropical West Africa · · · · · · · · · · · · · · · · · · · · · 27. Lasimorpha 17. Ovary 1-ovulate, rarely 2-ovulate; seeds without endosperm or rarely with a little endosperm 19. Petiole aculeate, with obvious spines; Malay Archipelago · · 28. Podolasia 19. Petiole smooth to scabrid-verrucose, never aculeate; tropical America 20. Leaf blade never fenestrate; spathe lanceolate, very long-acuminate and usually spirally twisted; ovary locules with (1–)2 to several ovules; neotropics · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 30. Urospatha 20. Leaf blade often perforated with a few perforations of irregular size between primary lateral veins; spathe fornicate; endemic to Brazil (coastal Bahia and Espirito Santo) · · · · · · · · · · · · · · · · · · · 22. Dracontioides 16. Leaf blade pinnatifid, pinnatisect, dracontioid or sometimes ± pedatifid; anterior division always pinnately divided, either pinnatifid, pinnatisect or yet more highly divided 21. Stem aculeate, aerial and scrambling to prostrate, internodes distinct, green · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 29. Lasia 21. Stem not aculeate, subterranean, internodes very abbreviated, not green 22. Leaf blade dracontioid, anterior division bipinnatifid or yet more highly divided; stem a depressed-globose tuber; spathe fornicate 23. Tropical America; flowers with perigone of 4–8 free tepals; berries smooth, red · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 21. Dracontium 23. Tropical southeast Asia; flowers without perigone; berries aculeate, dark green · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 24. Pycnospatha 22. Leaf blade pinnatifid, pinnatisect, or sometimes ± pedatifid, anterior division pinnatifid to pinnatisect; stem a vertical or horizontal rhizome; spathe erect, not fornicate, blade often spirally twisted apically 24. Tropical America; testa thick, verrucose; embryo curved · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 23. Anaphyllopsis 24. Southern India; testa membranous, smooth; embryo straight · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 25. Anaphyllum 5. Flowers unisexual, spadix clearly divided into basal female zone and apical male zone; tropical Africa 25. Leaf pinnatisect to tri- or quadripinnatifid; tepals free; spathe margins free 26. Leaf blade pinnatisect; stamens free · · · · · · · · · · · · · · · · · · · · · · · · · 32. Zamioculcas 26. Leaf blade bipinnatifid to quadripinnatifid, at least in lowest pinnae; stamen filaments connate · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 33. Gonatopus 25. Leaf entire, linear to cordate, sagittate or hastate; tepals connate into cup; spathe margins connate basally · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 34. Stylochaeton 4. Flowers without perigon of free or fused tepals 27. Flowers bisexual; spadix uniform in appearance with flowers of only one type (sometimes with sterile flowers at spadix base) 28. Helophytes from temperate regions of northern hemisphere; petiole sheath with long apical ligule · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 31. Calla 28. Climbing hemiepiphytes or sometimes epiphytes or very rarely rheophytes (few Rhaphidophora) from tropical regions; petiole sheath non-ligulate or ligule only short 29. Petiole usually very short with non-annular insertion; trichosclereids not present in tissues, leaf never perforated or lobed; primary lateral veins forming distinct submarginal vein · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 8. Heteropsis 29. Petiole well-developed with annular insertion and usually conspicuous sheath; tri- chosclereids present in tissues, or if absent (or nearly so) then leaf with conspicuously reticulate higher order venation and often perforated or lobed (Amydrium); primary lat- eral veins usually not forming distinct submarginal vein 84 T H E G E N E R A O F A R A C E A E

30. Trichosclereids rare or nearly absent; higher order leaf venation completely reticulate; ovary 1-locular, placenta 1, intrusive-parietal, ovules 2 · · · · · · · · · 13. Amydrium 30. Trichosclereids abundant; higher order leaf venation parallel to primary lateral veins, or only finest venation reticulate 31. Ovary 1-locular or incompletely 2-locular 32. Ovules anatropous, more than one 33. Ovules numerous, superposed on 2 (rarely 3) parietal placentas; seeds fusiform, straight, 1.3–3.2 mm long, 0.6–1.0 mm wide · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 14. Rhaphidophora 33. Ovules 2–4 (–6) at base of a single intrusive placenta; seeds curved, 3–7 mm long, 1.5–4.0 mm wide · · · · · · · · · · · · · · · · · · 15. Epipremnum 32. Ovules amphitropous to anatropous, solitary, basal 34. Adult leaf blade entire; palaeotropics · · · · · · · · · · · · · 16. Scindapsus 34. Adult leaf blade pinnatifid; neotropics (Amazonia) 18. Alloschemone 31. Ovary 2–5 locular 35. Seeds fusiform, claviform or lenticular, less than 3 mm long, endosperm pre- sent; ovules (2–)3-many per locule; leaf blade entire 36. Placenta basal; seeds fusiform to claviform; leaf blades thickly coria- ceous · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 20. Stenospermation 36. Placenta axile; seeds lenticular and flattened, strongly curved; leaf blades mostly membranous · · · · · · · · · · · · · · · · · · · · · · · 19. Rhodospatha 35. Seeds globose to oblong, 6–22 mm long, the raphe S-shaped; endosperm absent; ovules 2 per locule; leaf blade variously shaped, often perforated or pinnatifid or both · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 17. Monstera 27. Flowers unisexual; spadix clearly divided into basal female zone and apical or intermediate male zone, flowers very rarely in longitudinal rows (Spathicarpa) 37. Spadix fused laterally on both sides to spathe and entirely enclosed by it, forming a septum dividing the spathe into two chambers, with a single gynoecium on one side and the male flowers arranged in 2 rows on the other; very small, seasonally dormant plants endemic to western Mediterranean · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 87. Ambrosina 37. Spadix free or fused to spathe in various degrees but never fused laterally on both sides to spathe to form two internal chambers with a single gynoecium on one side and the male flow- ers on the other 38. Stamens of each male flower free or only the filaments connate 39. Spadix never entirely enclosed by spathe in a basal “ kettle “ formed of connate spathe margins (if spathe margins basally connate then plant never aquatic) 40. Higher order leaf venation parallel-pinnate 41. Upper part of spathe persisting as long as lower part; petiole sheath lacking ligule; ovary 1–many locular; thecae dehiscing by subapical pores or longi- tudinal slits; connective usually conspicuously thickened 42. Spathe variously shaped, never campanulate; plants tropical American or tropical Asian; peduncle usually short, if long then female flowers in single whorl (Aglaodorum) 43. Plant always terrestrial, rarely aquatic, never climbing or epiphytic; inflorescences not secreting resin at anthesis; endothecium with cell wall thickenings; ovary 1 locular or incompletely 2–5 locular; most tropical Asian (except Homalomena sect. Curmeria) 44. Seed without endosperm, embryo large; ovule 1, placenta basal or parietal 45. Inflorescence with short peduncle; female flowers in spirals; stem erect to repent; placenta basal; forest plants · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 72. Aglaonema 45. Inflorescence with long peduncle; female flowers in a single whorl; stem repent; placenta parietal; on tidal mudflats · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 73. Aglaodorum 44. Seed with copious endosperm, embryo relatively small; ovules several to many, placenta basal, parietal or axile K E Y T O G E N E R A 85

46. Male flower consisting of solitary stamen overtopped by flask-shaped pistillode; ovary 1-locular, placenta basal · · · · · · · · · · · · · · · · · · · · · · · · · · · · 46. Furtadoa 46. Male flower consisting of 2–6 stamens, pistillodes absent; ovary incompletely 2–5 locular, placentas parietal and axile · · · · · · · · · · · · · · · · · · · · · 47. Homalomena 43. Plant usually climbing or epiphytic; inflorescences secreting resin from spathe or spadix at anthesis; endothecium nearly always lacking cell wall thickenings; ovary completely 2–many locular, placenta axile to basal; tropical America · · · · · · · · 45. Philodendron 42. Spathe obconic to campanulate; plants from Southern Africa (naturalized in America and Asia); peduncle long, sometimes longer than leaves · · · · · · · · · · · · · · · · · · · · 77. Zantedeschia 41. Upper part of spathe marcescent or caducous at anthesis, lower part long-persistent; petiole sheath with long, marcescent ligule (except most Schismatoglottis spp.); ovary 1-locular; thecae dehiscing by apical pores, connective not conspicuously thickened 47. Placentas parietal; thecae truncate. 48. Spathe constricted; ovules anatropous to hemianatropous; petiole sheath usually not ligulate; upper part of spadix usually sterile. · · · · · · · · · · · · · · · 49. Schismatoglottis 48. Spathe not constricted; ovules hemiorthotropous to orthotropous; petiole sheath with long, marcescent ligule; spadix fertile almost to apex · · · · · · · · · · · · · · · · 50. Piptospatha 47. Placenta basal or basal and apical; thecae truncate or horned 49. Thecae truncate; placenta basal · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 51. Hottarum 49. Thecae horned; placenta basal or basal and apical 50. Stigma smaller than ovary; upper part of spadix sterile with a distinct appendix of hornless sterile flowers; spathe constricted or not; stamens never excavated apically. 51. Spathe not constricted; male flowers smooth or verrucose; sterile flowers between male and female flowers flattened · · · · · · · · · · · · · · · · · · 52. Bucephalandra 51. Spathe constricted; male flowers densely tuberculate; sterile flowers between male and female flowers subcylindric · · · · · · · · · · · · · · · · · 53. Phymatarum 50. Stigma as broad as ovary; upper part of spadix mostly fertile to apex and without a distinct appendix; spathe not conspicuously constricted; stamens all or mostly exca- vated apically 52. Stamens all excavated; placenta basal · · · · · · · · · · · · · · · · · · · · 54. Aridarum 52. Two lateral stamens of each male flower excavated and thecae horned, central stamen truncate and thecae hornless; placentas basal (fertile ovules) and apical (apparently sterile) · · · · · · · · · · · · · · · · · · · · · · · · · · · · 55. Heteroaridarum 40. Higher order leaf venation reticulate 53. Leaf blade dracontioid, leaf solitary in each growth period 54. Petiole usually aculeate; at least some of the ultimate leaf lobes trapezoid, truncate or shallowly bifid, veins not forming regular submarginal collective vein on each side 55. Peduncle long; ovary 1-locular · · · · · · · · · · · · · · · · · · · · · · · · · · · · 70. Anchomanes 55. Peduncle very short; ovary 2-locular · · · · · · · · · · · · · · · · · · · · 71. Pseudohydrosme 54. Petiole usually smooth, sometimes rugose but never aculeate; ultimate leaf lobes usually oblong-elliptic, acuminate, with primary lateral veins forming regular submarginal collective veins on each side 56. Ovary 1–4-locular; terminal appendix smooth, rugose, rarely verrucose or staminodial (appendix absent in Amorphophallus margaritifer and A. coudercii) · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 79. Amorphophallus 56. Ovary always 1-locular; terminal appendix staminodial, separated from male zone by naked axial region · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 80. Pseudodracontium 53. Leaf blade shape of various types but never dracontioid; usually several leaves present 57. Spadix fertile to apex, terminal appendix absent 58. Helophytes with robust, erect stems and an apical crown of sagittate to hastate (rarely tri- sect) leaves; tropical America · · · · · · · · · · · · · · · · · · · · · · · · · · · · 76. Montrichardia 58. Terrestrial, hemiepiphytic or epiphytic plants, leaf blade variously shaped; tropical Africa. 59. Leaf blade usually with pellucid resin canals (lines or points); plants mostly climbing hemiepiphytes; spathe boat-shaped, convolute basally; anthers lacking endothecial thickenings 60. Laticifers absent; flagelliform shoots absent; leaf blade always simple, acute to rounded at base; ovary 1–3-locular; spadix stipitate or sessile · · · 74. Culcasia 86 T H E G E N E R A O F A R A C E A E

60. Laticifers present; flagelliform shoots present; leaf blade oblong-lanceolate to cordate, sagittate, hastate, trifid or laciniate to pinnatifid; ovary 1-locular; spadix sessile · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 75. Cercestis 59. Leaf blade lacking pellucid resin glands; plants terrestrial; spathe ± fully expanded, not convolute; anthers with endothecial thickenings 61. Leaf cordate-sagittate or subtriangular, deeply sagittate or trifid, glabrous; spathe green; spadix entirely free of spathe · · · · · · · · · · · · · · · · · · · 69. Nephthytis 61. Leaf cordate-ovate, minutely hispid abaxially; spathe pure white; female zone of spadix adnate to spathe · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 78. Callopsis 57. Spadix with ± smooth terminal appendix 62. Laticifers anastomosing; tropical South America 63. Ovary 6- to 9-ovulate; leaf blade trisect to pedatisect; stem a subglobose tuber · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 58. Zomicarpa 63. Ovary 1- to 6-ovulate; leaf blade cordate-sagittate; stem a subglobose tuber or rhizome 64. Appendix slender 65. Stamen connective much prolonged, thread-like; stem tuberous; ovary 1- ovulate · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 61. Filarum 65. Stamen connective not at all prolonged; ovary 1-6-ovulate; stem a creeping rhizome · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 59. Zomicarpella 64. Appendix relatively thick and subcylindric; stem a creeping rhizome; ovary 1- ovulate · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 60. Ulearum 62. Laticifers simple; temperate Eurasia and palaeotropics 66. Spadix with zone of sterile flowers between male and female zones, rarely with a naked axis between female and male zones of spadix (Arum pictum) or with fertile zones contiguous (Dracunculus) 67. Placenta parietal to subbasal; leaf blade sagittate or hastate · · · · · · · 88. Arum 67. Placenta apical and/or basal; leaf blade variously shaped 68. Placentas basal and apical 69. Male zone of spadix contiguous with female zone; leaf blade pedatifid but lobes not spirally twisted upwards · · · · · · · · · · · · · · 90. Dracunculus 69. Male zone of spadix separated from female zone by subulate to filiform sterile organs; leaf blade variously shaped 70. Appendix covered with subulate to setiform sterile flowers; leaf blade pedatifid, lobes twisting upwards on each side · · 91. Helicodiceros 70. Appendix smooth; leaf blade oblong-lanceolate or sagittate-hastate · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 92. Theriophonum 68. Placenta basal 71. Lower spathe margins free (except Typhonium hirsutum) 72. Infructescence borne above ground level, berries dark red to purple, pericarp juicy; sterile zone between male and female zones of spadix relatively long, often partially naked; tropical and subtropical to warm temperate Asia to Australia · · · · · · · · · · · · · · · · · 93. Typhonium 72. Infructescence borne at or below ground level, berries white to pale lilac, pericarp firm, not juicy; sterile zone between male and female zones relatively short and covered entirely with subulate sterile flow- ers; Turkey, eastern North Africa, Near East, central Asia · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 89. Eminium 71. Lower spathe margins connate for an appreciable distance (entirely free in Biarum aleppicum) 73. Leaf usually solitary, blade deeply pedatifid to pedatisect; ovary 2–several-ovulate · · · · · · · · · · · · · · · · · · · · · · · 94. Sauromatum 73. Leaves several; blade linear to ovate, elliptic or obovate; ovary 1- ovulate 74. Leaf blade broadly elliptic, spathe tube septate · · 95. Lazarum 74. Leaf blade linear, ovate, elliptic-oblong or obovate; spathe tube not septate · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 96. Biarum 66. Spadix usually without sterile flowers (sometimes present in Arisaema) K E Y T O G E N E R A 87

75. Ovary several-ovulate; female zone of spadix free from spathe; spathe without a transverse septum separating male and female zones. 76. Flowers of both sexes always present in a single inflorescence; male flowers 1-androus; lower spathe margins connate; leaf blade ovate or sagittate · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 86. Arisarum 76. Flowers of both sexes sometimes present in a single inflorescence, but more often with male and female flowers appearing in separate inflorescences; male flowers 2–5-androus; lower spathe margins convolute; leaf blade normally tri- sect, pedatisect or radiatisect, rarely simple and ovate · · · · · 98. Arisaema 75. Ovary 1-ovulate; female zone of spadix adnate to spathe; spathe usually with transverse septum between male and female zones · · · · · · · · · · · 97. Pinellia 39. Spadix entirely enclosed by spathe in a basal “kettle” formed of connate spathe margins, plants always helophytic or aquatic 77. Female flowers spirally arranged (pseudo-whorl in Lagenandra nairii, whorled in L. gomezii) and free; spathe tube “kettle” with connate margins occupying entire spathe tube; spathe blade usually opening only slightly by a straight ot twisted slit; berries free, opening from base; leaf ptyxis involute · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 56. Lagenandra 77. Female flowers in a single whorl, connate; spathe tube kettle occupying only lower part of spathe tube, remainder also with connate margins (except Cryptocoryne spiralis), blade spreading or twisted; berries connate into a syncarp which opens from the apex; leaf ptyxis convolute · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 57. Cryptocoryne 38. Stamens of each male flower entirely connate into a distinct synandrium, synandrium rarely reduced to single stamen (Colletogyne endemic to Madagascar) or stamens free and basally connate with remote globose thecae (Gorgonidium endemic to Andean South America), or only filaments con- nate and then stigma stellate and 5–8-lobed (Spathantheum) 78. Laticifers simple 79. Synandria connate, thecae of adjacent synandria encircling pits in the spadix, each pit with a somewhat prominent upper margin; leaf peltate; Burma to India · · · · · · · · 99. Ariopsis 79. Synandria free; leaf not peltate; Africa, Madagascar or Americas 80. Higher order leaf venation parallel-pinnate or if reticulate then stem a creeping rhizome and plant from Amazonia (Bognera) 81. Ovules anatropous; primary lateral veins of leaf forming a single marginal vein, no submarginal collective vein present; plant from tropical America or continental tropical Africa. 82. Female zone of spadix free; plant from tropical west and central Africa · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 48. Anubias 82. Female zone of spadix entirely adnate to spathe; plant from tropical America. 83. Female flowers each with whorl of several staminodes; higher order leaf venation strictly parallel-pinnate · · · · · · · · · · · · · · · · · 35. Dieffenbachia 83. Female flowers without staminodes; higher order leaf venation reticulate · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 36. Bognera 81. Ovules orthotropous to hemi-orthotropous; primary lateral veins of leaf forming sub- marginal collective vein and 1–2 marginal veins; plants from temperate eastern North America or Madagascar. 84. Giant herbs (to 4m) with massive pseudostem of petiole sheaths; staminodes of female flower free; Madagascar (also naturalized in Pemba, Zanzibar and Mascarene Is.) · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 85. Typhonodorum 84. Relatively small herbs (less than 1m) without pseudostem; staminodes of female flower connate into a cup-like synandrodium; eastern North America · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 84. Peltandra 80. Higher order leaf venation reticulate; stem usually a subglobose tuber, if rhizomatous then plant Madagascan 85. Madagascan plants; seed lacking endosperm; ovary 1-locular; leaf venation with pri- mary lateral veins forming submarginal collective vein and 1-2 marginal veins on each side of blade 86. Stamens either completely connate with marginal thecae or only partially connate by filaments; bisexual flowers often present between male and female zones of the spadix · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 82. Carlephyton 88 T H E G E N E R A O F A R A C E A E

86. Stamens completely connate into truncate synandria or synandria reduced to one stamen 87. Synandria reduced to one stamen, thecae apical on conical filament; spadix fertile to apex; leaf blade always cordate · · · · · · · · · · · · 83. Colletogyne 87. Thecae apical on a truncate synandrium; spadix appendix present or not; leaf blade cordate, hastate, trifid, trisect or pedatifid · · · · · · · · 81. Arophyton 85. South American plants; seed with abundant endosperm; ovary with more than 1 locule (except Spathicarpa); leaf venation with primary lateral veins usually forming single marginal vein on each side, submarginal collective veins usually absent 88. Spadix free or only female zone adnate to spathe 89. Ovules anatropous 90. Ovules 2 per locule; leaf blade entire, linear to subsagittate; spadix, with terminal appendix of synandrodes · · · · · · · · · · · · · · · · 37. Mangonia 90. Ovules 1 per locule; leaf blade entire, pinnatifid to subdracontioid; spadix fertile to apex 91. Leaf blade pinnatifid to bipinnatifid or subdracontioid; synandria elongate; stigma capitate or lobed; staminodes of female flowers free (connate in Taccarum caudatum) · · · · · · · · · · · · · 38. Taccarum 91. Leaf blade usually pinnatifid, rarely entire; synandria short and domed; stigma deeply lobed; staminodes of female flowers free or connate · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 39. Asterostigma 89. Ovules orthotropous 92. Styles and synandria elongate; leaf blade pinnatifid or -sect or bipinnatifid or entire and ± cordate 93. Staminodes in female flowers filiform to subclavate; synandria with free filament apices or not; leaf blade pinnatifid, pinnatisect or bipinnatifid · · · · · · · · · · · · · · · · · · · · · · · · · · · 40. Gorgonidium 93. Staminodes in female flowers elongate-triangular; synandria entirely connate; leaf blade entire, ± cordate · · · · · · 41. Synandrospadix 92. Styles and synandria short and squat or synandria very flat; staminodes in female flowers obovate or trapezoid; leaf blade pedatisect or sub- palmatifid · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 42. Gearum 88. Spadix usually entirely adnate to spathe (male zone free in Spathantheum inter- medium) 94. Ovary 6–8-locular; female flowers below, male above; leaf blade entire or pin- nately lobed · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 43. Spathantheum 94. Ovary 1-locular; female and male flowers intermixed (2 central rows of male flowers, 2 outer rows of female flowers); leaf blade entire · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 44. Spathicarpa 78. Laticifers anastomosing 95. Plants climbing hemiepiphytes, sometimes creeping on ground in submature growth; inter- nodes long; berries connate into a syncarp · · · · · · · · · · · · · · · · · · · · · · · 67. Syngonium 95. Plants terrestrial or geophytic, rarely aquatic, not climbing; internodes very short; berries free from each other 96. Spadix without an appendix (present in Hapaline appendiculata, included here, occa- sionally absent in Colocasia esculenta, excluded here) 97. Ovary completely to incompletely 2- to several-locular with deeply intrusive parietal placentas (1-locular with basal placenta in Jasarum, Scaphispatha and a few species of Caladium and Xanthosoma); neotropical plants 98. Helophytes or terrestrial; leaf blade ovate, sagittate to hastate or pedatifid 99. Pollen shed in tetrads; style usually laterally thickened or expanded into a diaphanous mantle; leaf blade rarely peltate, sometimes trifid or -sect, or pedatifid or -sect 100. Spathe tube subglobose, inflated; female zone of spadix free; styles normally discoid (laterally swollen) and coherent (except Xanthosoma plowmanii); synandrodes (sterile flowers) between male and female flowers well-developed, ± prismatic · · · · · · · · · · · 65. Xanthosoma K E Y T O G E N E R A 89