BOOK-ARACEAE

THE GENERA OF ARACEAE S J Mayo J Bogner P C Boyce

THE GENERA OF ARACEAE S J Mayo J Bogner P C Boyce WITH CONTRIBUTIONS FROM J.C. French and R. Hegnauer ILLUSTRATIONS BY E. Catherine

© Copyright The Trustees, Royal Botanic Gardens, Kew First published 1997 ISBN 1 900347 22 9 Cover design by John Stone Book design by Jeff Eden Page make-up by Media Resources, Information Services Department, Royal Botanic Gardens, Kew Printed in The European Union by Continental Printing, Belgium.

CONTENTS - TO RETURN TO THIS PAGE PRESS C Dedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .vii Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi A GENERAL PART 1. History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 2. Vegetative Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 3. Vegetative Anatomy (by J.C. French) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 4. Inflorescence and Floral Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 5. Inflorescence and Floral Anatomy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 6. Fruits and Seeds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 7. Seedling Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 8. Embryology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 9. Cytology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 10. Palynology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 11. Phytochemistry and Chemotaxonomy (by R. Hegnauer) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 12. Ecology and Life Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 13. Pollination Biology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 14. Dispersal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 15. Geography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 16. Uses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 17. Cultivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 18. Conservation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 19. Fossil Record . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 20. Phylogenetic relationships within the Monocotyledons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 21. Phylogenetic relationships within Araceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 22. Previous classifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 CONTENTS v

CONTENTS B TAXONOMIC PART 23. Synopsis of the Classification of Araceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 24. Family Description of Araceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 25. Key to the Genera of Araceae and Acoraceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 26. Descriptions of the Tribes and Genera of Araceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 27. Description of Acoraceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289 28. References and selected taxonomic literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291 29. Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310 30. Appendix: Table 9. Fungal Parasites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318 Table 10. Schott’s (1860) classification of Aroideae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320 Table 11. Engler’s (1876b) classification of Araceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321 Table 12. Engler’s (1920b) classification of Araceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322 Table 13. Grayum’s (1990) classification of Araceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323 Table 14. Bogner & Nicolson’s (1991) classification of Araceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324 Table 15. Generic country lists . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325 Table 16. Colour plates: photo credits and vouchers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329 31. Index to scientific names . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331 32. Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342 33. Descriptions of new taxa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346 Colour Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347 vi T H E G E N E R A O F A R A C E A E

C DEDICATION We dedicate this book to the memory of Heinrich Wilhelm Schott and Heinrich Adolf Gustav Engler, the two great founding fathers of Araceae systematics, and also to Nicholas Edward Brown, whose studies of the family, while less widely known, were of the highest standard. D E D I C A T I O N vii

C FOREWORD By Dan H. Nicolson It has been almost 150 years since H.W. Schott pub- lished his monumental Genera Aroidearum (The Genera of Araceae). A new treatment of the family has been a desideratum for a century and urgently needed during the last fifty years. What has been particularly needed is an illustrated work that can be used by any- one to recognize unknowns and help learn the terms necessary for accurate understanding of the taxa. It is often taken for granted that the family is more or less completely known and information is readily available. The first problem is that the keystone works by Schott (1794-1865) and Engler (1844-1930) are expensive, if available for purchase, and not to be found in public libraries. If you are fortunate enough to find them, you discover the second problem. They are in Latin! In spite of such problems there has been a revival of interest in the family. An astonishing number of workers have been studying plants in the field and cul- tivation and applying new and sophisticated techniques. New tools, such as the scanning electron microscope (which revolutionized comparative study of pollen) and techniques, such as molecular systematics and cladistics, have revolutionized thinking about relationships. At last it has been possible to stimulate three peo- ple, scarcely more than entering middle life and with all the necessary modern training, to collaborate and synthesize what is known about the genera of Araceae, including at least one crisp, fresh and new full page drawing of each genus. Comparison of this work with its predecessor shows the distance we have travelled. Do not think that the last word has been spoken. New species, even genera, are still turning up and some of them are testing the hypotheses that we wish were facts. Smithsonian Institution, Washington, D.C. March 1994 viii T H E G E N E R A O F A R A C E A E

C PREFACE The Araceae, or aroids, are plants which are very famil- most beautiful and dramatic that the vegetable king- iar to everyone but paradoxically little known. dom has to offer. Monstera deliciosa, Epipremnum pinnatum ‘Aureum’ (syn. Epipremnum aureum), Philodendron scandens, The idea for this book germinated in 1980 when the Dieffenbachia maculata and Aglaonema commuta- first international workshop on the systematics of tum may be counted as among the world’s most Araceae was held at the Marie Selby Botanical Garden, popular house plants. But most rarely flower in their Sarasota, Florida, organized by Dr Michael Madison. domestic environment, and the fact that they are all Madison and Simon Mayo began a manuscript but the aroids is appreciated by relatively few people. Once project never came to fruition. In 1987, with the encour- their inflorescences appear, however, it is obvious they agement of Professor Grenville Lucas, then Keeper of the belong together, with their characteristic cowl-like Kew Herbarium, Mayo and Josef Bogner of the Munich spathe and central, fleshy spike, known as the spadix. Botanic Garden resolved to tackle the task anew and The most familiar aroid inflorescences are those of Peter Boyce joined us soon afterwards. Eleanor Anthurium andraeanum and Zantedeschia aethiopica, Catherine, the artist, completed the team at a later stage. which are used the world over as cut flowers. The format of this book is modelled on that of the The family is extraordinarily diverse in appearance, Genera Palmarum by Dr Natalie Uhl and Dr John with the foliage being probably its most widely appre- Dransfield. The delimitation of all the genera has been ciated feature. The perforated leaf of Monstera deliciosa critically re-examined in the light of modern studies. (the subject of our cover) is one of the most instantly Since A. Engler’s last monograph in “Das Pflanzenreich” recognizable plant images the world over, while the various new genera have been described and old ones velvet, pendent leaves of Anthurium waroqueanum reduced to synonymy. F. Gagnepain described are unforgettably elegant. The beauty of many other Pycnospatha from Thailand, H. Jumelle Arophyton and genera, such as Philodendron, Alocasia and Arisaema Carlephyton from Madagascar and S. Buchet is also very largely due to their superb foliage. In sharp Colletogyne, also from Madagascar. M. Hotta con- contrast, the inflorescences of certain aroids are quite tributed four new genera, Heteroaridarum, repulsive; that of Helicodiceros muscivorus, for exam- Pedicellarum and Phymatarum from Sarawak, and ple, resembles nothing so much as the rear end of a Furtadoa from Sumatra and the Malay Peninsula. D.H. decomposing mammal corpse, while so bizarre is the Nicolson and colleagues contributed Bognera and foul-smelling Amorphophallus konjac that some years Filarum from tropical America and Hottarum from ago it was chosen as the basis for a film dramatization Borneo. G.S. Bunting described the extraordinary of the “triffids” of science fiction. Towering over all aquatic Jasarum, and Lasimorpha, a synonym of such lesser monstrosities is Amorphophallus titanum, Cyrtosperma according to Engler, has been reestab- one of the vegetable wonders of the world, which lished by A. Hay. Hay also recently described attracts large crowds at botanic gardens when from Anaphyllopsis from tropical America and Lazarum from time to time the great tuber yields a huge phallic inflo- Australia. Despite these changes, the total number of rescence smelling of rotten fish. genera treated here (105 without Acorus) is much the same as that presented by Engler (108 without Acorus). Why is it then, that aroids are not better known as Cladistic analysis of morphological and molecular data a plant group? We think the reason is a dearth of acces- in recent years has, however, meant that the classifi- sible literature. Apart from Deni Bown’s excellent cation has changed considerably since Engler’s time. “Aroids. Plants of the Arum Family”, published in 1988 for the non-specialist reader, there have been no gen- We have deliberately laid the primary emphasis on eral and comprehensive treatments of the Araceae, the preparation of completely revised descriptions of either technical or lay, since the publication of Engler’s the genera, together with analytical illustrations for taxonomic monograph (in Latin and German) for his each. The plates are all original drawings by Eleanor “Das Pflanzenreich” series in the first two decades of Catherine, based on a combination of herbarium, spirit this century. We have tried to contribute towards fill- (from the Kew spirit collection) and living specimens, ing this gap with a general taxonomic treatment of the supplemented when necessary by photographs. The family which extends to the level of genus. While it is chapters of the General Part are intended to be sum- unashamedly technical, and may seem somewhat for- maries, in some cases quite brief, of various aspects of bidding to the general reader, we hope that the book’s Araceae which are of taxonomic and general interest. publication will encourage more people to study and In two cases, the treatments are much more detailed, enjoy these wonderful plants, which are among the namely chapter 3 on vegetative anatomy, and chapter 11 on phytochemistry and chemotaxonomy, which we P R E F A C E ix

were very fortunate to receive from Professor J.C. the genus in a general treatment of aroids. French and Professor R. Hegnauer, respectively. Their In contrast, we have not included any systematic contributions are the first detailed modern reviews in English of these subjects for the Araceae . account of the Lemnaceae, which according to the molecular work of French and colleagues, seem clearly We have omitted any general treatment of the mol- to be embedded within the Araceae (see chapter 20). ecular systematics of the family, which is being studied, Our reasons are again pragmatic. The taxonomy of the in particular, by Professor French and his colleagues at Lemnaceae has been comprehensively revised in recent Rutgers University, New Jersey. They have generously years by Professor E. Landolt and French’s results allowed us access to their most important phyloge- became available only in the final stages of preparation netic conclusions, thus greatly improving our of this book. discussion of the family’s phylogeny (chapter 21). Throughout the text we have employed the terms We have included material on Acorus, including a “aroid” and “araceous” as synonymous adjectives refer- generic treatment, despite the fact that we accept that ring to any member of the family, and the noun this genus does not belong to the Araceae. However, “aroid” likewise. The reader should therefore not inter- it was felt that it would be convenient to the non-spe- pret “aroid” as referring only to members of the cialist reader, who might expect to find something on subfamily Aroideae. x THE GENERA OF ARACEAE

C ACKNOWLEDGEMENTS A book of this kind cannot be prepared without the Milan Svanderlik, Mr Jeff Eden, Mrs Christine Beard, help of many people besides the authors. Dr Geoff Kite, Mrs Margaret Newman, Mr John Stone, Mr John Woodhams, Mr Michael Marsh, Mr John Hale, Our greatest debt is to Eleanor Catherine, whose Mr John Norris, Mr Phillip Brewster. professionalism, patience and superb facility in line illustration has produced such a unique and wonder- It is also a pleasure to thank Cássia Mônica ful set of plates. It is likely that her work will be the Sakuragui for preparing the original maps. most valuable contribution that this book can make to advancing knowledge of the Araceae and it is no exag- We are very grateful to the directors and staff of the geration to say that all aroid-lovers are in her debt. following herbaria for lending us material:- A, AAU, B, BK, BKF, BM, BO, BRUN, C, DNA, E, FI, G, HRB, K, The growth of activity in aroid systematics in recent KEP, L, LE, M, MO, NY, P, SAN, SAR, SEL, SGN, SI, U, decades has been notable for the friendly spirit of col- US, YUKU, YUNU. laboration that has always prevailed and because of this we have been able to call on the help of many J. Bogner wishes to thank the Directors of the fol- other specialist colleagues. To all these we are lowing institutes for financial support: the Smithsonian extremely grateful. Dr Dan H. Nicolson undertook the Institution (Washington DC, USA) for a field study in task of reviewing our manuscript and contributing a Sarawak, the Centre National de la Recherche foreword. The generic descriptions have been revised Scientifique (CNRS, Paris, France) for the publication by the following specialists who have been very gen- of the Araceae for “Flore de Madagascar et des erous in allowing us to use unpublished manuscripts Comores” and the Royal Botanic Gardens, Kew (UK) and in providing new data and insights on taxa and for visits to RBG Kew in connection with the prepa- character fields concerning which their knowledge is ration of this book. much more profound than our own:- Mr Julius Boos, Dr Thomas B. Croat, Professor James C. French, Dr The living plants used in the preparation of this Michael H. Grayum, Dr A. Hay, Professor R. Hegnauer, book were cultivated at the Botanischer Garten Munich Mr Wilbert Hetterscheid, Professor Niels Jacobsen, Dr and the Living Collections Department of the Royal Z. Kvaček, Dr Gitte Petersen, Professor Robin Scribailo, Botanic Gardens, Kew and we are very grateful to the Dr Mikhail Serebryanyi, Dr Elke Seubert, Dr M. Curators and staff of these institutes for their helpful Sivadasan, Professor Thomas Stützel, Dr Sue collaboration. Thompson, Professor Hans-Jürgen Tillich and Dr Guanghua Zhu. Professor Tom Ray contributed new Other colleagues should be mentioned who have data on shoot organization. These contributions have prepared the ground in some important way; Deni greatly enriched this book, but its shortcomings are Bown for her splendid pioneering book “Aroids: Plants necessarily our responsibility alone. of the Arum family”, Dr George Bunting, who was responsible for starting the modern renaissance in aroid We would also like to thank the following staff systematics, Dr Michael Madison, Mrs Betty Waterbury members of the Royal Botanic Gardens, Kew, who, and Mrs Libby Besse who with their friends and col- among many others, provided essential support:- leagues, started the International Aroid Society and its Professor Ghillean T. Prance, Dr Charles Stirton, excellent journal Aroideana, and finally Professor P.B. Professor Grenville Lucas, Dr Michael Lock, Dr John Tomlinson who provided the impetus for continuing Dransfield, Dr Phillip Cribb, Miss Mary Gregory, Mr the international workshops on Araceae by organizing the second (Harvard Forest 1984) and leading the third (Berlin 1987) meetings. A C K N O W L E D G E M E N T S xi

AGENERAL

C 1 HISTORY The word “arum” is derived directly some of the finest botanical artists of from ancient Greek “aron” and indi- the day (Riedl & Riedl-Dorn 1988). vidual species of the Araceae have The pencil drawings are of her- been recorded by many botanists barium specimens and material and historians since those ancient preserved in alcohol from times. Theophrastus (ca. herbaria all over Europe and 371–285 B.C.) recorded Arum the water colour plates were in his treatise (Prime 1960), painted from living plants and Hernandez (1790) grown at Schönbrunn, the described a number of tropi- imperial gardens near Vienna. cal aroids and their uses by The colour plates represent Aztec people. European each plant in astonishing Araceae were described in detail and thoroughness and detail by botanists such as are among the most impres- Fuchs (1542) and Ray (1686) sive analyses ever achieved in while R. Dodoens (1574) the medium of botanical illus- arranged all Araceae known to tration. Only relatively few him into a single group. were published (Schott J. P. de Tournefort (1700) 1853–1857, 1857, 1858, Peyritsch created a “class” without a name 1879). Today the Icones are part which grouped three European of the collections of the Vienna genera (Arum, Dracunculus and Natural History Museum. About 80 Arisarum), characterized by the pos- plates (including subfamily Lasioideae) session of a “monopetalous” flower. are lost, while Schott’s herbarium of This concept of the aroid inflores- Araceae, consisting of 1379 speci- cence as a flower with a single petal Figure 1. H.W. Schott (1794–1865). mens, was destroyed at the end of also influenced Linnaeus (1753, Photograph c. 1860. the Second World War (Schott 1984, 1754) who classified the known Riedl & Riedl-Dorn 1988). species according to his artificial Schott’s most important works sexual system. It was not until later that the inflores- were the account of Araceae in the Meletemata botan- cence was recognized as a spike (spadix) of tiny ica (Schott 1832), the Genera Aroidearum (Schott 1858) flowers surrounded by an often colourful bract and the Prodromus systematis Aroidearum (Schott (spathe). A.L. de Jussieu (1789) established the Araceae 1860). Some of his new genera were first published in as a natural family but recognized only a few small or the cultural periodical Wiener Zeitschrift für Kunst, rather broadly conceived genera, probably because of Literatur, Theater und Mode as part of a series of articles the paucity of good material of non-European taxa. entitled Für Liebhaber der Botanik (Schott 1829a–e, All the climbing species were grouped under the name 1830). He also published many papers on Araceae in the Pothos and most of the terrestrial species were placed Oesterreiches Botanisches Wochenblatt, which later in the genera Arum and Dracontium. became the Oesterreichische Botanische Zeitschrift (from Modern systematic studies of the Araceae began 1858, the 8th volume, onwards). with the work of the Austrian botanist and gardener Schott described many new genera and species Heinrich Wilhelm Schott. He was the first monographer and created the first major natural classification of the of the family and the first botanist to make careful whole family. Though his taxon concepts were nar- comparative studies of aroid inflorescences, flowers row, many of his genera and species have withstood and fruits. Using these observations he was able to the test of time. He created the basis of Araceae tax- put the family on a sound taxonomic footing. His work onomy, not only for Engler, who soon followed him is documented in his writings (Riedl 1965a, b) and by in studying the family comprehensively, but also for an outstanding archive of scientific illustrations (Schott succeeding generations. A notable aspect of Schott’s 1984). This consists of a collection of 4400 superb work was that he used a combination of herbarium coloured and black and white plates of Araceae, the material, living plants and field work in the study of a Icones Aroidearum, which were prepared at his direc- largely tropical plant group at a time when such a tion and personal expense and employed the talents of wide-ranging approach was most unusual. 2 THE GENERA OF ARACEAE

Riedl (1965a, b, pers. comm.) Other significant works on the has given details of Schott’s life family published during Schott’s and work. He was born on 7th lifetime include Kunth’s treatment January 1794 in Brünn in for his Enumeratio Plantarum Moravia (now Czech Republic). (Kunth 1841), which was the first At the age of seven he moved to post-Linnean treatment at species Vienna, where his father had level, and Blume’s Rumphia become head gardener at the (Blume 1836–1837), which was botanical garden of the important especially for Asian University. There he soon came genera and included very fine into contact with eminent coloured plates. botanists. N. von Jacquin stimu- The second great monographer lated and directed the early of Araceae was Adolf Engler. interest of the boy to study Right from the start he established plants and once, when severely himself as an authority on the ill as a youth, he received a visit family by a series of prodigious from Alexander von Humboldt works. His first major publication which made a lifelong impres- on the family outlined a new sys- sion upon him. tem on phylogenetic lines (Engler The young Schott attended 1876b) which was substantially lectures in botany and other different from Schott’s classifica- disciplines (agriculture, chem- tion, especially at the higher istry) at the University, but his ranks. The following year he pre- first employment was as an sented a pioneering comparative assistant gardener under the study of araceous shoot organi- direction of his father. In 1815 zation, based on original he became gardener for the col- Figure 2. H.A. Engler (1844–1930). observations (Engler 1877). His lection of Austrian flora at the Photograph taken when Engler was treatment of Araceae for Martius’s Belvedere Palace. Then, at von Flora Brasiliensis followed nineteen years old. Jacquin’s recommendation he (Engler 1878), and immediately was chosen to become a par- afterwards a comprehensive ticipant in the famous Austrian-Bavarian monograph at species level for the entire scientific expedition to Brazil which family appeared in De Candolle’s included the botanist Mikan, the Monographiae Phanerogamarum zoologist Natterer, the mineralogist (Engler 1879). His account for and botanist Pohl, the botanist Beccari’s Malesia (Engler 1883a) C.F.P. von Martius and zoologist was followed by the family treat- J. von Spix. ment for Die natürlichen Between 1817 and 1821 he Pflanzenfamilien (Engler 1887– worked in Brazil as a member 1889). At about this time he of this group of naturalists and also published two papers in made acquaintance with a the Botanische Jahrbücher rich tropical flora (Schreiber which further developed his 1822). His main duty was to phylogenetic interpretations establish a garden in Rio de of morphological and anatom- Janeiro to prepare plants for ical trends in the Araceae the long journey to Europe. (Engler 1883b, 1884). When he returned to Vienna With the initiation of the in 1821 he took up gardening Das Pflanzenreich series in again, eventually rising to 1900, Engler embarked on his become the Director of the second monograph of the entire botanical and zoological gardens family, completed in 1920 (Engler at the imperial palace of Schön- 1905, 1911, 1912, 1915, 1920a, brunn, a position he held until his 1920b, Engler & Krause 1908, 1920, death on 5th March 1865. Schott was Krause 1908, 1913) with the help of his made a Doctor “honoris causa” and assistant Kurt Krause (1883–1963). Engler a member of the Kaiserliche and Krause doubled Schott’s total Akademie der Wissenschaften for Figure 3. H.A. Engler in of about 900 species to around his work in botany and horticulture. his later years 1800. An important feature of this HISTORY 3

treatment was the large number of institute. He was by now an extremely illustrations, prepared by J. Pohl. able organizer and under his direc- These are mostly original but tion a new botanical garden and some were copied from Schott’s botanical museum were planned illustrations. The large living and built at Dahlem, then on collection of Araceae at the the outskirts of Berlin. Started Botanischer Garten Berlin- in 1897, the new garden was Dahlem was an important inaugurated in 1910. In 1902 resource for Engler’s studies. he journeyed to the Cape and He carried out little field Transvaal in South Africa, East work but he was an accom- Africa and Egypt. In 1905 he plished synthesiser of data visited South and East Africa from the notes and collec- again and also Zimbabwe, tions of others. Botanists from Zambia and the Zambezi all over the world sent him region, after which he went material and information. to India and Sri Lanka, where Diels’s (1931) biography he paid particular attention to gives a detailed account of economic plants and Araceae. Engler’s life and work, from His journey then took him to which we have selected the Bogor, Singapore, Malaysia, main facts. Engler was born on Burma, the Himalaya and 25th March 1844 in Sagan, Lower Calcutta. In 1913 he went on a Silesia (today Poland), the son of world tour that took him to a tradesman. In 1848 his mother Southwest Africa, China, Japan, took him to the provincial capital of Hawaii, California and New Breslau, where the young Engler England. He was awarded hon- grew up. From his early years he Figure 4. N.E. Brown (1849–1934). orary doctorates of the Universities took a great interest in natural his- Photograph c. 1900. of Cambridge, Cape Town, tory and during his University years Uppsala and Geneva and the gold studied with the famous palaeob- medal of the Linnean Society. He otanist and teacher H.R. Goeppert, later becoming a retired officially as Director at Berlin on 31st March teacher himself for a short time in Breslau (today 1921 but continued his scientific work until his death Wroclaw). In 1869 he became acquainted with A.W. on 10th October 1930. Eichler who had succeeded C. von Martius as editor of K. Krause continued to work on Araceae at Berlin the “Flora Brasiliensis” and this contact was to be of until 1942, publishing new species collected by var- great significance for his future career. In April 1871 he ious field botanists and collectors, especially those was employed as a scientific assistant at the Botanische working in South America, but he did not take up his Sammlungen (herbarium and living collections) in studies again after the Second World War ended in Munich. Here, under the guidance of Nägeli’s direc- 1945 and died in obscurity in 1963. A manuscript torship, he matured scientifically for eight years until he that he completed in 1942 for the second edition of was 34 years old. The botanic garden and herbarium, Die natürlichen Pflanzenfamilien was lost in 1943 together with the library (Bayerische Staatsbibliothek) when the Berlin Botanical Museum was largely provided excellent working conditions. He worked up destroyed by war action. his professorial qualification (habilitation) in 1872 and Nicholas Edward Brown (1849–1934), working at later taught at the University. Engler’s period at Munich the Herbarium of the Royal Botanic Gardens, Kew, was clearly the most creative of his life as far as the made significant contributions to Araceae systematics. Araceae were concerned, since it was there that he His largest and most important publication on made most of his important innovations and insights Araceae was the family treatment for Flora of Tropical into the family’s taxonomy. Africa (Brown 1901), but in addition to this he pub- In 1879 Engler became a professor ordinarius at lished various other flora treatments, many new the University of Kiel, where he also became Director species and various new genera. Many of his new of the botanical garden and institute. However, after the taxa were based on living material introduced into the death of Goeppert in 1884 he was appointed to suc- gardens at Kew. Brown’s work is particularly notable ceed his former professor at the University of Breslau for his meticulous descriptions based on accurate which included the directorship of the botanic garden. observation and it is evident that his grasp of Araceae There he began work on Die natürlichen Pflan- systematics was profound. Brown was an almost exact zenfamilien together with K. Prantl. After five years, he contemporary of Engler. He began work at the Kew moved to Berlin in October 1889 to become the Herbarium as an assistant in 1873 and retired in 1914 Director of Germany’s largest botanical garden and as an Assistant Keeper. 4 THE GENERA OF ARACEAE

Joseph Dalton Hooker (1883), in Bentham & These events have done a great deal to foster interna- C Hooker’s Genera Plantarum, published a classification tional collaboration. of Araceae based largely on Schott’s (1860) work which was later revised and modified by John Hutchinson Bogner (1979a) updated the Engler classification, (1934, 1959, 1973), another Kew botanist. Hutchinson adding newly described genera and taking account of made an impact on Araceae taxonomy because his new synonymy. Nicolson (1983) published an English treatment included a key to all genera in English. translation of Engler’s classification, including the However, aroid specialists have found his system to be accepted genera described since 1920. Bogner & artificial and it was never widely used. Nicolson (1991) published a revised synoptic key to all the genera, which incorporates a number of impor- T. Nakai (1943) published a new classification in tant changes from Engler’s concepts, particularly in the which he recognized the Pistiaceae, Cryptocorynaceae subfamilies Pothoideae and Lasioideae. M.H. Grayum and Acoraceae as separate families from the Araceae. (1984, 1990) presented a new phylogenetic classifica- He also made important contributions to the system- tion which is especially notable for recommending the atics of the genus Arisaema. removal of Acorus from Araceae and the large scale reorganization of Engler’s subfamilies. After 1945 a period of relative inactivity followed, until the 1950s when George S. Bunting, Monroe Grayum’s work, in particular, has been a stimulus Birdsey, Dan H. Nicolson and Josef Bogner began to other workers to initiate new studies of taxonomi- working on the family. M. Hotta began to publish on cally significant character fields, largely due to his Araceae in the 1960s (Hotta 1965, 1966a–b, 1967) and comprehensive review of the technical literature. later he presented a classification for Eastern Asian Recent comparative character surveys of particular sig- and Malesian genera and an influential study of veg- nificance are the studies of J.C. French (anatomy, also etative and floral morphology (Hotta 1970, 1971). with P.B. Tomlinson), T. Ray and P. Blanc (shoot mor- Since then he has made many important investigations phology), G. Petersen (cytology), M. Grayum of aroids in the Malay Archipelago region (Hotta 1976, (palynology), H.-J. Tillich (seedling morphology), D. 1981, 1982, 1984, 1985, 1986a–b). Other important Barabé, L. Chrétien, S. Forget and M. Labrecque (flo- studies carried out during this period are those by G. ral anatomy), W.N. Carvell (floral anatomy), E. Seubert Thanikaimoni (pollen morphology, 1969) and C. (seed anatomy) and J.C. French, M. Chung & Y. Hur Marchant (1970, 1971a–b, 1972, 1973, cytology). (cpDNA molecules). Floristic and revisionary studies in the 1950s and 1960s that should be mentioned are those of H.C.D. There are many centres of research on Araceae de Wit (Cryptocoryne, Lagenandra), H. Riedl (Flora systematics now active throughout the world, includ- Iranica, Eminium), Bunting (Spathiphyllum) and ing those in Brazil (I.M. Andrade, C. Barros, E. Nicolson (Aglaonema). Gonçalves, A. Lins, M. Nadruz Coelho, M.L. Soares, F. Ramalho, C. Sakuragui, J. Waechter). Calicut (M. In the 1970s and 1980s, the pace of work on the Sivadasan), Copenhagen (N. Jacobsen, G. Petersen), systematics of Araceae increased. In 1978 the Kew (S.J. Mayo, P.C. Boyce, G. Kite), Kunming (Li International Aroid Society was formed in Florida, USA Heng), Kyoto and Kagoshima (M. Hotta), Leiden (W. and the journal Aroideana founded, originally under Hetterscheid), Moscow (M. Serebryanyi), Munich (J. the energetic editorship of M.T. Madison. Aroideana Bogner), Rutgers University, New Jersey (J.C. French, generated a significant expansion of scientific and M. Chung, Y. Hur), St. Louis (T.B. Croat, M.H. Grayum, horticultural interest in the family and continues to G. Zhu), Sydney (A. Hay), Tokyo (J. Murata), play an important role. A series of international work- Washington DC. (D.H. Nicolson), and until recently, at shops on Araceae systematics was established, Yaoundé (C. Ntépé-Nyame†). beginning at Sarasota (1980), and continuing at Harvard Forest (1984), Berlin (1987), Moscow (1992), With this increase in taxonomic activity, new revi- Tokyo (1993), and Kunming (1995), with further meet- sions and floristic studies are now in progress which ings planned for Sydney (1998) and St Louis (1999). should eventually lead to a much more complete knowledge of the family. HISTORY 5

C 2 VEGETATIVE MORPHOLOGY Root Pinellia, and hypogeal stolons in Colocasia, Cryptocoryne, some Spathiphyllum spp. and Roots in Araceae are always adventitious and dimorphic Lasimorpha. roots are often found in climbing hemiepiphytes, e.g. Shoot organization shows a range of interesting vari- Monstera deliciosa, Philodendron bipinnatifidum. In ation within the family and can be taxonomically useful some genera, e.g. Arisarum, Arum, Biarum, (Engler 1877, Blanc 1977a,b, 1978, 1980, Ray 1986, Cryptocoryne, specialized contractile roots occur which 1987a–c, 1988, 1990). In virtually all genera the mature prevent the stem from rising too near to the soil surface; stem is a sympodium composed of sympodial units (arti- more details of root structure are given in Chapter 3. cles) each of which has a more-or-less determinate structure, beginning with a prophyll and ending with an inflorescence or aborted inflorescence. Foliage leaves Stem and cataphylls (reduced sheath-like leaves) occur in a sometimes very regular sequence within each unit. The stem varies from an aerial elongated axis with Continued growth of the stem takes place in most gen- extended internodes, as in the many climbing hemiepi- era by the development of a “continuation shoot” in the phytes, to a hypogeal rhizome or subglobose tuber. axil of the leaf (foliage leaf or cataphyll) situated at the Climbing genera with long internodes are commonest second node below the spathe insertion. In the subfam- in the more primitive tribes, i.e. those with bisexual ily Orontioideae it arises at the first node below the flowers. Geophytes are found throughout the family spathe. The production of more than one inflorescence but are especially common in the most advanced sub- to form a floral sympodium commonly takes place by the family, the Aroideae. Abbreviated aerial stems, resulting development of short units consisting of a prophyll, a in rosulate plant forms, are also commonly found, as spathe and a spadix. The first unit arises in the axil of the in many epiphytic species of Philodendron and Anthurium. Some, generally larger, species have an Continuation shoot arborescent habit, in which the main axis is a fleshy (Alocasia, Xanthosoma) Spathe or fibrous (Philodendron) stem, or a pseudostem of petiole sheaths Prophyll Spadix (Arisaema, Typhonodorum). Shoot Peduncle types apparently specialized for vege- tative reproduction occur in various Sympodial leaf forms. Flagelliform shoots, or “flagel- lae”, equivalent to aerial stolons, have been observed in Amydrium, Cercestis, Pedicellarum, Philodendron, Pothos, Rhaphidophora, Rhodospatha, and Syngonium among others. They consist of branches (usually in the form of con- tinuation shoots) in which the internodes become very much longer and more slender than in the flowering zone of the stem, and the leaves often become reduced in size, sometimes to Monopodial leaf small, scale-like cataphylls. Flagelliform shoots grow rapidly, and thus encounter new host trees on which flowering stems later develop. Bulbils, which appear to be dispersed by birds, occur in Remusatia, while tubercles occur in Amorphophallus bulbifer, Dracontioides, Dracontium, and Figure 5. Schematic diagram of a common form of shoot architecture in Araceae. 6 THE GENERA OF ARACEAE

leaf immediately below the spathe and succeeding ones pinnatisect and radiatisect. Sometimes the posterior divi- in the axils of the prophylls. These floral sympodia have sions of pedatifid and pedatisect leaves are twisted a range of structural variation which may be quite com- spirally so that the leaf segments resemble a spiral stair- plex; extreme forms are found in Homalomena (Ray case (Eminium, Helicodiceros). Bipinnatifid, tripinnatifid 1988). Anomalous shoot organization of an apparently and partially quadripinnatifid leaves also occur. unique type occurs in Gymnostachys, while in most tribe Dracontioid leaves may be viewed as elaborated forms Potheae and Heteropsis the flowering axes occur as lat- of sagittate, hastate, trisect or pedatisect leaves in which eral short shoots on monopodial main vegetative axes of the anterior and posterior divisions of the leaf blade apparently indeterminate growth. have become highly dissected during ontogeny by dif- ferential growth of the margin, necrosis of parts of the Leaf blade, or a combination of both processes. The mature leaf can be compared to an umbrella which has been In virtually all genera the leaf is clearly differentiated into blown inside out by a gale — the spokes correspond to an expanded blade, petiole and petiole sheath; excep- a much-divided system of major leaf ribs and the canopy tions are Gymnostachys and some Biarum species. The is rent into numerous “tattered” lamina lobes which sheath normally clasps the subtended internode, at least adhere to the ribs and are more regularly shaped at the basally, and has an annular insertion (except many tips of the major leaf veins. Gonatopus has a somewhat Potheae and most Heteropsis). The foliage leaves that similar leaf form while Zamioculcas and certain species occur nearest the end of sympodial shoot articles (sym- of Anaphyllum and Anaphyllopsis have completely pin- podial leaves) often have short or very reduced sheaths, natisect leaves. The first foliage leaf of a seedling is particularly when the article apex aborts and fails to always entire (sagittate–hastate) in the genera develop an inflorescence, e.g. in Philodendron. Anaphyllopsis, Anaphyllum, Anchomanes, Dracontium, and Pycnospatha, but divided (dracontioid type) in the The terminology adopted here to describe leaf shape genera Gonatopus and Amorphophallus. requires some explanation as it differs from traditional practice. For descriptive purposes, the leaf is divided into Heteroblasty is a striking and sometimes taxonomi- the anterior division, corresponding to that part of the cally useful feature (e.g. Madison 1977a) of a number of blade surrounding the midrib, and two posterior divi- climbing genera (Cercestis, Monstera, Philodendron, sions, which, when present, are those portions of the Pothos, Rhaphidophora, Rhodospatha, Syngonium). It leaf blade which extend basally on each side of the occurs both in ontogeny from seedling to the mature petiole insertion. In many genera, e.g. those comprising plant and in association with the development of flagelli- the Monsteroideae, there are no posterior divisions and form shoots. A very striking form of heteroblasty is the blade is composed entirely of the anterior division. shown in certain genera (e.g. Monstera, Pothos) where In other taxa, such as the Lasioideae, many species have the juvenile leaves have very short petioles and their deeply sagittate leaves with very strongly developed blades are held flat against the host tree in a regular, over- posterior divisions, sometimes exceeding the anterior lapping sequence giving the appearance of roof shingles division in length. In strongly sagittate leaves, each pos- or tiles; these are consequently known as shingle plants. terior division usually has a well developed basal rib, Monstera dubia is a well known example which also has which performs the same mechanical support role as the beautifully variegated leaf blades in this growth phase. midrib does for the anterior division. In cordate and cordate-sagittate leaves the basal ribs may be short or Perforated (fenestrate) leaves are another peculiar- even absent, with the individual primary lateral veins ity of Araceae in genera such as Dracontioides, arising independently at the base of the midrib. On the Monstera, Rhaphidophora, and in juvenile leaves of other hand, pedately divided leaves, as seen for exam- Anchomanes and certain species of Cercestis. A num- ple in Philodendron goeldii or Sauromatum venosum, ber of genera have species with peltate leaves (tribe have a central, undivided anterior division while the Colocasieae, Anthurium, Caladium, Homalomena). posterior divisions are represented by the lateral series of pedate segments on either side. Here the basal ribs The midrib is almost always present, being absent are represented by the “arms” on which the segments of only in Gymnostachys and Pistia. The major veins the posterior divisions are inserted and which arch back which compose the midrib and basal ribs and branch from the midrib insertion at the apex of the petiole. laterally from them are termed primary lateral veins. Secondary, tertiary and higher orders of lateral veins are Leaf blade size and shape is exceedingly diverse. recognized by their relative thickness and/or their hier- Size may range from diminutive (e.g. Ambrosina) to archical level of branching. The primary lateral veins gigantic (e.g. Alocasia, Amorphophallus, Anchomanes, may be arcuate-parallel (e.g. Ambrosina), pedate (e.g. Cyrtosperma, Xanthosoma ). Shape varies from linear Sauromatum) or radiate (e.g. many Arisaema spp.) but (Biarum, Jasarum) to dracontioid (tribe Thomsonieae, most commonly are pinnately arranged. Even in pedat- Anchomanes, Dracontium, Pseudohydrosme, ifid (-sect) and radiatisect leaves, the primary lateral Pycnospatha, ), through elliptic, ovate, cordate, sagittate, veins of each segment are generally pinnate. hastate, trifid or trisect, pedatifid, pinnatifid, pedatisect, Except in deeply divided leaves, the primary lat- eral veins always run throughout the leaf blade, ultimately joining together at the leaf apex (Ertl 1932). VEGETATIVE MORPHOLOGY 7

The primary lateral veins generally run to the margin guish it from true, grass-type parallel venation, which first, where they form a marginal vein which then in Araceae occurs only in Gymnostachys. A third type runs to the leaf apex. In some species, most primary of fine venation (“colocasioid venation”) has been rec- lateral veins curve arcuately within the margin to fuse ognized by previous authors for tribes Colocasieae and together at the apex, and in these cases only the low- Caladieae. In this pattern the finer veins branch almost ermost primaries run into the margin to form a at right angles from the primary lateral veins and then marginal vein. In other genera either one or several arch strongly towards the leaf margin, often fusing on of the primary lateral veins form a submarginal col- the way to form a more-or-less sinuose interprimary lective vein (brochidodromous pattern) which lies collective vein and then finally joining into a submar- parallel to the marginal veins. ginal collective vein. Intermediates occur between most recognized types. Venation patterns and their devel- The finer venation may be reticulated (e.g. opment require new investigation, particularly because Anthurium) or may run essentially parallel to the pin- of their great potential for facilitating identification of nately arranged primary lateral veins (e.g. species and genera. Ertl’s study (1932) is the only large Philodendron). This is often referred to as “parallel” or scale comparative survey yet made of this important “striate” venation in aroid literature. In this book we character field. refer to the latter type as parallel-pinnate, to distin- ANTERIOR DIVISION Primary lateral vein Midrib Submarginal collective vein Petiole insertion ANTERIOR POSTERIOR POSTERIOR DIVISION DIVISION DIVISION B Basal ribs Petiole ANTERIOR DIVISION POSTERIOR POSTERIOR DIVISION DIVISION Petiole Basal ribs C Petiole A Figure 6. Leaf division types: A, no posterior division development; B, moderate posterior division development; C, extreme posterior C division development. 8 THE GENERA OF ARACEAE

C 3 VEGETATIVE ANATOMY By James C. French The vegetative anatomy of the Araceae is among the growth is relatively limited and often dependent upon most diverse of any family of monocotyledons. Stem contact with a substrate. With no contact, growth ceases vasculature is the most diverse of any monocotyledon precociously, whereas prolonged contact stimulates group and virtually every known type of secretory struc- elongation. Anchor roots never attain the enormous ture occurs, including resin canals, laticifers, extrafloral lengths of feeder roots. In Monstera (Madison 1977a) nectaries, mucilage cavities and intravaginal squamules. the stems not in contact with a substrate produce only a tuft of dried anchor roots, while those in contact pro- Root duce anchor roots 20–30cm in length. The latter may surround tree trunks, for example. Root hairs serve to The first root to emerge from the seed is a short-lived attach the root to a substrate, often forming a pseudo- extension of the radicle. Subsequent roots arise from parenchymatous layer (Went 1895). Some anchor roots the stem and may form lateral roots. The stem is the may lack root hairs and appear to be cemented by normal site of root origin in Araceae, a pattern referred dried mucilage, as in Syngonium. to by Troll (1949) as “secondary homorhizy” and typ- ical of monocotyledons. Feeder roots are thicker than the associated anchor roots and are capable of considerable elongation. In Anchor and feeder roots some species of Monstera they may extend 30m to the forest floor (Madison 1977a). Feeder roots are consid- Aerial roots of epiphytic and climbing Araceae are ered positively gravitropic, although their tips do not often specialized into anchor roots (Haftwurzeln) that hang vertically in some cases (Linsbauer 1907). Feeder serve to attach the plant to the substrate which pro- roots may hang freely in the air for their entire length vides support, and feeder roots (Nährwurzeln) which or they may adhere to a tree trunk, e.g. Syngonium extend to the soil and supply water and dissolved (Troll 1941) even though no root hairs are produced. nutrients (Went 1895). Mucilage may play a role in their adhesion as it is often more abundant than on anchor roots (Went 1895). A number of anatomical, morphological and physio- Feeder roots do not generally branch before reaching logical differences between anchor and feeder roots have the soil unless injured. However, Philodendron been described (Schimper 1888, Lierau 1888, Wettstein melanochrysum and some Syngonium species nor- 1904, Gaulhofer 1907, Linsbauer 1907, Porsch 1911, mally produce aerially branched feeder roots (Troll Solereder & Meyer 1928, Funke 1929, Goebel & Sandt 1941). The lateral roots of decapitated feeder roots also 1930, Birdsey 1955, Guttenberg 1968, Madison 1977a). lack root hairs and show positive gravitropism. Once Both anchor and feeder roots typically arise close to the the typically unbranched feeder roots of most species node, although the former may arise along the internode enter the soil they branch profusely. as well in some species of e.g. Anthurium, Epipremnum, Monstera, Rhaphidophora and Scindapsus (Madison In Monstera (Madison 1977a), dimorphic roots do 1977a, Croat & Baker 1978, Hotta 1971). According to not appear until plants have reached about a metre Troll (1941), the association between anchor and feeder above the ground and the stem is 7–10 mm thick. After roots can be highly specific, as in the monophyllous a prolonged period without contact, as in a pendent sympodium of Philodendron. In P. scandens a feeder shoot, only tufts of non-growing anchor roots are pro- root develops near the insertion of the foliage leaf, while duced. Hinchee (1981) studied root development in anchor roots develop nearby at the insertion of the sub- Monstera before and after entry into the soil. Unfortu- adjacent prophyll. In Araceae with root dimorphism nately Hinchee used the term Haftwurzeln erroneously, anchor roots are more numerous than feeder roots at the applying it to feeder roots prior to their entry into the same node. soil, and reserved the term Nährwurzeln for their sub- terranean branches. This misapplication of terminology The morphology and physiology of anchor and invalidates comparisons that are made with previous feeder roots are markedly different. The former are typ- studies of true anchor and feeder roots (Guttenberg ically relatively narrow, agravitropic and appear to be 1968) since Hinchee did not study Haftwurzeln. negatively phototropic, which tends to bring them into contact with the substrate. Numerous root hairs are typ- Some climbing Araceae, such as Scindapsus ically produced on the side adjacent to the substrate, but pothoides (= S. hederaceus) do not show a clear dis- may develop all over the root (Troll 1941). Anchor root tinction between anchor and feeder roots (Went 1895). In these instances anchor roots develop all over the stem, but are larger and longer at the nodes and may reach to the soil. Internodal roots are also found in VEGETATIVE ANATOMY 9

Anthurium sect. Polyphyllium (Croat & Baker 1978), as Nest roots in A. clidemioides. Some feeder roots of Araceae are capable of contraction (Wettstein 1904, Rimbach 1922). Another type of aerial root has been described in Anchor roots, on the contrary, do not appear to con- Anthurium ellipticum and is characteristic of many tract. Rimbach found that contraction was absent in other species of Anthurium sect. Pachyneurium, the feeder roots of Monstera deliciosa but present in “bird’s nest” or “litter basket” anthuriums. This is the so- Philodendron bipinnatifidum where he observed that called nest root (Schimper 1888, Bruhn 1910). These a feeder root contracted by about 20% over 5 months. exhibit negative gravitropism (Troll 1941) and branch profusely forming an “impenetrable” nest. The devel- Root growth opment of these roots was studied by Bruhn (1910) who found that their formation was typically inhibited The average length of the elongation zone of when the roots were surrounded by moist moss or Philodendron species has been observed as 20–50 earth. Bruhn concluded that they resulted from dam- mm, with the maximum found in P. selloum (= P. age to the root apex. However, more recent field bipinnatifidum), amounting to over 90 mm. Hinchee studies suggest that nest roots are mainly for feeding. (1981) reported that the elongation zone in Monstera In these epiphytes the rosulate leaves form a “basket” aerial (probably feeder) roots may be as long as 105 into which leaves, twigs and other detritus gather. The mm. Actual rates of growth for Philodendron aerial nest roots seem to be especially adapted to exploit roots were less than 10 mm/day for most species. The this food resource, growing directly up into the mass large roots of P. selloum (= P. bipinnatifidum) grew of detritus and ramifying within it (Croat 1991). exceptionally fast, from 7–21.5 mm/day. Daily growth rates in Monstera deliciosa ranged from 3–40 mm/day Contractile roots for aerial roots. According to Hinchee (1981) aerial roots of Monstera exhibit an exponential increase in These roots occur in numerous Araceae which possess growth rate during aerial growth. Once roots enter a tuberous habit, for example Typhonium (Banerji the soil their rate of growth declines and maintains 1947), Arum (Rimbach 1897), other genera of subfam- a relatively constant value. ily Aroideae (Rimbach 1898), subfamily Orontioideae (Lysichiton, Orontium, Symplocarpus), some genera Adventitious roots of subfamily Lasioideae (Hotta 1971), and some other aquatic species, as in Cryptocoryne, but not Calla Adventitious roots arise from the inner stem cortex at (Dudley 1937). some distance from the shoot apex, and they often have a broad attachment to the vascular cylinder of the Detailed studies of root contraction have been made stem, e.g. in some Pothoideae and Monsteroideae in two species of Arum. Rimbach (1897) followed the (French & Tomlinson 1981a, b). The adventitious roots life history of Arum maculatum in the field, presum- of Pothos, Pothoidium and Heteropsis have a close spa- ably in Germany, from seed germination to adult plant tial relationship to leaf traces. In Pothos and Pothoidium and studied in particular the phenology of shoot and each lateral leaf trace diverges from the central cylin- root growth. Lamant & Heller (1967) studied the mech- der to the leaf through the region of the root trace anism of contraction in Arum italicum. They observed attachment. In Heteropsis only the mid-vein leaf trace that contraction was related to: 1) radial expansion of diverges through the region of root attachment (French cortical parenchyma, which results in longitudinal ten- & Tomlinson 1981a). sion, and 2) release of such tension by the collapse of proximal outer cortical tissues and shortening of the The origin of lateral roots in Araceae is typical of stele. Similar mechanisms occur in other contractile many monocotyledons and has been studied in roots and were discussed in more detail by Ruzin Anthurium, Colocasia, Monstera and Zantedeschia. (1979) and Jernstedt (1984). Galil (1978) found active Pistia is atypical in several respects. Firstly, lateral roots lateral contractile roots in shallow, horizontally atten- arise very close (110–350 µm) to the root cap and apex uated rhizomes of Arisarum, which pull the rhizome junction (Charlton 1983). The lateral root primordia of laterally rather than downwards. A second type of rhi- Pistia tend to arise regularly along the length of the zome, which grows vertically at deeper levels, lacks xylem strands in the central cylinder, but no evidence contractile roots and is non-mobile. was found for regular spacing between the rows of root primordia (Charlton 1983). Secondly the roots of Root epidermis Pistia are unusual in the Araceae in their development of a persistent pocket or “Tasche”, which is a kind of The epidermis of most aroid roots is composed of a “substitute” root cap. This structure also occurs in some single layer of cells which nearly always includes both other aquatic monocotyledons, e.g. Lemna, Spirodela, ground cells and root hairs. More than one epidermal Eichhornia (Guttenberg 1968). layer has been reported for a number of genera, 10 T H E G E N E R A O F A R A C E A E

including Aglaonema, Anthurium, Gonatopus and Root cortex Homalomena (Lierau 1888, Solereder & Meyer 1928), but these results were largely based on studies of In the roots of many Araceae an exodermis is present mature roots only, which do not provide conclusive beneath the epidermis, which becomes the outer pro- evidence. Root hairs were reported to be absent from tective layer if the latter is lost (French 1987b). The subfamily Calloideae (sensu Krause 1908) but the exodermis or “Interkutis” is suberized and the outer walls entire family is clearly not “without root hairs” as may become thickened as in some Anthurium species. described by Cronquist (1981). In fact two types of Patterns of cell shape are variable, with some roots hav- root hair development have been described (Leavitt ing exclusively elongated exodermal cells, as in Calla. 1904). In Type I, any protodermal cell may form a Other roots have alternating long and short cells, as in root hair, as in Aglaonema, Arisaema, Caladium, Monstera deliciosa (Sinnott & Bloch 1946, Hinchee 1981) Dieffenbachia and Zantedeschia. In Type II, root hairs or Anthurium (Leitgeb 1865). Although an exodermis is arise only from specialized trichoblasts, as in present in most Araceae examined so far (Olivier 1881, Anthurium and Monstera. Root hairs are usually not Schimper 1888, Solereder & Meyer 1928, Hinchee 1981, formed on feeder roots of hemiepiphytes until they French 1987b), it is by no means universal. An exoder- penetrate the soil (Solereder & Meyer 1928, Lierau mis is reportedly absent from Pistia (Guttenberg 1968). 1888, Guttenberg 1968, Hinchee 1981). However, anchor roots form abundant root hairs, in some cases Beneath the exodermis a specialized multilayered from every cell (Guttenberg 1968). In many species the sclerotic hypodermis develops in the roots of seven root hair tips of anchor roots are modified by branch- genera: Anubias, Cercestis, Culcasia, Furtadoa, ing or are flattened and become cemented to the Homalomena, Montrichardia and Philodendron substrate and to each other, possibly by exudates from (Lierau 1888, Solereder & Meyer 1928, French 1987a). the root (Richter 1901, Lierau 1888, Madison 1977a). The organization of the ground tissue of the cortex of Mucilage has frequently been observed on roots of Araceae varies considerably among different genera. In climbing Araceae (Went 1895, Guttenberg 1968). species with aerial roots the outer cortex contains Unusual scales have been described on the roots of chloroplasts. Collenchyma has also been reported in Pothos and Monstera resulting in a striped appearance this region in feeder roots of both Monstera (Hinchee (Richter 1901). 1981) and Philodendron (Porsch 1911) as well as other root climbers. The ground tissue of the cortex is usu- One of the most distinctive anatomical features of ally composed of relatively thin-walled, unlignified the roots of some Anthurium species is the presence parenchyma cells. In many aquatic or semi-aquatic of a true multiple epidermis which resembles the genera and some others, large schizogenous intercel- velamen of Orchidaceae and some other mono- lular spaces, or lacunae, are present. These occur in cotyledons (Guttenberg 1968) in having a thick Amorphophallus, Anchomanes, Calla, Colocasia, water-absorbing layer that is white when dry. Limited Dracunculus, Philodendron and Syngonium (Lierau developmental studies have shown that the proto- 1888). The feeder roots of Syngonium develop large derm of these species exhibits tangential cell divisions intercellular spaces, while the anchor roots do not resulting in multiple layers of epidermis (Guttenberg (Went 1895). Trichosclereids have been reported in 1968). In some Anthurium species the velamen has intercellular spaces only in Monstera deliciosa, long been recognized (Leitgeb 1865, Tieghem 1867, Rhaphidophora decursiva and Scindapsus pteropodus Lierau 1888, Schimper 1888). The multiple epidermis (=Rhaphidophora pteropoda) (Lierau 1888, Solereder & of Anthurium gracile exhibits a white velamen when Meyer 1928) despite their common occurrence in other dry (Croat 1984). Engler (1920b) noted that the mul- organs. Thick-walled cells, some with pits, are men- tiple epidermis of some Anthurium species, tioned by several authors (Lierau 1888) as occurring in particularly in section Pachyneurium, contains dead the cortex of some Monsteroideae, but have not been cells with fibrous thickenings and granular structures described in detail. on the innermost tangential wall layer. Densely cyto- plasmic “passage cells” have also been observed, A variety of secretory tissues occurs in the cortex. which are similar to cells found in some Orchidaceae Resin canals are present in the cortex of Philodendron, (Deshpande 1956). Fibrous thickenings have been Homalomena and Furtadoa (Trécul 1865, 1866, Lierau observed in A. affine, A. crassinervium, A. maximum, 1888, Porsch 1911, Engler 1912) and in Culcasia and A. willdenowii, among others, according to Lierau Cercestis (French 1987b). More details of organization (1888), Leitgeb (1865), Schimper (1888) and Solereder are given in the section on resin canals. Laticifers also & Meyer (1928). Some Anthurium species exhibit a occur in the root cortex of a limited number of species, smooth-walled multiple epidermis (A. huegelii (= A. including Syngonium (Weiss 1866). Numerous crystal- hookeri), A. cucullatum (= A. andicola); Schimper containing and tanniniferous idioblasts also occur in the 1888, Leitgeb 1865). root cortex (Solereder & Meyer 1928). The inner cortex typically contains narrower cells arranged in concentric rings adjacent to relatively smaller intercellular spaces in Amorphophallus, Colocasia, Dieffenbachia, Homalomena, Philodendron, Scindapsus, V E G E T A T I V E A N A T O M Y 11

Syngonium and other genera (Solereder & Meyer 1928). the root caps of dimorphic aerial roots. Earlier work- The inner cortical region of many Monsteroideae and ers (Haberlandt 1914) reported the presence of normal some Pothoideae contains a sheath of brachysclereids, “statolith” starch grains in the columellas of gravitropic fibres or sclerotic parenchyma, one to several layers feeder roots of Monstera deliciosa (and other species), thick (Tieghem 1867, Lierau 1888, Went 1895, Solereder which has been confirmed by Hinchee (1981). In & Meyer 1928, Sinnott & Bloch 1946, Hinchee 1981). agravitropic clasping roots it is generally agreed that the This sheath may be in direct contact with the endoder- size of the columella is smaller than in feeder roots and mis as in Spathiphyllum kochii (Lierau 1888) or separated fewer “statocytes” are present (Haberlandt 1914). from it by one or more cell layers. The latter condition Several reports contend that columella starch grains occurs in Epipremnum pinnatum (Tieghem 1867), sev- are “sluggish or motionless” in clasping roots eral species of Rhaphidophora (Solereder & Meyer (Gaulhofer 1907, Haberlandt 1914), however Linsbauer 1928), Pothos celatocaulis (= R. korthalsii) (Lierau 1888) (1907) found mobile grains. and P. scandens (Went 1985). Development of the mature cortex was studied in An endodermis with a Casparian strip is probably Monstera deliciosa by Bloch (1946), Sinnott & Bloch present in roots of all Araceae (Solereder & Meyer (1946) and Hinchee (1981). The exodermis has alter- 1928). Particularly in older aerial roots, the walls of nating short and long cells resulting from unequal the endodermis become thickened and suberized, thus divisions that occur 2.5–3.5mm from the apex. Beneath obscuring the Casparian strips. In most species suber- the exodermis certain cells in longitudinal files undergo ization begins first in the endodermis adjacent to the unequal divisions resulting in a smaller basal cell which phloem and only later progresses to the region adjacent then develops into a trichosclereid (Bloch 1946). The tri- to the xylem. This pattern is typical of monocotyledons. chosclereid initials develop adjacent to intercellular spaces into which the cell arms elongate intrusively. Vascular cylinder of roots Bloch (1946) found trichosclereid initials only at the distal ends of files, but Hinchee (1981) reports finding The vascular cylinder of roots of Araceae is typically small, densely cytoplasmic, putative trichosclereid ini- cylindrical and exhibits an alternating, radial arrange- tials elsewhere. Asymmetric cell divisions were also ment of xylem and phloem. However, lobing of the associated with the formation of raphide cells by Kovacs cylinder occurs in some species of Philodendron and & Rakovan (1975), but were not observed by Hinchee in Cercestis. Some species of Epipremnum and (1981). Sinnott & Bloch (1946) emphasized the role of Philodendron have converging V-shaped xylem unequal cell division, internal environment and cell strands with narrower phloem strands in their angles position in the differentiation of cells of the exodermis and between them. Anomalous organization of vas- and trichosclereids. In contrast, the development of the cular tissues occurs in some Monstereae (Monstera, inner layer of brachysclereids adjacent to the endoder- Rhaphidophora, Scindapsus), Anthurium and mis is not dependent on unequal cell division. Similar Philodendron. In these genera the vascular region of short sclereids develop at the surface in response to the central cylinder consists of interspersed strands of wounding of the root, by redifferentiation of their phloem and xylem. The latter are generally one ves- walled cortical cells, which Sinnott & Bloch (1946) inter- sel in width while the phloem strands comprise several preted as an effect of a changed environment. wide sieve elements (Hinchee 1981). Ground tissue of the central cylinder often becomes sclerotic in older Periderm formation roots. A pericycle has been reported (Hinchee 1981) and a pith region may be present or absent. Meyer A cork cambium is generally formed in the cell layer (1925) followed the course of the inner vascular tissue below the exodermis in aerial roots of Monstera (Richter in roots of Anthurium and Monstera and found that 1901, Hinchee 1981), Anthurium and Rhaphidophora phloem and xylem strands remained separate in (Olivier 1881), and replaces the exodermis as the out- Anthurium wagenerianum. In Monstera deliciosa ermost protective layer. In Monstera deliciosa the connections were observed between vascular strands, periderm begins to develop about 60 mm from the root both in the periphery and the centre of the root. apex in feeder roots and 10–15 mm from the apex of subterranean lateral roots (Hinchee 1981). Following Root apex cork formation, epidermal cells collapse in Monstera and Philodendron (Engler 1912). Cork may consist of Relatively few studies of apical organization have been alternating layers of thin-walled and sclerenchyma cells made in Araceae. Guttenberg (1968) described an open in Monstera (Hinchee 1981). In Philodendron the phel- pattern of apical organization. Hinchee (1981) demon- logen arises under layers of the sclerotic hypodermis and strated the presence of a quiescent centre in both the forms a wide layer of periderm in some species (Engler feeder root and its subterranean branches. 1912). Lenticels also occur in the roots of Anthurium and other species (Weisse 1897). The lenticels of More attention has been given to the structure of Monstera contain filling tissue as well as closing layers. 12 T H E G E N E R A O F A R A C E A E

Stem Calla palustris, have no endodermis and widely sep- arated central cylinder bundles. Root traces are Stem anatomy and morphology are frequently corre- typically inserted along the periphery of the central lated in Araceae, particularly with respect to the cylinder and serve as useful markers. overall organization of the vascular system (Engler 1920b, French & Tomlinson 1981a–d, 1983, 1984). The vascular bundles in the cortex generally Conducting tissue tends to be highly condensed in belong to two different systems: 1) leaf traces that tuberous species and in rhizomes with short intern- traverse the cortex but may remain there for a longi- odes, compared with scandent plants. Climbing tudinally short distance, never exceeding one genera are largely concentrated in subfamilies complete internode, and 2) cortical vascular bundles Pothoideae and Monsteroideae, but climbing that constitute a separate system, persisting in the hemiepiphytes occur in subfamily Aroideae as well, cortex over many internodes. including Cercestis and Culcasia (tribe Culcasieae), Philodendron (tribe Philodendreae) and Syngonium In all species in which leaf traces are followed (tribe Caladieae). basipetally (using cinematographic techniques), they appear to enter the cortex from the leaf at the node Grayum (1984) pointed out that it is sometimes dif- and traverse it before entering the central cylinder, ficult to distinguish between rhizomes and tubers in the where they ultimately attach to axial bundles. There Araceae. Tubers are common in subfamily Aroideae is considerable variation in the relative distance over (tribes Areae, Arisaemateae, Arisareae, Arophyteae, which leaf traces remain in the internode before Caladieae, Colocasieae, Nephthytideae, Spathicarpeae, entering the central cylinder. In most species the Thomsonieae, Zamioculcadeae, Zantedeschieae, major leaf traces enter the central cylinder close to Zomicarpeae). Hotta (1971) drew a distinction between the point of leaf insertion after traversing the cortex two types of tuber in Araceae, the Amorphophallus at an acute angle, as in Anthurium polyschistum type which is also found in Zamioculcas and (French & Tomlinson 1981a). The minor leaf traces Gonatopus, and the Arisaema type, which also occurs may remain in the inter-node for a variable distance, in Remusatia. The principal difference between them in some cases extending to the base of the subjacent lies in the way new tubers form and the ease with internode. Fibre bundles enter the cortex from the which they are separated. In Amorphophallus, succes- leaf in Homalomena and Pothos (and Acoraceae) sive tubers develop from each other without and end blindly in the inner cortex (Tieghem 1867, separating, while in Arisaema new tubers are sepa- Falkenberg 1876, French & Tomlinson 1981a,d) in a rated from the older tissue by cork layers. manner similar to palms and some other mono- cotyledons (Zimmermann & Tomlinson 1967, Stem epidermis Zimmerman, Tomlinson & LeClaire 1974). The stem epidermis exhibits many of the same fea- A separate cortical vascular system is present in a tures as that of the leaf with regard to occurrence of few genera and tribes, including Anthurium, Caladium, trichomes, which are discussed in the section on Chlorospatha, Dieffenbachia, Heteropsis, Monstereae, leaves. Older stems of many Araceae develop a peri- Pothoidium, Pothos, Philodendron, Syngonium and derm in the outer cortex or epidermis which Xanthosoma (French & Tomlinson 1980, 1981a–d, 1983, ultimately leads to the death of the epidermis (French 1984, Grayum 1984). In addition to the persistent cor- & Tomlinson 1981a–d, 1983). tical bundles there may also be a series of minor and intermediate leaf traces. It is not uncommon for minor Stem cortex leaf traces to fuse with the cortical system and remain separate from the central cylinder. The cortex is delimited from the central cylinder by its less dense clustering of vascular bundles, and Cortical vascular bundles show anatomical diversity sometimes by the presence of an endodermis. The amongst the genera examined. However, they can usu- principal sources of histological variation in the cor- ally be recognized in single sections by the relatively tex are: 1) width, 2) vascular organization, and 3) small amount of vascular tissue as compared to that of various specializations of the ground tissue. In many leaf traces. The latter tend to be wider and if they are Monsteroideae, some Pothoideae and other genera persistent they are located nearer to the central cylin- like Culcasia, Cercestis and Philodendron, an espe- der. In many Caladieae with a cortical vascular system, cially distinct inner cortical boundary is present the cortical bundles are narrow and form a highly anas- because the sclerotic sheaths of the peripheral vas- tomosing system, as in Chlorospatha longipoda and cular bundles of the central cylinder are fused Xanthosoma tarapotense. Laticifers are typically present together. In many Monsteroideae the inner layers of and are associated with the phloem in Caladium, the cortex also become sclerotic. Some species, like Dieffenbachia, Philodendron and Xanthosoma. The ground tissue of the cortex generally consists of unlignified parenchyma with small intercellular spaces, except in aquatic genera such as Calla and Montrichardia, where a more extensive system of schizogenous intercellular spaces develops. V E G E T A T I V E A N A T O M Y 13

Trichosclereids typically develop in the intercellular Organization of stem central cylinder spaces of subfamily Monsteroideae, Pothos and Pothoidium (French & Tomlinson 1981a). Chloroplasts The central cylinder contains ground tissue and vas- are present in the ground tissue beneath the epider- cular bundles that are generally dispersed throughout. mis. A continuous ring or separate strands of In many species, the bundles are somewhat more collenchyma may develop in the peripheral cortex closely spaced towards the periphery and have a of only certain genera, including Syngonium (Birdsey smaller diameter than those closer to the centre of the 1955), Dieffenbachia (Tieghem 1867), Aglaonema, axis. The ground tissue is rarely lignified except in Asterostigma, Homalomena, Philodendron, Pothos, Pothoidium, Heteropsis and in Philodendron Schismatoglottis, Spathantheum and Zantedeschia subgen. Pteromischum, where usually the peripheral (Engler 1920b). A wide variety of secretory tissues is and sometimes the entire ground tissue is lignified. A present, including resin and mucilage canals and cav- wide range of secretory elements and idioblastic cells ities, laticifers and a variety of idioblasts. Many of occurs both in the ground tissue and in association these structures also occur in other parts of the plant with the vascular tissue. and are discussed later. The general principles of vascular organization in The inner boundary of the stem cortex is some- Araceae follow those of palms and other monocotyle- times demarcated by an endodermis with casparian dons (Zimmermann & Tomlinson 1967). According to strips, and has been described from Rhaphidophora these concepts there are two kinds of vascular bundles celatocaulis (= R. korthalsii) (Solereder & Meyer in the central cylinder: 1) leaf traces that ultimately 1928), Scindapsus pictus and Anthurium (Solereder & enter a leaf when followed distally, and 2) axial bun- Meyer 1928) as well as various species of Amydrium, dles that remain in the central cylinder. The leaf traces Monstera, Orontium, Rhaphidophora, Spathiphyllum, have protoxylem, which easily distinguishes them from Symplocarpus, tribe Schismatoglottideae, tribe axial bundles, which have only metaxylem. Leaf traces Peltandreae and Pistia (French & Tomlinson 1981a–d, of some genera such as Cercestis and Syngonium can 1983). There is some positive correlation between be distinguished from axial bundles by the presence of the aquatic and rhizomatous habits and the occur- associated laticifers (French & Tomlinson 1981c, 1983). rence of an endodermis, but there are also aquatic Leaf traces have a collateral organization throughout genera that lack an endodermis, including Calla, the family. Cryptocoryne and Lagenandra. Stem endodermis has been sought in the majority of Araceae genera and Four general types of axial bundles can be recog- shown to be absent (French & Tomlinson 1980, nized: simple collateral bundles, compound bundles, 1981a–d, 1983, 1984). intermediate forms between compound and amphivasal bundles, and amphivasal bundles, includ- When an endodermis is present it may have sev- ing frequently branched forms that seldom have a eral patterns of distribution, including unusual types complete cylinder of xylem. There is a structural con- (French & Tomlinson 1981a–d, 1983). It may encircle tinuum between all four categories (Tieghem 1867, the entire central cylinder, interrupted only where French & Tomlinson 1981a–d, 1983, 1984). Compound leaf traces depart, as in Orontium and Symplocarpus. bundles in Araceae were first observed by Tieghem. He In Scindapsus and Rhaphidophora the endodermis correctly interpreted the presence of these in some may develop only on the ventral side of the central species of Philodendron and in Dieffenbachia (French cylinder, in association with a vascular plexus of root & Tomlinson 1981d, 1984), but did not investigate traces to which the roots attach. An unusual pattern Montrichardia, Rhodospatha and Stenospermation, is found in tribes Schismatoglottideae and which also have compound bundles. Peltandreae. The endodermis with casparian strips has two different patterns in these tribes: 1) it sur- Tieghem thought that all Araceae with unisexual rounds only the individual peripheral axial bundles of flowers had compound bundles. However, in virtually the central cylinder (i.e. no endodermis is present all cases, e.g. Syngonium (French & Tomlinson 1983), around interior bundles or around the entire central these are not true compound bundles but various types cylinder), or 2) it surrounds all of the axial bundles of of amphivasal bundles that branch and anastomose the central cylinder but does not surround the entire among themselves. The essential difference between central cylinder or individual leaf traces. The first pat- the two types hinges on whether the distinct vascular tern occurs in some species of Schismatoglottis, components can be recognized, as in compound bun- Hottarum and Typhonodorum. The second is present dles, or whether the xylem elements tend to form a in some species of Aridarum, Bucephalandra, cylinder surrounding a core of phloem, as in amphivasal Peltandra, Phymatarum, Piptospatha and Schisma- bundles. In most genera the distinction is clear. toglottis. The occurrence of an endodermis around However, in Philodendron a wide array of bundle orga- individual vascular bundles is unusual in angiosperms nization is present from simple collateral to compound (cf. Gunnera) but does occur in some ferns and in and amphivasal (French & Tomlinson 1984), and Equisetum (Esau 1965). numerous intermediate types are also present. Axial bundles in Dieffenbachia show a similar variation in organization and cannot readily be categorized. 14 T H E G E N E R A O F A R A C E A E

Simple collateral axial bundles occur infrequently in Typical amphivasal bundles with a core of phloem Araceae and tend to be present in genera with bisex- that is irregular or circular in transverse section are ual flowers (Tieghem 1867), although there are also present in Acorus and Araceae, as in many numerous exceptions. From the systematic and mor- Colocasieae and Caladieae. The phloem is sur- phological viewpoints, the distribution of collateral rounded by xylem that contains tracheary elements bundles is interesting. They occur in all genera with having variable patterns of organization; they may be bisexual perigoniate flowers except Gymnostachys, relatively narrow and contiguous (Cryptocoryne) or tribe Spathiphylleae, and subfamilies Orontioideae and exist as individual elements, or clusters that have vari- Lasioideae. Most genera with collateral bundles have able spacing (Philodendron). elongated internodes and many are climbers. Most genera with bisexual perigoniate flowers but lacking Course of vascular bundles collateral bundles, such as Spathiphyllum, Holochlamys, subfamily Orontioideae, most Lasioideae The three-dimensional organization of stem vasculature and Gymnostachys, have relatively congested intern- of Acorus and about sixty genera of Araceae has been odes, suggesting a strong correlation between this type analyzed using cinematic techniques (French & of vascular bundle and habit. Except for Philodendron Tomlinson 1980, 1981a–d, 1983, 1984). These methods and Syngonium, Araceae with unisexual flowers are are necessary to understand the complex organization rarely scandent and amphivasal bundles predominate. of stem vasculature in monocotyledons, which con- Exceptions include collateral bundles in the climbing tains numerous vascular bundles in a dispersed or prostrate plants of the genera Culcasia and Cercestis. arrangement. A wide range of types of organization The presence of collateral bundles tends to be corre- was revealed and these can be grouped conveniently lated with prominent sclerenchyma sheaths, which are into four patterns. well developed in many species. Considerable scle- renchyma occurs in the rigid stems of Pothos, Pattern 1 is similar to that of some palms such as Pothoidium, Heteropsis and Philodendron subgen. Rhapis (Zimmermann & Tomlinson 1967), and involves Pteromischum. The habit of these genera is also dis- a continuing axial bundle apparently following a sig- tinctive in the Araceae because the climbing stems moid curve through the axis, giving rise to branches give rise to lateral branches which arch outward and which become leaf traces as they bend towards the hang down forming shrubby or tangled masses of periphery. Developmental studies in palms have shown branches. It is possible that increased stem rigidity in that the leaf trace actually branches to form the con- these genera is functionally correlated with their habit tinuing axial bundle. The Rhapis-type pattern occurs in (Grayum 1984). Acorus, which has amphivasal bundles (Mangin 1880). Mangin’s remarkable study is probably the earliest The second major type of vascular bundle in Araceae accurate account of the “palm” pattern in mono- is the compound bundle, recognized by Tieghem (1867), cotyledons and has unfortunately been long neglected. which typically consists of two to five or more simple In Acorus, axial bundles tend to become part of a collateral bundles clustered together within a common branched network of peripheral vascular bundles after bundle sheath. Two major characters are used to iden- the departure of leaf traces. This pattern has also been tify the compound bundle. First, component vascular observed in palms such as Chamaedorea. bundles tend to remain clustered as they follow a com- mon course through the stem, although the distance The palm type occurs in Anthurium polyschistum over which the components remain together is variable. and the rhizome of Spathiphyllum cannifolium, which Second, individual components consist of discrete col- both have simple collateral bundles (French & lateral bundles with their phloem strands directed Tomlinson 1981a). Culcasia saxatilis and Philodendron towards the centre of the cluster. In some Araceae there hederaceum also have this basic organization. are compound bundles with discrete components that remain together over relatively long distances, as in Pattern 2 is generally similar to the palm type Rhodospatha (except R. venosa), Montrichardia, because at regular intervals a continuing axial bundle Philodendron (some sections only), Stenospermation branches from the leaf trace. The principal difference and Dieffenbachia (some species). is the occurrence of unpredictable branching and anas- tomosis of axial bundles between departures of leaf In other Araceae there are complexes of vascular traces. This pattern occurs widely in Araceae. It is tissue that migrate together but have less discrete com- present in subfamilies Calloideae and Orontioideae, ponents, in that the phloem tends to form a solid core most Lasioideae, and most Aroideae (French & rather than separate strands. Examples include Tomlinson 1980, 1981a–d, 1983, 1984). Highly con- Zamioculcas, adult stems of Cercestis, many species of densed pattern 2 vascular systems are typical of Philodendron and some species of Dieffenbachia. species with tuberous stems and rhizomes, or erect These examples lead to the conclusion that it is not stems with condensed nodes, which are common in possible to precisely circumscribe compound bundles subfamilies Lasioideae and Aroideae. Individual axial from a structural viewpoint, since there is a continuous bundles in pattern 2 typically have some variation of intergradation with amphivasal bundles. the amphivasal pattern of organization. V E G E T A T I V E A N A T O M Y 15

Pattern 3 exhibits a relatively distinct organization occur in climbing Cyclanthaceae (French, Clancy & and is restricted to a few genera of Araceae. It has not Tomlinson 1983). been described in any other monocotyledon. The basic feature of pattern 3 involves the formation of a series Three major patterns of bud trace insertion are rec- of narrow “bridges”, or lateral branches from both leaf ognizable in the Araceae. In the first pattern two clusters traces and axial bundles basally, which then aggregate of bud traces follow an arching course from the bud to form a new axial bundle that distally departs as a leaf into the central cylinder and through the subperipheral trace into a leaf. The pattern is referred to as the “basal or peripheral area. The second pattern also involves the aggregation” type because axial bundles appear to arise formation of distinct clusters, but these migrate around by the aggregation of a series of bridges or branches. the outside of the central cylinder in the inner cortex. Basal aggregation occurs in Anadendrum, Pothoidium, They subsequently enter the central cylinder with leaf Pothos and certain Monstereae, including Amydrium, traces and often form a cluster around an individual Epipremnum, Monstera, Rhaphidophora, and trace. The third pattern is largely inconspicuous and Scindapsus. nondescript. It involves the fusion of a few traces directly with the surface bundles of the central cylinder Pattern 4 includes a relatively small number of gen- opposite the bud. This pattern can intergrade with the era with distinct compound vascular bundles, including first type and is typically associated with the rhizoma- Dieffenbachia, Montrichardia, Rhodospatha (all but tous or tuberous habit. The formation of extensive bud R. venosa), Stenospermation, some Philodendron spp., trace systems (patterns 1 and 2) in climbing species of and genera such as Cercestis and Zamioculcas, with Araceae appears to be an adaptation which ensures intermediate conditions. adequate vascular connections with lateral branches that have no other immediate source of nutrients. In the compound bundles of Montrichardia, Philodendron, Rhodospatha and Stenospermation, the Internode development individual components are relatively distinct and can remain independent over several millimetres. A pre- This has been studied in a small number of climb- dictable pattern of changes in course and number of ing species, using measurements of cell length, mitotic components in a bundle was not detected, in contrast index and marking studies of elongating internodes to the compound bundles of Cyclanthaceae (French, (Fisher & French 1976, 1978, French & Fisher 1977b, Clancy & Tomlinson 1983) and Pandanaceae French 1977). The stems of these Araceae contain so- (Zimmermann, Tomlinson & LeClaire 1974). called uninterrupted meristems, in which an acropetal Components may fuse together laterally, separate from wave of cell maturation proceeds through successive the compound bundle or may split apart in various internodes, involving an end to cell division, an abrupt patterns. New bundles may be added from surround- increase in cell length, and ultimately cessation of ing compound or simple bundles. In these genera there growth. Marking experiments show that elongation is an apparently random series of interchanges continues for a longer time and at a slightly higher between components both within one bundle and rate in the upper region of the internodes of between adjacent bundles. This unpredictable feature Anthurium, Epipremnum and Philodendron, leading to also occurs in Dieffenbachia, in which sclerenchyma is differential growth of the upper region (French & minimal and bundle components are not well sepa- Fisher 1977b, French 1977). rated in some species. Leaf In Dieffenbachia and in some Philodendron species there are closely spaced bifurcations of bun- dles and loose association of components, which intergrade with amphivasal bundles. A similar situation exists in Cercestis (adult stem) and in Zamioculcas. Bud traces Epidermis Both the organization of bud traces and their pattern Trichomes and larger emergences such as prickles of insertion into the main axis vary considerably in are highly unusual in leaves or stems of Araceae Araceae. There are systematic correlations as well as (Solereder & Meyer 1928, Grayum 1984) and detailed associations with the basic biology of stem growth. anatomical descriptions are usually lacking. Unicellular Unfortunately this subject has not received much hairs occur in some species of Xanthosoma (e.g. X. study in monocotyledons so there is minimal infor- pubescens) and short 2-celled hairs are reported for mation for broad comparisons. While most Araceae Anubias barteri (Solereder & Meyer 1928). Multicel- have one bud per node, some, like Xanthosoma and lular hairs occur in abundance on leaves and stems of Dracontium, develop multiple buds and others like some neotropical Homalomena species (section Remusatia produce bulbils (Engler 1920b, Möbius Curmeria), a few species of Schismatoglottis and in 1935). Some similar types of bud trace insertion also Pistia. Trichomes are also reported for leaves of Bognera (Madison 1980), Callopsis, and some species 16 T H E G E N E R A O F A R A C E A E

of Stylochaeton (Bogner 1984c, Mayo 1985a) but ata are present on the adaxial surface only, while detailed anatomical details are not available. Large, Orontium is entirely epistomatic. The guard cells are scale-like or filamentous emergences are present on typically level with the other cells of the epidermis, leaves of some species of Philodendron section but they may also be sunken, as in Orontium, Philodendron (syn. sect. Polyspermium) (Krause 1913, Ariopsis and Pistia. Engler 1920b). Scale-like emergences are found on stems of Syngonium podophyllum var. peliocladum Ground epidermal cells tend to have a polygonal (Birdsey 1955, Croat 1982). The petioles, leaf blades outline, with 5–8 straight sides and the surface usually and stems of some genera are armed with sharp prick- lacks papillae or trichomes. Undulate cell walls occur in les, as in Anchomanes, Cyrtosperma, Lasia, one or both epidermal layers and over all or part of the Lasimorpha, Podolasia, Pseudohydrosme and occa- anticlinal walls in a few genera. Exceptionally wide sionally Nephthytis (Grayum 1984). Some neotropical ground epidermal cells reaching 94 µm in width occur species of Homalomena have numerous recurved in Zamioculcas, whereas in other Araceae they range prickles on their petioles. Prickle-like emergences also from 36–71 µm (Pant & Kidwai 1966). Epidermal cells occur on the stem of Montrichardia (Engler 1911, 1912, of some genera are papillate, giving the leaves a velvety 1920b). sheen (Dalitzsch 1886, Engler 1920b, Solereder & Meyer 1928, Birdsey 1955). The papillate epidermis of the The cuticle of Araceae varies considerably in thick- spathe of Arum *is highly unusual in having intercellu- ness and morphology (Solereder & Meyer 1928), but no lar spaces between the epidermal cells. This feature is systematic study has been made. Cuticular ridges occur part of a syndrome of adaptive characters linked to the in a number of genera including Culcasia, pollination mechanism and is correlated with the Dieffenbachia, Peltandra, Pothos, Spathiphyllum and absence of stomata in the spathe (Knoll 1923). others (Dalitzsch 1886, Webber 1960, Pant & Kidwai 1966). One or both leaf surfaces may have ridges. Waxy Leaf mesophyll deposits on the epidermis have been reported from Caladium, Colocasia, Lagenandra, Remusatia The mesophyll of Araceae is predominantly bifacial, (Gonatanthus) (Solereder & Meyer 1928), Syngonium with a thicker spongy layer below the palisade (Da- (Birdsey 1955), Carlephyton glaucophyllum and litzsch 1886). Isobilateral leaves are rare and include Xanthosoma violaceum (Bogner pers. comm.). The some species of Anthurium and Montrichardia in types of wax crystalloids in Araceae have a greater which the spongy layer is virtually absent and the pal- similarity to those of dicotyledons than monocotyle- isade is nearly uniform. In Typhonodorum two palisade dons (Behnke & Barthlott 1983). Hydrophilic globules layers are separated by a region of aerenchyma. Acorus occur on the upper epidermis of Orontium and lower calamus (Acoraceae) exhibits a band of 3–5 layers of epidermis of Peltandra. Franke (1967) and Maier- chlorophyllous isodiametric cells beneath the epider- Maercker (1981) have studied transpiration through mis, and a central region of large air cavities, which the cell walls and cuticle of Zantedeschia. significantly vary in size in different populations (Kaplan 1970, Röst 1979b). The organization of stomatal guard and subsidiary cells in the leaves of Araceae has been partially sur- The palisade parenchyma cells are generally short, veyed in mature leaves (Dalitzsch 1886, Solereder & relatively wide and may have arms, as for example in Meyer 1928, Webber 1960, Pant & Kidwai 1966, Bunting Gonatopus (Solereder & Meyer 1928). Long narrow 1968, Grear 1973, Grau 1983). Not surprisingly, there is palisade cells are of more limited occurrence and have great variability in the number of subsidiary cells between been observed in some species of Anthurium taxa and among individuals. The range is from zero (Dalitzsch 1886, Solereder & Meyer 1928). (anomocytic) as in Orontium, to two (paracytic), or four to eight (tetracytic) in Rhaphidophora (a more complete The spongy mesophyll consists of flattened cells list is given by Grayum 1984). The development of sub- with arms that are arranged in tiers, as in Anthurium sidiary cells in Araceae appears to follow the perigenous and Anubias. The spongy mesophyll has a distinctive pattern, that is, to have guard and subsidiary cells from chambered structure in leaves of many species of the same parent cell, which is typical of monocotyledons. Homalomena, Philodendron, Piptospatha, Very few aroid stomata have been examined develop- Schismatoglottis and Typhonium (Engler 1920b, Solereder mentally, however, which prevents the application of & Meyer 1928, Bunting 1968). In Typhonodorum stellate Tomlinson’s (1974) developmental system of classifica- parenchyma occurs in the partitions of such chambers. tion. Araceae lack the “oblique” cell divisions found in other monocotyledons such as palms (Tomlinson 1974). In some Araceae, a hypodermis is present, consist- ing of one to several layers of parenchyma cells lacking In the leaves of most aroids the stomata are either chloroplasts, as in Anthurium and other genera present only on the abaxial surface (hypostomatic) or (Dalitzsch 1886, Solereder & Meyer 1928). In are largely concentrated there (see Grayum 1984 for Philodendron Engler (1912) reported a cell layer with- a complete list). Shaw (1993) has studied the distri- out chloroplasts containing a red pigment lying bution of stomata on the leaves of Monstera deliciosa. adjacent to the lower epidermis. Williams (1994) made In Pistia the leaves are largely epistomatic, i.e. stom- a more recent study of some Philodendron species. V E G E T A T I V E A N A T O M Y 17

Leaf vasculature Philodendron, Schismatoglottis, Zantedeschia and most Caladieae and Colocasieae (Dalitzsch 1886, Early studies of leaf venation by Engler led him to Birdsey 1955). In some genera such as Amorpho- conclude that it was not a highly significant character phallus and Pseudodracontium, the vascular bundles for the classification of Araceae (Engler 1920b). have collenchymatous bundle sheaths (Dalitzsch Nevertheless, he used venation as a major character in 1886). In others, the support tissue is primarily bun- the definition of his subfamilies, especially his dle sheath sclerenchyma and may be fused together Philodendroideae and Aroideae (Engler 1920b), with to form a peripheral layer (Anthurium, parallel-pinnate and reticulate venation respectively. Rhaphidophora, Spathiphyllum). Fibre bundles may be present, with or without phloem, as in Ertl (1932), based on a survey of over 40 genera, Anthurium. The vascular bundles may be rather dis- developed an explanation for the apparently variable tant from one another and have large sclerenchyma nature of venation in Araceae based on differences in sheaths, as in Cyrtosperma and Lasia. In Orontium spacing between the primary lateral veins. No detailed and Calla mechanical tissue is absent from the peti- study has been made since, and although Hotta (1971) ole (Solereder & Meyer 1928). essentially endorsed Ertl’s conclusions, this hypothe- sis must be regarded as tentative. Ertl concluded that Leaf tubercles and regeneration overall leaf size and shape, that is, broad versus linear, as well as degree of dissection, have an important Tubercles regularly develop at the juncture of leaflet relationship to venation pattern. Although these mor- and petiole in Pinellia ternata (Hansen 1881, Linsbauer phological and developmental correlations are 1934, Troll 1939), at the apical end of petiole in important, systematic position is also significant. Some Typhonium bulbiferum (Sriboonma et al. 1994) and at the groups, such as tribes Colocasieae and Caladieae, first and second order divisions of the leaf of have a characteristic venation pattern that is indepen- Amorphophallus bulbifer (Troll 1939). Tubercles in dent of leaf shape. In these tribes there is a tendency Pinellia may also form spontaneously along the petiole for quaternary veins to arch over tertiary veins (Ertl or can be induced in the basal part by cutting into seg- 1932, Birdsey 1955). ments (Linsbauer 1934). Tubercles may develop in Typhonium violifolium at the leaf apex, the petiole apex Ertl identified two extremes of venation pattern, and at the apex of the sheath (Sriboonma et al. 1994). Groups I and II, and an intermediate Group III. The continuous intergradation between the extremes is Regeneration of tubers, leaves and roots from leaf used as an argument for a single basic plan for ara- segments is well known in Zamioculcas zamiifolia ceous vasculature. Group I has basically parallel or and Gonatopus boivinii (Engler 1881, Schubert 1913, parallel-pinnate (striate) venation and contains Cutter 1962). Isolated entire leaflets of Zamioculcas Gymnostachys, Calla, some Monsteroideae such as and Gonatopus spontaneously develop a basal Stenospermation and Spathiphyllum, and the tribes of swelling, followed by the formation of roots and up to Aroideae grouped by Engler (1920b) into subfamily 3 buds, over a 6–9 week period for Zamioculcas. Leaf Philodendroideae (Aglaonemateae, Anubiadeae, regeneration in Gonatopus is more rapid. The results Dieffenbachieae, Homalomeneae, Peltandreae, of experimental manipulation of isolated leaflets grown Philodendreae, Schismatoglottideae and Zantedeschieae), in culture show that any part of the compound leaf is together with tribes Cryptocoryneae and Ambrosineae. capable of regeneration, and that new shoots arise in Group II has a reticulate pattern, superficially like association with the cut ends of the largest veins, pref- that of dicotyledons, and includes Anthurium, sub- erentially at the proximal end. When midvein tissue is family Lasioideae, and most other tribes of subfamily present, regeneration at lateral veins is absent, even Aroideae including Pistieae. Group III contains though there is no vascular connection with the mid- Pothos, some Monsteroideae such as Monstera, and vein. When only lateral veins are present, regeneration tribes Caladieae and Colocasieae. Ertl’s work was a proceeds at their proximal ends (Cutter 1962). Plantlets pioneering study and there is a need for a new study may also regenerate from leaf blades in of leaf venation using modern terminology and a Schismatoglottis (Bogner, pers. obs.). broad taxonomic approach. Geniculi (pulvini) Araceae petioles contain a variety of arrangements of vascular bundles but no detailed systematic com- A pulvinus (swelling) or geniculum (joint) occurs parison has been made. Transverse sections of at the distal end of the petiole in nearly all Pothoideae, petioles are shown in works by Engler (1905, 1911, Monsteroideae and certain Aroideae (Anubias, 1912, 1915, 1920a, b), Ertl (1932), Troll (1939) and Bognera, tribes Culcasieae, Zamioculcadeae, rarely in Birdsey (1955). The mechanical support tissue of the Philodendron and Homalomena). Details are given in petiole may be in the form of a continuous periph- the generic and tribal descriptions in this volume. The eral ring as in Asterostigma, Homalomena, Montrichardia, Philodendron, Spathantheum, or sep- arate peripheral strands of collenchyma as in Anchomanes, Dieffenbachia, Dracontium, 18 T H E G E N E R A O F A R A C E A E

mechanical tissue in the geniculum consists of col- The leaf anatomy of Acorus has received much lenchyma in a peripheral ring or collenchymatous attention (Dalitzsch 1886, Solereder & Meyer 1928, Ertl bundle sheaths, whereas elsewhere in the petiole it is 1932, Kaplan 1970, 1973a). Röst (1979b) was particularly formed by sclerenchyma (Dufour 1886, Dalitzsch interested in the biosystematics of Acorus calamus and 1886). The petiole of Gonatopus boivinii has a genicu- A. gramineus and he discriminated between the two lum in the middle which can reorient the blade. Some using a variety of leaf characters, including the organi- reorientation is also possible from a second geniculum zation of chlorenchyma, aerenchyma and sclerenchyma. near the base of the leaf (Troll 1939). Three morphologically distinct varieties of A. calamus Ligules are recognized (Röst 1979b). Each can be distinguished by ploidy level and on the basis of one leaf characteris- Ligules were regarded by Engler as a modified leaf tic, namely the number of air canals in a prescribed area sheath with a free extension at its tip. They occur in of a transverse section (0.62mm2) of the lamina above the Calla, various Schismatoglottideae, Dieffenbachia and sheath (Röst 1979b). The diploid variety has relatively few some species of Philodendron. A membranous “stip- canals (less than 17), the tetraploid has numerous canals ule” has been reported in Pistia (Engler 1920a). (more than 77), and the triploid variety is intermediate (26–51 canals). Using Röst’s criteria it appears that Leaf structure of Acorus (Acoraceae) Kaplan’s (1970) Iowa population is the native American diploid (Acorus calamus var. americanus) and the Much attention has been devoted to the leaf of Wisconsin population is the triploid (A. calamus var. Acorus, especially to its early development. Kaplan calamus). No tetraploids were studied by Kaplan (1970). (1970) presented several conclusions based on a rig- The triploid is sterile, mainly Eurasian, and probably was orous comparative developmental study of Acorus. He introduced to North America by European settlers. concluded that its unifacial leaves are distinct from those of the Araceae because of the intense adaxial Special Topics meristematic activity that begins during apical growth. Adaxial growth leads to a “bulge” that arches over the Trichosclereids shoot apex. The leaf of Acorus is thus a foliar structure and not a flattened petiole. These are filamentous, branched sclerenchyma cells (sclereids) up to 7 mm long which develop in the Earlier workers mistakenly interpreted young leaves intercellular spaces of vegetative and floral tissues of all as having two successive apices, one that led to a genera of subfamily Monsteroideae except Anadendrum hyponastic arching primordium and a second that over- and Heteropsis, and two genera of subfamily Pothoideae, topped the original, a pattern referred to as sympodial Pothos (Tieghem 1867, Engler 1920, Solereder & Meyer growth. Kaplan (1970) observed that continued adax- 1928, Nicolson 1959) and Pothoidium (French & ial meristematic activity and suppression of the Tomlinson 1981a). Trichosclereids rarely develop in marginal meristems leads to the formation of the radi- roots. They played a significant role in Engler’s classifi- ally flattened leaf. A narrow secondary midrib does, cation of the Araceae, which emphasized vegetative however, develop. Morphologically and developmen- anatomical features. tally the leaf of Acorus is similar to the phyllode of Acacia, which also develops an active adaxial meris- The term trichosclereid was first applied by Bloch tem. The phyllode theory of Arber, which attempted to (1946) and Sinnott & Bloch (1946) in their classic interpret leaves of Acorus as flattened petioles, is thus study of idioblast development in Monstera deliciosa. based on an erroneous interpretation. A variety of other terms had been used previously including “raphide”, which was in common use in the It is interesting that leaves of different populations early 19th century to refer to anything needle-shaped of Acorus calamus studied by Kaplan (1970) differ in (Nicolson 1959, 1960b). The term was confusing their degree of apical growth, which is correlated with because raphides of calcium oxalate, which are crys- radial expansion. The Iowa population (probably a tals, also occur in the same tissue. Schott correctly diploid) exhibited more prolonged apical growth and illustrated trichosclereids in various monsteroid gen- less radial growth than the Wisconsin population era, but referred to them as “raphides” (Schott 1832, (probably a triploid). One would predict that the Schott 1858). Hasskarl (1842) carried this confusion tetraploid should exhibit even shorter apical growth one step further by naming a monsteroid genus and greater radial expansion. Vascular differentiation in Rhaphidophora on the basis of correctly illustrated the leaf of Acorus begins first in the centre of the axis trichosclereids. and later is bidirectional, as is the activity of a plate meristem at the leaf margins. Subsequent growth in Schleiden (1839) correctly identified “raphides” (= length is accomplished predominantly through the trichosclereids) as “Bastzellen”, and more detailed activity of a basal intercalary meristem. accounts by other anatomists followed (Sueur 1866, Tieghem 1867, Wiesner 1875). Tieghem (1867) made V E G E T A T I V E A N A T O M Y 19

a number of observations of trichosclereids as part of Laticifers a general survey of the anatomy of the Araceae, and was the first to notice that they occurred primarily in Early studies (Hanstein 1864, Trécul 1865, 1866) monsteroid taxa, an observation used by Engler in demonstrated the presence of articulated non-anasto- defining subfamily Monsteroideae taxonomically mosing laticifers in most subfamilies, including (Engler 1920b). However, Engler emphasized the Calloideae, Lasioideae, Orontioideae and Aroideae, and presence of trichosclereids to such an extent that he articulated, anastomosing laticifers in tribes Caladieae placed several genera which lack these cells but oth- and Colocasieae (except Ariopsis). Both clear and milky erwise display clear monsteroid affinities in his latex was described by early workers, who detected subfamily Pothoideae. These genera, which included abundant tannins in the latex. The clear, tannin-rich Anadendrum, Epipremnopsis (= Amydrium) and latex of some Araceae has been compared with the Heteropsis, are today widely recognized as belonging exudate produced by articulated cells in the phloem of to subfamily Monsteroideae (Nicolson 1984a, Bogner some Leguminosae (Trécul 1865, 1866, Esau 1974), & Nicolson 1991, this volume), and trichosclereids which may also have perforated end walls. are now known to occur in Amydrium. All early workers considered Araceae to have lati- As a result of various studies, a rather general cifers that produce latex. However, Solereder & understanding of the morphology, occurrence and Meyer (1928) took a more neutral position, refraining development of trichosclereids has emerged (Nicolson from the use of the term laticifers. They used the 1960b). In the Araceae, trichosclereids have several terms “secretory files” or “secretory tubes” to describe distinctive features although there is considerable the non-anastomosing and anastomosing laticifers, quantitative variation among species. They begin respectively. development as small, nearly isodiametric cells which form 1–4 hair-like outgrowths that elongate into the Engler (1920b) emphasized the presence of anas- surrounding tissue, forming an irregular network tomosing laticifers in defining his subfamily within the intercellular spaces. Their branches grow Colocasioideae (equivalent to our tribes Caladieae and apically, are often of irregular shape and may them- Colocasieae combined). However, Weiss (1866) found selves branch. that non-anastomosing laticifers are present in roots of Syngonium; he made no further observations on other Trichosclereids that have two branches have a peg- genera of Engler’s subfamily Colocasioideae. Engler like central body which identifies the location of the (1920b) also defined his subfamilies Pothoideae and original cell and is the site of attachment in the tissue. Monsteroideae in part because of their lack of laticifers. In contrast, other types of sclereid, such as libriform Individual “secretory” cells, not arranged in files, are fibres, may be long and filamentous but do not have present in these taxa, but their chemical and structural branches arising from a cell body. Trichosclereids may similarity to laticifers remains unstudied. also have four outgrowths resulting in an H-shape. The number, diameter and length of the branches are Laticifers in Araceae typically occur in association the primary sources of variation. In Spathiphyllum tri- with vascular bundles, lying on the periphery of the chosclereids appear to be longer and narrower than phloem in the leaf and stem. In stems, laticifers occur those from Rhaphidophora and other Monstereae only in leaf traces or cortical bundles, not in axial (Tieghem 1867, Nicolson 1960b). It has long been rec- bundles (French & Tomlinson 1981c–d, 1983). In ognized that in tribe Spathiphylleae there are more roots, laticifers also occur in association with phloem, numerous trichosclereids in each intercellular space but may be present in the xylem, pericycle or ground than in tribe Monstereae. Estimates of over 50 tissue (Weiss 1866, Lierau 1888). They are visible in (Solereder & Meyer 1928) and between 15–20 sections because of dark-staining contents and a rel- (Nicolson 1960b) have been made. This contrasts atively wide lumen compared to adjacent phloem greatly with the 1–4 trichosclereids in each intercellu- cells. Laticifers may occur in the root cortex in lar space found in tribe Monstereae (Nicolson 1959). In Philodendron and various Caladieae and Colocasieae addition, some trichosclereids intergrade with astroscle- (Weiss 1866, Porsch 1911). reids in foliage leaves, e.g. in Monstera. The most complete study of the systematic occur- The distribution of trichosclereids in the plant body rence of laticifers is by French (1988 and unpublished follows a general pattern throughout subfamily observations), who examined leaves and inflorescences Monsteroideae (sensu Engler 1920b). They are usually of 75 genera. Anastomosing laticifers are limited to the present in the stem, petiole, leaf blade and inflores- tribes Caladieae, Colocasieae and Zomicarpeae. cence, but have been reported in the roots in only Articulated, non-anastomosing laticifers occur in Calla, three species (Solereder & Meyer 1928). It is not Orontium and almost all Aroideae (except tribes uncommon for trichosclereids to be absent from one Pistieae, Stylochaetoneae, Zamioculcadeae). Tribe part of the plant body (Nicolson 1959, 1960b). In some Cryptocoryneae have been previously reported as lack- Epipremnum species and Monstera punctulata, tri- ing laticifers, but recent unpublished work has chosclereids are absent from the leaf blade (Nicolson demonstrated their presence in the stems, roots and cat- 1959, Madison 1977a). aphylls of both Cryptocoryne and Lagenandra, though not in the foliage leaves (Sivadasan, pers. comm.). 20 T H E G E N E R A O F A R A C E A E

Limited cytological observations of laticifers were and Homalomena species. Additional reports of resin made by Molisch (1899, 1901) and Southorn (1964). canals and cavities in these two genera were made by No clear picture of latex cytology has yet emerged. Tieghem (1867, 1872, 1885), Möbius (1885), Dalitzsch Schmid (1882) observed nuclei, small dense spheres (1886), Leblois (1887), Lierau (1888), Went (1895), and larger unidentified elliptical particles approxi- Porsch (1911), Engler (1920b) and Pohl (1932a, b). mately the size of a nucleus in laticifers of Caladium. The most complete study is a family-wide survey by Molisch (1899, 1901) made the most extensive French (1987b), which included 91 genera and 250 cytological study, using fresh latex. He observed so- species and showed that resin canals occur in the roots called “Blasenkerne”, which represent nuclei having of only five genera, Culcasia, Cercestis, Homalomena, sac-like protrusions, in several genera, including Furtadoa and Philodendron. In the leaf of Culcasia the Aglaonema, Philodendron, and Xanthosoma. Molisch canals are easily visible with the unaided eye and also found what he called abundant leucoplasts in the exhibit a variety of differences in their length and ori- latex of Steudnera colocasiifolia, as well as biconvex entation depending on the species (Knecht 1983). They particles (7–16 µm in diameter) of unknown identity. are also visible without magnification in leaves of In other studies Molisch (1901) noted that in latex of Philodendron. Recent studies have also confirmed their Amorphophallus konjac (syn. A. rivieri), large (16 µm) presence in the stems of Culcasia, Homalomena (cav- membrane-bound six-sided crystalloids are found. In ities), Philodendron, and Cercestis (French & the only recent study, Southorn (1964) reports the Tomlinson 1981a, c–d, 1983). presence of vacuoles containing particles in Brownian motion in Dieffenbachia. Fox & French (1988) found Pohl (1932a, b) and Mayo (1986b, 1991) investigated that abundant terpenoid particles were responsible the resin canals that are present in the inflorescence of for the white appearance of the latex of tribe Philodendron. The resin is released onto the spathe or Caladieae, analogous to the white latex of spadix surface and plays a role in the floral biology in Euphorbiaceae. The clear or cloudy latex of tribe Philodendron by glueing the pollen onto the pollinator’s Colocasieae lacks the abundant terpenoid particles. body (Gottsberger & Amaral 1984, Grayum 1990, Mayo Larger latex particles were also found in the latex of 1991). In species of Philodendron subgen. Meconostigma “colocasioids” (i.e. tribes Caladieae and Colocasieae (e.g. P. bipinnatifidum studied by Pohl 1931a, b as P. sel- combined), as described by Molisch, and their occur- loum), one end of the canal is close to the inner surface rence was found to be systematically significant of the spathe and releases resin shortly before spathe clo- (French & Fox, unpublished results). sure. In Philodendron subgen. Philodendron resin is secreted onto the spadix surface from numerous resin Resin canals and cavities canals located just below the axial epidermis, around the bases of the stamens and staminodes. In these The presence of resin canals in Araceae has been species, the pollen and resin are mixed even before documented from the mid-nineteenth century (Trécul appearing at the spadix surface (Mayo 1986b). 1865, 1866, Tieghem 1867). However, Solereder & Meyer (1928) proposed the neutral term “secretion Mucilage cells, canals, and cavities spaces” to refer to these canals in Araceae because in their opinion there was a lack of chemical evidence for Mucilage contains various hydrophilic substances the presence of resin. Resin typically contains large and can readily be distinguished from resin using amounts of various terpenoids and other lipophilic appropriate histochemical procedures. It may occur in substances and is thus chemically distinct from various idioblasts such as raphide tubes (Solereder & mucilage, which is hydrophilic (Fahn 1979). It is syn- Meyer 1928), together with calcium oxalate crystals, thesized and accumulated by canals and cavities which or in large mucilage cells, as in Amorphophallus are of limited occurrence in Araceae (Trécul 1865, (Abranowicz 1912, Wakabayashi 1957a, b). Mucilage 1866, Tieghem 1867, 1872, 1885). Resin is generally also occurs in canals and cavities which may form considered to be part of the plant’s chemical defence schizogenously, as in Epipremnum pinnatum (Leblois against herbivores. 1887) or lysigenously as in Colocasia (Solereder & Meyer 1928). Mucilage canals and cavities may have a Resin canals are composed of a central cavity sur- distinct epithelium of smaller, more densely stained rounded by 1–3 layers of synthetically active epithelial cells, and some epithelial cells may project into the parenchyma. A sclerotic or collenchymatic protective locule or grow into it like a tylose (Engler 1920b). sheath may surround the epithelium (Möbius 1885). Secretory cavities differ from canals principally in their Mucilage canals are of relatively limited occur- shape which ranges from spherical to irregular. Both rence in Araceae. They occur in Aglaonema, canals and cavities are reportedly schizogenous in ori- Alloschemone, Alocasia, Anthurium, Cercestis, gin (Leblois 1887, Engler 1920b). Resin canals were Colocasia, Epipremnum, Monstera, Philodendron, first reported in Araceae by Trécul (1865, 1866) who Remusatia, Rhaphidophora, and Xanthosoma found them in leaves, stems and roots of Philodendron (Solereder & Meyer 1928, French & Tomlinson 1981a–d, 1983, 1984, Boyce, pers. obs.). V E G E T A T I V E A N A T O M Y 21

Extrafloral nectaries and punctations as specialized or conspicuous, compared with dicotyledons. The leaf apex and hydathodes have Zimmermann (1932) exhaustively reviewed the been described in various species of Araceae by occurrence of extrafloral nectaries in angiosperms Müller (1919), who recognized three types of hydath- and listed reports of possible extrafloral nectaries in odes: the Philodendron type, with normal or slightly Arisaema (Knuth 1909), Colocasia (Solereder & larger stomata; the Alocasia type, with large stomata, Meyer 1928) and Philodendron, but nectar secretion and finally the Colocasia type with gigantic macro- is listed as very doubtful. Madison (1979a, b) reported scopic stomata through which guttation occurs. nectar production by a ring of extrafloral nectaries at the base of the leaf lamina in Philodendron myrme- The Philodendron type occurs in all subfamilies; the cophilum (= P. megalophyllum), and sweet secretions Alocasia type is more restricted, and the Colocasia type are produced also by the extrafloral nectaries of pro- was found only in Ariopsis, Colocasia and Steudnera. phylls in many species of Philodendron, and on the More complete lists have been given by Müller (1919) outside of the spathe in P. goeldii (Bogner, pers. and Grayum (1984). In Lasia hydathodes are present on obs.). In a description of the leaf of Araceae, Engler leaf emergences. Further details of hydathode structure (1920b) made no specific mention of extrafloral nec- have been given by Ducharte (1859), Dalitzsch (1886), taries, but “sessile superficial glands” were reported Minden (1899) and Gentner (1905). in Alocasia by Gardiner (1883, 1889) and Belin- Depoux (1978). The anatomy of the “glandular” Intravaginal squamules structures in Culcasia was described by Solereder (1919) who considered them possibly to be extraflo- These are glandular, flattened, non-vascularized, ral nectaries (Solereder & Meyer 1928). multicellular scales of uncertain function, ranging from approximately 1 mm to over 25 mm in length. They Solereder (1919) and others (Tieghem 1867, occur in the leaf axils of Acorus (Acoraceae), Bachmann 1880, Dalitzsch 1886, Gentner 1905, Engler Philodendron, Cryptocoryne and Lagenandra (Irmisch 1920b) have described the mature structure of the 1858, Engler 1912, 1920a, b, Lawalrée 1945, Ritterbusch unusual “punctations” on the leaves of various species 1971, Blanc 1978). Velenovsky (1907) reported squa- of Anthurium. They are often referred to as glands or mules in many Araceae, which must be an error. glandular punctations and Dalitzsch (1886) reported Similar structures also occur in Spirodela (Lemnaceae) seeing secretions released internally, but this has not and in all Alismatiflorae except Scheuchzeria, where been confirmed. Each mature punctation consists of they are replaced by hairs (Tomlinson 1982). In section, three regions: 1) a basal region of 1–3 curved layers of squamules of Araceae and Alismatiflorae often have a prismatic cells in files embedded in mesophyll, 2) a mucilaginous and glandular appearance (Kaplan 1973a central region of large, radially elongated cells lacking Tomlinson 1982). Kaplan (1973a) reported that in dense contents, and 3) an outer cluster of compact cells Acorus, two squamules arise from the protoderm of the with dark brown or black contents beneath a stomate. axillary meristem and expand rapidly before the pro- No secretion has been observed to exude from the sur- phyll is initiated. face punctations and the chemical identity of the dark contents is unknown. Solereder (1919) pointed out the Mineral crystals and crystal idioblasts similarity of these structures to the cork warts of some dicotyledons, also mentioned by Bachmann (1880). Hydathodes A wide range of calcium oxalate crystals and crystal- containing idioblasts is present in Araceae, although few Hydathodes occur in the leaf tips of many Araceae, detailed studies have been made of their development, and have attracted considerable attention because of distribution or biological significance (see Seubert 1993 the high rate of guttation in some species. Guttation for a recent study in Araceae). Mineral crystals have been has been studied frequently in Colocasia, in which viewed both as waste products and as agents of plant rates of over 100 ml/day/leaf are not unusual (Molisch protection against herbivores (Esau 1965, Madison 1903). The hydathodes of Araceae are often localized 1979a). Calcium oxalate crystals occur most commonly in the leaf tip, the “Vorläuferspitze” (precursor tip), as raphides and druses, less commonly as crystal sand, which contains numerous stomata on both surfaces “ styloids” or prisms Histochemical tests by Sunell & (Müller 1919); hydathodes may also be found else- Healey (1981) confirmed that the raphides of Colocasia where on leaves. The stomata may be relatively esculenta contain calcium oxalate. X-ray diffraction stud- normal in size and appearance, or may be larger than ies of raphides of Monstera deliciosa (Al-Rais, Myers & surrounding stomata. The hydathode also contains Watson 1971), Xanthosoma sagittifolium (Cody & Horner underlying ground and vascular tissue with numer- 1983) and Colocasia esculenta (Sunell & Healey 1981) ous tracheids (epithem). However, Gardiner (1883) demonstrated monoclinic calcium oxalate monohydrate. did not regard the epithem of Calla and Zantedeschia Comparable studies have not been made on the other types of presumed calcium oxalate crystals in Araceae. 22 T H E G E N E R A O F A R A C E A E

No silica or calcium carbonate crystal inclusions have Alternatively both ends may project into separate spaces been demonstrated. Unidentified inclusions (probably with the middle of the cell loosely held by mesophyll not calcium oxalate) are present in the parenchyma of cells or vascular tissue (Sunell & Healey 1981). tubers of Amorphophallus konjac (syn. A. rivieri) (Abranowicz 1912). The presence of mucilage in biforines has been recognized since Turpin’s study (1836) and its swelling The structure of raphides in Araceae is apparently is generally considered to be the motive force for the unusual in angiosperms (Cody & Horner 1983) because expulsion of the contents of the biforine (Sakai & they consist of twinned crystals, H-shaped in trans- Hanson 1974). In some species raphides are expelled verse section and are often barbed laterally (Sakai & individually (Colocasia) by piercing and subsequent Hanson 1974). Similar shaped raphides occur in some rupturing of the thin papillae. In others expulsion Lemnaceae. The systematic significance of these find- occurs after the papillae are ruptured by other means ings is unclear because there have been very few (Sakai & Hanson 1974), as in Dieffenbachia. By con- general surveys of raphide structure in angiosperms trast, in Alocasia the cell wall of the biforine breaks at using the scanning electron microscope, which is nec- the base of the cell and the entire mass of crystals and essary for more rigorous comparative studies. mucilage is ejected, followed by swelling and individ- ual dispersal of the raphide crystals. Raphide-containing idioblasts are variable in length, wall specialization and manner of raphide release. There The occurrence of biforines in Araceae has been are two general types in Araceae: 1) thin-walled raphide summarized by Solereder & Meyer (1928), by Nicolson idioblasts, and 2) thick-walled idioblasts called biforines. (1959) who added many new observations, and by Thin-walled raphide idioblasts range in shape from iso- Grayum (1984). No detailed family-wide survey has yet diametric to very elongated (3–5 mm long in Monstera been made, but existing data (De Bary 1884, Dalitzch deliciosa, Kovacs & Rakovan 1975). The elongated forms 1886, Solereder & Meyer 1928, Grayum 1984) suggest have been referred to as “Raphidenschläuche” (Solereder that biforines are characteristic of genera with unisexual & Meyer 1928) and may contain many bundles of flowers and generally absent in bisexually-flowered gen- raphides oriented at various angles and embedded in era. Among the latter, biforines have so far been observed mucilage. They are of common occurrence in Araceae only in Anthurium, Orontium and Symplocarpus. and have been reported by many authors (e.g. Dalitzsch 1886, Porsch 1911, Solereder & Meyer 1928, Hinchee Calcium oxalate is also the major component of 1981). The elongated type may intergrade with cells that druses, which have a wide systematic occurrence in the are much shorter at maturity and tend to grow rapidly family. Druses vary considerably in size, sometimes by apical intrusive growth, allowing them to exceed the reaching 90 µm in Zamioculcas (Solereder & Meyer length of surrounding cells. There are reports of raphide 1928). They also vary in morphology and include atyp- idioblasts anastomosing during development in ical forms in leaves of Anthurium, Dieffenbachia, Pistia Anthurium scandens, A. scherzerianum and A. triphyl- and other genera (Solereder & Meyer 1928) and may lum (Samuels 1923) and in other genera, such as occur in small rounded epidermal cells in Anthurium, Monstera and Dieffenbachia (Solereder & Meyer 1928). Culcasia, Pothos, Schismatoglottis and many other gen- Samuels challenged the idea that the long tubes in many era. Prisms are of limited occurrence in Araceae, and Araceae are single cells, proposing instead that they have been reported only in subfamilies Pothoideae and form by the anastomosis of cells in files. No subsequent Monsteroideae (Solereder & Meyer 1928, Nicolson authors have addressed this issue and Samuels’ inter- 1959, Seubert 1993). esting findings are unconfirmed. Raphides are reportedly absent from Acorus (Acoraceae). Developmental studies of raphide crystals and crys- tal-containing idioblasts have been made by Samuels The second type of raphide idioblast is the biforine, (1923), Becker & Ziegenspeck (1931), Rambour (1965), a term coined by Turpin (1836) to refer to spindle- Mollenhauer & Larson (1966), Rakovan, Kovacs & shaped or cylindrical cells with thick, unlignified walls Szujko-Lacza (1973), Kovacs & Rakovan (1975), and (Middendorf 1983). No specific function for these cells Hinchee (1981). According to Kovacs & Rakovan has been demonstrated, but some authors have (1975), raphide idioblasts in Monstera arise by unequal described a “blowgun” release of raphides (Middendorf cell division. The smaller cell develops into the raphide 1983). One or both ends of the cell may be blunt, idioblast which has densely staining cytoplasm and a pointed or papillose, and with much thinner walls. The larger nucleus than the surrounding cells. Hinchee thinnest part of the end wall in Colocasia esculenta is 0.1 (1981), however, did not observe asymmetric divisions µm thick (Sakai & Hanson 1974). The variable orienta- giving rise to raphide idioblasts in Monstera. Kovacs & tion of fibrillar material causes the cell wall to appear Rakovan (1975) noted that raphides develop as soon as layered, except in Alocasia (Sakai & Hanson 1974). The the young idioblast cells are the same size as sur- dimensions of some biforine cells are 50 x 150 µm in rounding cells, before the phase of rapid elongation. Colocasia esculenta (Sakai & Hanson 1974) and 40 x 90 µm in Alocasia. Biforines like those of Colocasia escu- Seubert (1993) has recently carried out a survey of lenta may have one end embedded in tissue and the rest crystal types occurring in the seeds of Araceae as part of the cell projecting into an intercellular space. of a comprehensive study of Araceae seed anatomy. Her illustrations reveal a fascinating diversity of novel crystal forms. V E G E T A T I V E A N A T O M Y 23

Occurrence of vessels Parthasarathy (1980). Behnke’s work on sieve element C plastids is of considerable interest. In an extensive Vessels with scalariform or reticulate perforation survey of the Araceae, Behnke (1995) confirmed the plates have long been recognized in the roots of some widespread occurrence of the typically monocot plas- Araceae, such as Monstera deliciosa, and tid type P2 in the family. The Araceae have a relatively Philodendron and Alocasia species (Solereder & rich array of sieve element plastid ultrastructure, with Meyer 1928, Kundu 1942), while other genera like type P2c found in 14 species and P2cfs in one species. Caladium and Pistia have been considered vessel- A P2 subtype with both cuneate protein and starch less (Solereder & Meyer 1928). Solereder & Meyer grains was found in 110 of the 126 species surveyed. were apparently the first to identify vessels in the Unexpectedly Pistia was found to have S type plastids stems of Araceae, in Pothos scandens and P. rumphii. thus completely different from all other monocotyle- In Culcasia scandens they found no vessels but tra- dons. The S type has heretofore been found only in cheids with wide lumens. Later Cheadle (1942) dicotyledons. Conversely only two dicotyledons have examined 7 genera and 8 species and found vessels in been found with P2c plastids, Asarum and Saruma. the roots only, with scalariform perforation plates in all The phylogenetic significance of these findings is dis- species. Subsequently Hotta (1971) reported vessels in cussed at length by Behnke (1995). stems of a few species, but in the roots of nearly all those he examined except Arisaema, Homalomena, Starch grains Pothos and Rhaphidophora. Hotta’s sample was not very broad taxonomically, but represents the most Starch grain morphology in Araceae has been extensive to date. Vessels occur in stems of several examined by Reichert (1913), Czaja (1969, 1978a, b) species of Pothos, Epipremnum, Rhaphidophora and and Seubert (1993). Reichert reported that starch grains Scindapsus. Hotta (1971) emphasized the difficulty in Arum, Arisaema, Dracunculus and Zantedeschia encountered in recognizing vessel members in all belonged to the same type, while in Dieffenbachia Araceae, principally because of the similarity between a remarkably different kind is present. Despite gross scalariform pitting and scalariform perforations on the structural differences all five genera have starch with long, oblique end walls of tracheids. He also referred similar physical and chemical properties. Reichert to some tracheids as “vesselform” tracheids because of (1913) also studied starch grains in Alocasia, their similarity to vessels. Amorphophallus, Peltandra and Colocasia. Those of Peltandra were found to resemble the Dieffenbachia Phloem cytology type, while grains in Amorphophallus and Colocasia were of the Arum type. Grains of Alocasia were “pecu- The phloem of Araceae has been examined using liar” and not assigned to either type. More recently, light microscopy in studies by Shah & James (1971) as Seubert (1993) has surveyed starch grain morphology well as in investigations by Lesage (1891). Recent in Araceae seeds. Using Czaja’s terminology (Czaja ultrastructural studies have been made by Behnke 1969), she recognized 11 different types in a survey of (1969, 1981), Bonzi & Fabbri (1975, 1978) and over 70 genera. 24 T H E G E N E R A O F A R A C E A E

C 4 INFLORESCENCE AND FLORAL MORPHOLOGY The inflorescence of the Araceae is composed of an short stipe, modified attractive spathe and continuation unbranched spike bearing flowers, the spadix, sub- shoot development at the second node below the tended by a bract called the spathe. The flowers are spathe. Gymnostachys and subfamilies Orontioideae usually numerous, very small, sessile in all genera and Pothoideae seem to represent earlier, less uniform except Pedicellarum, and lack floral bracts. They are phases of organization. The typical araceous pattern generally spirally arranged and usually tightly packed, has given rise to a wide range of variant forms in dif- although in some species of Pothos ser. Goniuri, ferent genera, which can be seen to represent an Pedicellarum, Amorphophallus (male and female flow- evolutionary trend of increasing integration towards a ers), tribe Dieffenbachieae (female flowers) and most synflorescence or pseudanth. The major phyletic modi- species of Arisaema and Arisarum (male flowers), they fications are: 1) loss of perigone in the flowers; 2) may be somewhat distant from one another. specialization of flowers on the spadix into a lower female zone, upper male zone and, often, one or sev- The spathe is, strictly speaking, the last leaf of a eral zones of sterile flowers, entirely naked axial zones flowering article. It is usually a specialized attractive and smooth or staminodial terminal appendices; 3) dif- organ although in a few genera (Gymnostachys, ferentiation of the spathe into a lower, convolute tube Orontium) is inconspicuous. The internode between and an upper, expanded blade. spathe and spadix (spadix stipe) is usually very short or absent, while the peduncle – the internode between Spathe and spadix modifications are closely related spathe and last foliage leaf or cataphyll – is much so that the spathe may be seen evolutionarily as longer. In some primitive genera, however, this arrange- becoming increasingly integrated into the inflores- ment is reversed (Gymnostachys, subfamily cence itself, until in extreme cases, such as tribe Orontioideae, some Pothos species). It seems likely that Cryptocoryneae, Ambrosina, Pistia and Pinellia, fusion an important stage in aroid evolution involved the com- and still more elaborate modifications have brought bined development of a relatively long peduncle and about division of the spathe into separate chambers. Spadix Spathe blade Male zone Flowers Sterile zone Stipe Female zone Stipe Spathe Spathe constriction Peduncle Spathe tube Peduncle Figure 7. Inflorescence types: A, bisexual flowered spadix with a simple, undifferentiated spathe; B, unisexual flowered spadix with a spathe divided into a limb (blade) and convolute lower tube. I N F L O R E S C E N C E A N D F L O R A L M O R P H O L O G Y 25

Other notable specializations of the inflorescence tribes of subfamily Aroideae mentioned above. Anthers C include the wide range of odours found in different are almost always extrorse (introrse in Zamioculcas, genera, colour patterns, especially on the spathe, and latrorse in Pedicellarum). Theca dehiscence may be by the relative persistence of different regions of the a longitudinal or rarely transversal slit (most genera spathe. In Philodendron, for example, the entire with bisexual flowers and some unisexual-flowered spathe persists until fruit, but in tribes Colocasieae, genera: Anubias, some Areae, Arisaema, Arisarum, Caladieae, Peltandreae and Schismatoglottideae the Stylochaeton) or by apical or subapical pores or short spathe blade withers or drops off immediately fol- slits. In many genera of subfamily Aroideae dehiscence lowing anthesis while the spathe tube persists. In of each theca is by a subapical stomial pore and this many Monstereae the entire spathe withers or drops morphology is frequently correlated with the extrusion off soon after flowering, a behaviour which is corre- of pollen in strands. Similar structures occur in lated in this tribe with the presence of numerous Amorphophallus and Dracunculus. protective trichosclereids in the style tissue. The gynoecium usually varies between 1- and 3- Terminal appendices of the spadix are found in tribes locular, and when unilocular often shows traces of 2- Areae, Arisaemateae, Colocasieae, Schismatoglottideae, or 3-merous origin through the presence of a several- Thomsonieae and Zomicarpeae, sporadically elsewhere lobed stigma (e.g. Typhonodorum) or more than one in the family. The function of the appendix, where placenta (e.g. Schismatoglottis). Gynoecia with more investigated, is to produce odours to attract pollina- than 3 locules are less common but are found in tribe tors (osmophore, Vogel 1963, 1990). The appendix is Spathicarpeae (1–8 locular) and in Philodendron (2–47 either clearly composed of staminodes (e.g. locular). Placentation varies from axile to parietal, Pseudodracontium) or is partially to entirely smooth basal, apical or basal and apical (the latter in with no vestiges of floral organs (e.g. Arum). Dracunculus, Helicodiceros, Heteroaridarum and Theriophonum), with many intermediates. Ovules may Flowers in Araceae may be 2- or 3-merous. In be anatropous, campylotropous, orthotropous or inter- perigoniate flowers the tepals, when free, are orga- mediate between these types. Funicle trichomes are nized in two whorls. The tepals are usually usually present (French 1987c) and secrete a clear, more-or-less fleshy and fornicate apically (except sub- mucilaginous substance which in many genera (e.g. family Pothoideae) and in some genera or sections tribe Monstereae, Philodendron) entirely fills the ovary (Anadendrum, Holochlamys, Pedicellarum, locules; this secretion appears to play a role in pollen Spathiphyllum sect. Massowia, Stylochaeton) they are tube growth (Buzgó 1994). The style may be narrowed fused into a cup-like structure. Stamens in perigoniate and elongated (e.g. Dracontium) but in most genera flowers and in the naked bisexual flowers of most is relatively inconspicuous externally. However, there Monsteroideae have essentially the orthodox structure is very often a thick stylar region between the ovary of distinct (usually flattened) filament, basifixed anther locules and stigma (e.g. Philodendron, Mayo 1989b). and slender, inconspicuous connective. In the uni- In tribe Monstereae this stylar region is especially well sexual flowers of many tribes of subfamily Aroideae, developed and densely filled with trichosclereids. Here however, filaments are typically very short or lacking, the style seems to substitute functionally for a perianth and there is a thick, fleshy connective which probably in protecting the sexual organs of the flower. Stigmas acts as an osmophore (Aglaonemateae, Culcasieae, are always wet in Araceae and in some genera Homalomeneae, Montrichardieae, Nephthytideae, (Anthurium, Arum, several Lasioideae) produce con- Philodendreae, Zantedeschieae). Stamens of tribes spicuous nectar droplets at anthesis. In Anubiadeae, Caladieae, Colocasieae, Dieffenbachieae, Amorphophallus, Dieffenbachia and some and Peltandreae are essentially similar but are always Spathicarpeae, the lobing of the stigma can be very fused into synandria. In tribe Arophyteae the stamens pronounced, or the stigma relatively massive. In sub- may be fused or not and exhibit a diversity of struc- family Monsteroideae stigmas vary from subcapitate ture. Large connectives also occur in tribe to conspicuously elongated, either transversely (e.g. Spathicarpeae but their different morphology suggests Anadendrum) or longitudinally. that they are not homologous with those of the other 26 T H E G E N E R A O F A R A C E A E

C 5 INFLORESCENCE AND FLORAL ANATOMY Inflorescence and flower anatomy of Araceae emphasis on floral vasculature. One aim of their stud- received little attention after the completion of Engler’s ies has been to establish whether unilocular gynoecia last monograph (Engler 1920b) until very recently. in Araceae are always pseudomonomerous (see also Knoll (1926) made a detailed study of inflorescence Eckardt 1937) or if truly 1-carpellate gynoecia occur; anatomy in Arum with particular reference to struc- thus far no examples of the latter have emerged. tures and adaptations concerned with pollinator French (1985a,b, 1986 a,b, c, 1987c) has made family- behaviour. Pohl (1932a, b) and Mayo (1986b, 1989b) wide surveys of endothecial thickenings, stamen and studied inflorescence and floral anatomy in Philo- ovule vasculature and ovular trichomes. These have dendron and demonstrated the existence of spathe further confirmed the distinctness of Acorus and tissues adapted for opening and closing movements, revealed some taxonomically useful character variation resin secretion of various types from resin canals in the within the family. Carvell (1989a, b) has made a spathe and spadix and a wide variety of gynoecial detailed survey of floral anatomy in the bisexual and and androecial structures. Eyde, Nicolson & Sherwin bisexual-tepalate flowered genera. (1967) made the first general survey with a study of the flowers of 18 genera (including Acorus), which con- Vogel (1963, 1978) made many fascinating obser- centrated more on those with bisexual or vations, including histological studies, of the terminal bisexual-tepalate flowers. They established that in flo- appendix in various genera (Alocasia, Arisaema, Arum ral anatomy, as in other characters (Grayum 1987), and others) and the spathe limb of Cryptocoryne, in Acorus differs markedly from the Araceae and con- relation to their biological function as osmophores. firmed the absence of floral bracts subtending the Eyde et al. (1967), Vogel (1963, 1990), Mayo (1986b, flowers. Barabé and coworkers (Barabé1982, 1987; 1989b) and Chauhan (unpublished results) demon- Barabé & Chrétien 1985, 1986a, b; Barabé, Chrétien & strated the existence of papillate and sculptured cell Forget 1986; Barabé & Forget 1987, 1988a, b, 1993; surfaces in the epidermis of the spathe, androecia and Barabé, Forget & Chrétien 1986, 1987; Barabé & gynoecia of Amorphophallus, Homalomena, and Labrecque 1983, 1984, 1985; Barabé, Labrecque & Philodendron species; this feature is probably very Chrétien 1984) have made a series of detailed floral widespread in the family. Ittenbach (1993) has made a anatomical studies of different genera with particular very interesting study of the anatomy and micromor- phology of some African Amorphophallus species. I N F L O R E S C E N C E A N D F L O R A L A N A T O M Y 27

C 6 FRUITS AND SEEDS The fruits of Araceae are typically juicy berries, hemiepiphytic species to new sites (e.g. neighbouring although rarely drier and leathery. The infructescence trees) by birds or mammals. is usually cylindric or sometimes globose. The berries are most commonly red or orange (see family descrip- The amount of endosperm in the seed varies con- tion) and are almost always free. Exceptions are siderably within the family and has long been regarded Syngonium, in which the berries form an indehiscent as a useful taxonomic character at tribal level. However, syncarp, and Cryptocoryne which has an apically a thorough comparative anatomical study has been dehiscent syncarp. In Lagenandra the berry actively lacking until very recently (Seubert 1993). Among a opens at the base to release the seeds, but aroid berries wealth of other interesting new observations, Seubert’s are otherwise indehiscent. study shows that endosperm may be present in mature seeds of some groups e.g. subfamily Lasioideae, as The various mechanisms observed for protection of only a very thin layer. Many intermediate conditions the developing fruits and seeds have been discussed by exist between the presence of copious endosperm and Madison (1979a). In the Monstereae, which have bisex- absence of endosperm. Absence of endosperm is often ual but non-perigoniate flowers, the thick stylar region correlated with the presence of a well-developed is filled with trichosclereids which protect the devel- plumule and the largest seeds of Araceae are of this oping seeds. At maturity the stylar region is shed to type, e.g. Orontium, Typhonodorum. reveal the seeds. In perigoniate genera like Anthurium the perigone clearly plays a protective role and keeps The seeds are usually straight, but in Lasioideae pace during growth of the developing berry. The lat- and Monstereae they are often curved, sometimes ter only becomes fully exposed at maturity by extrusion strongly so. In a few genera the plumule is also highly from the flower. In Lysichiton, also perigoniate, the developed (Cryptocoryne, Gonatopus, Nephthytis, stylar region and tepal apices protect the young berry, Orontium, Typhonodorum). In Cryptocoryne ciliata, eventually breaking off to reveal the ripe seeds (Hultén which grows in freshwater tidal zones, between 20 & St. John 1931). and 40 cataphylls are formed in the embryo, and these may serve to fix the seed to the substrate, preventing In many unisexual-flowered genera the protective it from being swept away with the ebb and flow of function is assumed by the persistent spathe or spathe tides. Seeds with a well developed plumule usually tube. Spathe growth continues around the developing lack endosperm and have only a very thin, papery fruits until maturity when the spathe may split open testa or none at all (Nephthytis). Such large embryos (Alocasia, Dieffenbachia) or absciss at the base generally contain chlorophyll at maturity and are only (Philodendron), exposing the infructescence of white viable for a short time. or coloured berries. In other monoecious genera, how- ever, the spathe is marcescent and plays no role in The testa may be smooth, rough, verrucose or costate, fruit protection. In such cases (e.g. Arum) protection thin or thick and in subfamily Lasioideae is often very may possibly be through the presence of toxic chem- hard and thick with prominent sculpturing. In tribes ical compounds in the berries. Ambrosineae, Areae, Arisaemateae and Arisareae, most genera have a prominent fleshy strophiole (aril), and in The seeds are often embedded in mucilaginous tribe Colocasieae smaller but distinct strophioles also pulp (secreted by the ovular and placental trichomes). occur. Arillate seeds have been observed in Philodendron In Anthurium the inner layer of the pericarp may subgen. Meconostigma (Mayo 1986b, 1991). In also be mucilaginous and in other genera the outer Zomicarpa the swollen funicle remains connected to the integument becomes mucilaginous. This makes the mature seed (Peyritsch 1879, Bogner, pers. obs.). In Pistia seeds sticky and aids the dispersal of epiphytic and a double operculum is formed by both integuments. 28 T H E G E N E R A O F A R A C E A E

C 7 SEEDLING MORPHOLOGY Four different types of seedling can be recognized in The cotyledonar hyperphyll functions as a storage the Araceae (Tillich 1985, 1995, Seubert 1993). organ and perhaps has a very limited haustorial role when some endosperm is present. The cotyledonar In Type 1, the seeds have copious endosperm, the sheath, hypocotyl and primary root tend to be reduced cotyledonar hyperphyll functions as a haustorium, the or often completely absent. cotyledonar sheath is condensed and short, the hypocotyl and primary root are well developed and In Acorus (Acoraceae) the seedling bears no simi- first leaf is either a cataphyll or a foliage leaf. This type larity to any form found in the Araceae. The cotyledonar is quite common in the family, e.g. Arisaema, Arum, hyperphyll is subdivided into a long, cylindrical, assim- Calla, Gymnostachys, Pinellia, Zantedeschia. ilatory portion and a small haustorial tip. The first plumular leaves are unifacial and ensiform and the Type 2 is similar to Type 1, but the cotyledonar primary root is well developed. Tillich (1985, 1995) sheath is broadened and blade-like, green and assimi- has pointed out the resemblance between the seedling latory, whereas the cotyledonar hyperphyll is a minute of Acorus and those of the Juncaceae, Melanthiaceae, haustorium, e.g. Colocasia, Philodendron, Xanthosoma. and Typhaceae. Type 3, which occurs only in Pistia and is very sim- In Type 2 and in Acorus, the root collar bears con- ilar to the seedling morphology of Lemna (Lemnaceae), spicuous, long and densely disposed rhizoids. In Type differs from Types 1 and 2 in that the primary root and 1 rhizoids are found only sporadically (e.g. in hypocotyl are undeveloped. Cryptocoryne cognata but not in C. ciliata). In Type 4 the seeds have little or no endosperm. S E E D L I N G M O R P H O L O G Y 29

C 8 EMBRYOLOGY The embryology of Araceae has not been studied on crassinucellate. However, in a strict sense most Araceae a broad comparative scale, but excellent reviews have so far investigated have tenuinucellate ovules, except been published by Grayum (1984, 1991b), on which Symplocarpus and Calla. An endothelium, derived from this chapter is based. Thorough treatments of all the inner surface of the inner integument, is reported embryological aspects of a single taxon have been from most genera investigated. The seed of Acorus has published for Peltandra virginica (Goldberg 1941), a perisperm derived from the nucellus in addition to the Theriophonum minutum (Parameswaran 1959) and endosperm, and in this differs from all Araceae. Synandrospadix vermitoxicus (Cocucci 1966); other genera are known less completely. Jüssen (1928) made Linear megaspore tetrads are the commonest type, the most important single contribution to date, but but T-shaped tetrads are also found, sometimes both generalizations about the family’s embryology are still types occurring in the same species. The mature based on rather fragmentary coverage. embryo sac is usually of the 8-nucleate type (10–12 nucleate in Nephthytis). Embryogeny is clearly under- Acorus (Acoraceae) has a secretory anther tapetum, stood in only nine genera. Onagrad and asterad thus differing from the Araceae, which have the embryogeny are known in monoecious genera and periplasmodial type (sensu Clausen 1927). Pollen caryophyllad and solanad types are known in bisexual mother cell division is probably always of the succes- genera. Cellular and free-nuclear types of endosperm sive type in Araceae. development both occur but after a review of the avail- able facts, Grayum (1984, 1991b) concluded that The nucellar epidermis of virtually all Araceae divides endosperm development in Araceae is best interpreted to form a nucellar cap, usually 2–3 cell layers thick (1 in as a form of the helobial type. Pistia), and for this reason Araceae have been treated as 30 T H E G E N E R A O F A R A C E A E

C 9 CYTOLOGY This chapter is based on Petersen’s (1989, 1993) recent Petersen (1989) considered that the basic numbers comprehensive review of aroid cytology. The 2n chro- x=14 or x=12 must have been the starting points for mosome numbers given in Table 1 and in the generic the derivation of all the modern chromosome num- descriptions are also from her results, supplemented by bers in the Araceae. It is possible, however, that these unpublished data from Dr. Marcelo Guerra (pers. basic numbers may represent secondary basic num- comm.). Reliable diploid (2n) numbers and intraspecific bers which arose by chromosome doubling from aneuploid derivatives are given, with dubious num- hypothetical basic numbers x=7 or x=6. Petersen bers placed in brackets. The chromosome numbers in regarded the widely occurring number x=14 as of Araceae vary greatly between genera, from 2n=14 ancient origin. She also considered that the number (Ulearum) to 2n=168 (Arisaema). Within a single genus x=12 might be primitive since the diploid number the diploid number may be highly variable, as in 2n=24 occurs in tribe Potheae. Cryptocoryne (2n=20 to 2n=132), or in Arisaema (2n=20 to 2n=168). Some genera, on the other hand, Reduction in diploid number occurs in various mor- have very stable diploid numbers. Anthurium, the phologically advanced genera. A good example is tribe largest genus of the family, is surprisingly uniform cyto- Areae in which the commonest diploid numbers are logically, with the great majority of species having a 2n=26 and 2n=28 but the genera Biarum and diploid number of 2n=30. Typhonium both include species with diploid num- bers as low as 2n=16. The distribution and variation of chromosome num- bers among the genera suggests that chromosome Aneuploid changes at the diploid level followed by number has increased in some phyletic lines to a high polyploidy or aneuploidy at the polyploid level have level of polyploidy and in others has been greatly taken place in some taxa, e.g. Cryptocoryne ciliata reduced. The most advanced tribes, such as Areae, (2n=22, 33), Cryptocoryne cordata (2n=34, 68, 85, 102). Arisaemateae and Cryptocoryneae, tend to have the highest numbers. In the more primitive genera chro- The size and shape of the chromosomes are also mosome number tends to be both less variable and less quite variable. Chromosome length varies from 1–17 extreme, being neither very high nor very low. µm, depending on the genus. The chromosomes have Examples of this type are Pothos (2n=24, 36) and been observed to differ greatly in size and shape within Spathiphyllum (2n=30, 60). a single genome in certain cases, e.g. the New and Old World species of Homalomena. Homalomena Petersen (1989) considered that no modern Araceae speariae (New World) has 42 chromosomes, of which species constitutes a true primary diploid. In the mor- 40 are small (1–2 µm) and one pair is nearly three phologically most primitive genera the lowest diploid times longer (ca. 5 µm). In Old World species there is number is 2n=24. In contrast, the highly derived no large pair of chromosomes. Chauhan and Brandham Ulearum sagittatum has the lowest diploid number (1985), in a study of Amorphophallus cytology, showed known in the family (2n=14), evidently the result of some appreciable size differences within the karyo- phyletic reduction. types of certain species. A primary basic number of x=7 has been proposed Ramalho (1994) has made a recent study of Araceae by Jones (1957), Larsen (1969) and Marchant (1973). chromosomes in Pernambuco, Brazil. C Y T O L O G Y 31

Table 1. List of diploid (2n) and assumed basic (x) chromosome numbers in the genera of Araceae. Chromosome numbers arranged in euploid series are separated by a semicolon; dubious numbers are in brackets (data from G. Petersen, pers. comm., updated from Petersen 1989 and Guerra, pers. comm.). 2n x Family Acoraceae 22, 44; 24, 36, 48 11, 12 1. Acorus Family Araceae I. Subfamily Gymnostachydoideae 48 12 1. Gymnostachys II. Subfamily Orontioideae 26 (24, 28) 13 2. Orontium 28 14 3. Lysichiton 30, 60 (28) 15 4. Symplocarpus III. Subfamily Pothoideae 24, 36 12 Tribe Potheae no data 12 5. Pothos 24 10, 12, 14, 15 6. Pedicellarum 7. Pothoidium 20, 40; 24, 48, 84; 28, 56; 30, 60, 90 Tribe Anthurieae 8. Anthurium IV. Subfamily Monsteroideae 30, 60 15 Tribe Spathiphylleae 60 15 9. Spathiphyllum 10. Holochlamys 60 15 Tribe Anadendreae 11. Anadendrum 28 14 Tribe Heteropsideae 12. Heteropsis 60 15 Tribe Monstereae 60, 120 (42, 54, 56) 15 13. Amydrium 60 (56, 84) 15 14. Rhaphidophora 60 (42, 56, 58, 64, 70, 112) 15 15. Epipremnum 60 (24, 48, 56, 58, 70) 15 16. Scindapsus 84 14 17. Monstera 28, 56 14 18. Alloschemone 28 14 19. Rhodospatha 20. Stenospermation V. Subfamily Lasioideae 26 13 21. Dracontium 26 13 22. Dracontioides 26 13 23. Anaphyllopsis 26 13 24. Pycnospatha 26 13 25. Anaphyllum 26 13 26. Cyrtosperma 26 13 27. Lasimorpha 26 13 28. Podolasia 26 13 29. Lasia 52 13 30. Urospatha 32 T H E G E N E R A O F A R A C E A E

VI. Subfamily Calloideae 36, 54, 72 18 31. Calla 34 17 VII. Subfamily Aroideae 34, 68 17 Tribe Zamioculcadeae 32. Zamioculcas 28, 56 14 33. Gonatopus Tribe Stylochaetoneae 34, 68 17 34. Stylochaeton 34 17 Tribe Dieffenbachieae 35. Dieffenbachia no data 17 36. Bognera 34 17 Tribe Spathicarpeae 34 17 37. Mangonia 34 17 38. Taccarum 34 39. Asterostigma no data 17 40. Gorgonidium 34 17 41. Synandrospadix 34 42. Gearum 15, 16, 17, 18 43. Spathantheum 28; 30; 32; 34; 36; 48 (26) 44. Spathicarpa 20 Tribe Philodendreae 40 19, 20, 21 45. Philodendron 38; 40, 80; 42 Tribe Homalomeneae 24 46. Furtadoa 48, 72 47. Homalomena 13 Tribe Anubiadeae 26, 39, 52 13 48. Anubias 26 13 Tribe Schismatoglottideae 26 c. 13 49. Schismatoglottis c. 26 13 50. Piptospatha 26 12 51. Hottarum 24 52. Bucephalandra no data 18 53. Phymatarum 10, 11, 14, 15, 17, 18 54. Aridarum 36, 72 55. Heteroaridarum 20; 22, 33, 66, 88, 132; 28, 42; 30; 34, 68, 10 Tribe Cryptocoryneae 85, 102; 36, 54, 72, 90 13 56. Lagenandra 7 57. Cryptocoryne 20 7 26 Tribe Zomicarpeae 14 14 58. Zomicarpa 28 13, 14, 15, 16 59. Zomicarpella 11 60. Ulearum 28 11, 13 61. Filarum 22; 26; 28; 30; 32 13 22 14 Tribe Caladieae 22; 26, 39, 52 13, 14 62. Scaphispatha 26 63. Caladium 28 (24, 26) 18, 20 64. Jasarum 26; 28 20 65. Xanthosoma c. 20 66. Chlorospatha 36; 40, 60 67. Syngonium 40 68. Hapaline c. 40 Tribe Nephthytideae 69. Nephthytis 70. Anchomanes 71. Pseudohydrosme C Y T O L O G Y 33

Tribe Aglaonemateae 40, 60, 80, 100, 120 (70, 110) 20 C 72. Aglaonema 40 20 73. Aglaodorum 42, 84 21 Tribe Culcasieae 42 21 74. Culcasia 75. Cercestis 48 24 Tribe Montrichardieae 32 16 76. Montrichardia 36 18 Tribe Zantedeschieae 77. Zantedeschia 26, 39; 28 13, 14 26 13 Tribe Callopsideae 78. Callopsis 38, 76; 54 (40) 19, 27 54, 108 27 Tribe Thomsonieae 54 27 79. Amorphophallus 80. Pseudodracontium 56, 112 14 112 14 Tribe Arophyteae 81. Arophyton 28, 42, 56 14 82. Carlephyton 83. Colletogyne 22 11 Tribe Peltandreae 28, 42, 56, 70, 84 14 84. Peltandra 24; 28 14 85. Typhonodorum 28 14 56 14 Tribe Arisareae 16 (14, 18) 8 86. Arisarum 16; 18, 36, 54; 20; 26, 52, 65; >100 (14) 8, 9, 10, 13 26, 52, 104 13 Tribe Ambrosineae c. 78 c. 13 87. Ambrosina 16; 20; 22; 24; 26; 32; 36; 74; 96; 98* ? Tribe Areae 26, 52 13 88. Arum 20; 22, 24, 48, 72; 26, 39, 52; 28, 42, 56, 89. Eminium 70, 112, 140, 168 (64) 10, 11, 12, 13, 14 90. Dracunculus 91. Helicodiceros 28, 84 (80, 86) 14 92. Theriophonum 28 14 93. Typhonium 28, 42, 56 14 94. Sauromatum 28, 42, 56 14 95. Lazarum 28, 42, 56 14 96. Biarum 28, 42, 56, 70, 84 14 Tribe Arisaemateae 28 14 97. Pinellia 98. Arisaema Tribe Colocasieae 99. Ariopsis 100. Protarum 101. Steudnera 102. Remusatia 103. Colocasia 104. Alocasia Tribe Pistieae 105. Pistia * No attempt has been made in this case to arrange the numbers into euploid series in this genus. 34 T H E G E N E R A O F A R A C E A E

C 10 PA L Y N O L O G Y Thanikaimoni (1969) and Grayum (1984, 1985, 1992a) associated with beetle pollination and spinose pollen have given detailed comparative surveys of aroid pollen with fly pollination (Grayum 1984, 1985, 1990). Beetles structure. These studies showed that palynological char- (Phaeochrous camerunensis, Scarabaeidae-Rutelinae) acters are important for the suprageneric taxonomy. and blowflies (Calliphoridae) were both observed as pollinators in Amorphophallus maculatus which has Aperture type is the feature of most general value. almost smooth pollen grains (Bogner 1976a). Bisexual-flowered genera have monosulcate, exten- sive-sulcate, meridionosulcate (zonate), diaperturate The shape is globose to ellipsoid, boat-shaped or or forate grains, but monoecious genera have inaper- hamburger-shaped (the latter in Gonatopus). The polar- turate grains. The only exceptions seem to be tribe ity is heteropolar to isopolar or apolar. Usually the Spathiphylleae (bisexual, inaperturate), Anadendrum pollen grains are found in monads, only two genera (bisexual, inaperturate), and tribe Zamioculcadeae (Xanthosoma, Chlorospatha) shed the pollen grains in (monoecious, extended-monosulcate to zonate). tetrads, arranged either tetragonally or serially (Chlorospatha longipoda). Ornamentation may be smooth (psilate), scabrate, foveolate, reticulate, spinulose, spinulose-reticulate, Grain size (measurements given here and in the spinulose-pilate to papillate, spinose, fossulate, rarely generic descriptions are mostly taken from Grayum, gemmate, verrucate, retiverrucate, areolate, rugulate 1984, 1992a) varies considerably, from small (12 µm in to tuberculate, striate, striate-verrucate, striate-reticu- Homalomena versteegii) to very large (114 (120) µm in late, striate to plicate or baculate. Spiny pollen grains Pseudohydrosme gabunensis) but the majority of the are common in Araceae, and have been considered genera (68%) are medium-sized (between 25–50 µm (Grayum 1984) as an adaptation for aiding attachment diam., mean 37 µm). to insect vectors which have hairy bodies. In many other genera, smooth pollen grains are extruded in In 73% of species examined the pollen contains strands composed of many grains glued together by starch although this may vary within a single genus; in pollenkitt. These strands also adhere to insect bodies, Schismatoglottis some species have starchy pollen and sometimes through the aid of sticky secretions within others do not. Nine out of ten genera with monosulcate the inflorescence (e.g. Philodendron). On the basis of pollen grains were found to be starchless which suggests existing observations, smooth pollen is almost always that this is the primitive type in Araceae. P A L Y N O L O G Y 35

C 11 P H Y TO C H E M I S T RY A N D C H E M OTA X O N O M Y by Robert Hegnauer Several recent reviews have dealt with chemical aspects distribution of kestose- and isokestose-types of sucrose of the biology of Araceae (Hegnauer 1963, 1986; fructosides in the basal parts of fresh monocotyledo- Dahlgren & Clifford 1982; Dahlgren et al. 1985; Bown nous stems; no such oligofructans (oligofructosans) 1988). Many references to the chemical characters and were present in Acoraceae (Acorus gramineus) or the ethnobotany of the family are available in the two Araceae (Arisaema atrorubens (= A. triphyllum), treatments by Hegnauer and in chapters 9 and 10 of Dieffenbachia picta (= D. maculata) and Peltandra Bown’s book. The present review supplies only the virginica were investigated), Lemnaceae, Alismataceae, most essential bibliography, and for additional refer- Dioscoreaceae, Sparganiaceae and Arecaceae (see e.g. ences the reader should refer to these three sources. Dahlgren et al. 1985: 277). 1. Phytochemistry Sakai & Hayashi (1973) studied the distribution of starchy and sugary leaves in monocots. Japanese Mineral deposits and primary metabolites Acoraceae and Araceae belong to those taxa which temporarily store sugars and non-starchy polysaccha- Oxalic Acid rides, but little if any starch, in the leaves. Acorus gramineus, Amorphophallus konjac, eight species of Aroids produce large amounts of oxalic acid, most Arisaema, Calla palustris, Colocasia esculenta, of it being deposited as crystals of calcium oxalate. Lysichiton camtschatcensis and Pinellia ternata were Raphide bundles (see also chapter 3), i.e. agglomera- investigated. Starch was seen occasionally only in tions of large, needle-like crystals lying parallel to one Acorus gramineus and in two species of Arisaema. another, are the typical crystal form of the family. These Araceae, which have “sugary leaves” thus differ from raphide bundles usually occur singly, embedded in the starchy leaved families Alismataceae, mucilage within large idioblasts (see section on Dioscoreaceae and Tricyrtidaceae, but resemble most Irritants in Araceae). Other types of calcium oxalate Japanese members of the Liliiflorae (sensu Dahlgren & crystals are found in aroids, such as druses, and in Clifford 1982). Acorus (Acoraceae), a genus lacking raphides, cells containing a solitary crystal are situated in rows accom- Mucilages panying fibres. For more detailed studies of calcium Vegetative parts of the Araceae mainly store starch oxalate crystals in Araceae, see Seubert (1993). According to Molisch (1918), who investigated (Czaja 1969, 1978a, b; for economically important Amorphophallus rivieri (= A. konjac), Caladium Araceae see also Brücher 1977, Mansfeld 1986, Palmer nymphaeifolium (probably a variety of Colocasia escu- 1989, and for taro – Colocasia esculenta – p. 54 and lenta), Monstera deliciosa and Sauromatum guttatum table 4.28 in Sunell & Arditti 1983). In some taxa starch (= S. venosum), aroids also tend to accumulate mod- is accompanied by mucilages in considerable amounts, erate to large amounts of soluble oxalates in leaves. In which usually consist mainly of glucomannans this respect they are similar to Lemnaceae, Helobiae (= (Hegnauer 1963, 1986). Ohtsuki (1967) showed that Alismatiflorae) such as Stratiotes aloides and Vallisneria subterranean parts of most Acoraceae and Araceae spiralis, and Zingiberales (Canna, Musa, Maranta). contain much starch but only negligible amounts of Silica, aluminium and heavy metals are not known to glucomannans (Acorus calamus, Arisaema atrorubens be accumulated significantly by members of the fam- (= A. triphyllum), A. japonicum (= A. serratum), A. ser- ily. Tubers of Eminium spiculatum and Arisarum ratum, A. thunbergii, Lysichiton camtschatcensis, vulgare were found not to contain soluble oxalic acid; Pinellia ternata, P. tripartita, Amorphophallus cam- tartaric and citric acid were detected in both species panulatus (= A. paeoniifolius) and A. kiusianus were (Ahmed et al. 1968). investigated). In certain other species of Amorphophallus starch is partly (A. bulbifer) or largely Carbohydrates (A. konjac, A. oncophyllus, A. variabilis) replaced by glucomannans, which are located in giant idioblasts. A The carbohydrates stored by Araceae have been procedure to isolate starch-free glucomannans from investigated many times. Pollard (1982) studied the tubers of Amorphophallus rivieri (= A. konjac) was described by Wootton et al. (1993); this glucomannan had a Gluc : Man ratio of 58 : 42. Literature concerning mucilages of Araceae is partly contradictory (Ahmed et al. 1968; Sunell & Arditti 1983: 36 T H E G E N E R A O F A R A C E A E

54–55). This is not surprising since mucilages are com- Pectins plex mixtures of heteropolysaccharides and very The following remarks concern the occurrence of difficult to obtain in pure form. Crude mucilages are always mixed with variable amounts of proteins and pectins in non-lignified cell walls. Jarvis et al. (1988) pectic and other substances (see section on Pectins). If showed that dicots and part of the monocots have pri- large amounts of mucilage are present in large mary walls with more than 150 mg of galacturonans idioblasts it is relatively easy to obtain a fairly pure per gramme of cell wall preparations (high contents). mucilage, as is the case with the glucomannans of Grasses and other Commeliniflorae had low (< 50 some species of Amorphophallus. If the major storage mg/g) galacturonan contents. Alocasia and Lemna compound is starch, however, the water-soluble “poly- (Ariflorae) and all investigated Alismatiflorae and saccharides” will always be of different origin, either as Liliiflorae belonged to the high-content group; some pectic substances, true mucilages such as those con- monocot taxa had intermediate (50–150 mg/g) galac- tained in raphide idioblasts, some cell wall turonan contents (see section on Mucilages). hemicelluloses or non-mucilaginous substances like proteins. According to Amin (1955) and others (see Secondary metabolites Sunell & Arditti 1983), mucilages purified from taro tubers (in several cultivars) are essentially branched Saponins, phenolic compounds including arabino-galactans with an approximate gal:arab ratio of flavonoids, cyanogenic glucosides, the widespread 8:1 to 11:1; this is perhaps the mucilage of the raphide occurrence of constituents which cause skin irritation idioblasts. The mucilages isolated in 3–4% yields from and painfully acrid sensations on mucous membranes tubers of Eminium spiculatum and Arisarum vulgare (mouth, throat, eyes) and calcium oxalate raphides by Ahmed et al. (1968) were mixtures of pectic sub- may be considered the key chemical characters of the stances, hemicelluloses and true mucilages. family (Hegnauer 1963, 1986; Bown 1988). Starch grain morphology Saponins Czaja (1969, 1978a, b) investigated rhizomes, Saponins are by no means ubiquitous but were tubers, corms, stems and seeds of many araceous taxa shown to be probably present in a number of taxa. for the presence and structure of starch grains and Hegnauer (1963) stressed the lack of chemical knowl- observed two main types: large grains of the so-called edge about araceous saponins; the occurrence of envelope-layer type (Hüllen-Lage-Stärkekörner) and steroidal sapogenins was and still remains uncertain small granules which are aggregated in compound (Hegnauer 1986). Nevertheless, the occurrence of grains. Two main subtypes of compound grains were steroidal saponins was mentioned for Montrichardia distinguished by Czaja: highly compound grains and Pinellia (Dahlgren & Clifford 1982) and for Arales (hochzusammengesetzte Stärkekörner, Czaja 1969: without indicating genera (Dahlgren et al. 1985). This 37–38), composed of 400 and more small granules, discrepancy is most probably caused by the fact that and compound grains (höher zusammengesetzte Altman (1954 in Hegnauer 1963) reported Stärkekörner, Czaja 1969: 34–37) with usually up to 10 Montrichardia arborescens to be a rich source of granules having a diameter of more than 6µm; these steroidal sapogenins. His analytical method, however, latter granules usually belong to the envelope type was unreliable. Haemolytic and foam-producing sub- (Hüllen-Stärkekörner). Highly compound grains were stances, generally believed to be saponins, were shown observed by Czaja (1978a) in all araceous seeds; 54 to be present in many species (Clark & Waters 1934, species from 17 genera were investigated. Rhizomes, Fontan-Candela 1957, Hegnauer 1963, Schroeter et al. tubers and stems are less homogenous (90 species 1966), but authors screening for steroidal saponins from nearly 40 genera were investigated). Some store (Villar Palasi 1948, Anzaldo et al. 1957, Wall et al. large grains and some store compound grains. Czaja’s 1954–1961) never obtained indications for this type of statements (1969: 36–38, 1978a: 60) concerning the saponin nor could they detect or isolate steroidal two subtypes of compound grains mentioned above sapogenins after hydrolysis (Marker et al. 1947, Wall et are sometimes discordant (e.g. Anthurium, Colocasia, al. 1954–1961). According to Villar Palasi (1948), Pinellia). On consulting the author’s table 1 (descrip- saponins are probably present in some parts of Alocasia tion of starch grains) and the figures of starches of odora and Arum italicum, according to Anzaldo et al. Amorphophallus campanulatus (= A. paeoniifolius), (1957) in Acorus calamus (Acoraceae), Amorphophallus A. oncophyllus and A. variabilis given in Ohtsuki’s campanulatus (= A. paeoniifolius) and Colocasia escu- (1967) paper, it becomes clear that Czaja’s two sub- lenta, and according to Wall et al. (1954–1961) in Acorus types of compound starch grains often merge into calamus (Acoraceae), several taxa of Colocasia, a one another. The highly compound type in pure form species of Philodendron and Symplocarpus foetidus is probably less frequent in vegetative parts of (confirmed by Segelman & Farnsworth 1969). All these Araceae than is suggested by Czaja’s list (1978a,b). For researchers failed to detect saponin-like substances in other detailed studies of starch types in Araceae, see Seubert (1993). 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 37

the investigated plant parts of Arisaema triphyllum, and ferulic acid are common in the family it is worth Arisarum vulgare, Amorphophallus bulbifer, several mentioning that they never seem to be bound to the taxa of Colocasia, Dieffenbachia cordata, Dracunculus cell wall polysaccharides. Harris & Hartley (1980) showed canariensis, Monstera deliciosa, Montrichardia that p-coumaric, ferulic and diferulic acids occur com- arborescens, Orontium aquaticum, Peltandra virginica, bined with non-lignified cell walls in all investigated Pistia stratiotes and Zantedeschia aethiopica. No members of Commelinidae (sensu Cronquist 1981) and appreciable amounts of saponins were present in in Arecaceae, Philydraceae, Pontederiaceae and Anchomanes difformis or Cyrtosperma senegalense (= Haemodoraceae in the strict sense. Araceae (Arum Lasimorpha senegalensis) (Delaude-Hulst 1974), or italicum, Pistia stratiotes and Sauromatum venosum have Typhonium brownii (Simes et al. 1959). Summarizing, been investigated) and Acoraceae (Acorus calamus), like it may be stated that saponin-like substances occur Alismatidae and typical Liliidae, lack this character. occasionally in Araceae but that the chemistry of the Acorus calamus has p-hydroxybenzoic acid combined saponins is still totally unknown. with non-lignified cell walls; this feature occurs erratically in vascular plants. Phenolic amines are dealt with in the Phenolic compounds section on Biogenic amines and alkaloids. Phenolic compounds occur in large amounts and are Recapitulating, it may be said that proanthocyani- structurally and biosynthetically diverse. Bate-Smith dins and C-glycoflavones and derivatives of caffeic acid (1968) investigated hydrolysed leaf extracts of 24 ara- belong to the main phenolics of the family, but that gly- ceous plants. Ten of them contained procyanidins coflavones may be replaced in certain taxa or (formerly leucocyanidins), seven contained quercetin, populations by flavonol glycosides (Stylochaeton; rutin six kaempferol, ten caffeic acid, seventeen p-coumaric in middle European Arum maculatum; kaempferol acid, twelve sinapic acid and eleven contained ferulic 3,7-bisglycoside in Gymnostachys anceps) or flavone- acid; in the case of Orontium aquaticum the presence O-glycosides (tribe Areae sensu Bogner 1979a) as main of scopoletin was indicated. A much more comprehen- leaf flavonoids (Williams et al. 1981, Hegnauer 1986, sive survey of proanthocyanidins, cinnamic acids, Williams & Harborne 1988). Leaves of Eminium spic- anthocyanins and flavonoids of Araceae was published ulatum yielded the C-glycoflavones vitexin, isovitexin by Williams et al. (1981). According to these investi- and its 7-glucoside (= saponarin), iso-orientin and its gators, procyanidins occur in leaves of nearly half the 7-galactoside, vicenin-1, the flavone-O-glycosides lute- investigated taxa and C-glycoflavones are the character- olin 3’-glucoside, luteolin 7-glucoside and chrysoeriol istic leaf flavonoids of the family. In some taxa they are 7-glucoside and the flavanone glycoside eriodictyol 7- accompanied or even replaced by O-glycosides of glucoside (Shammas & Couladi 1988). flavonols or flavones. Sulphates (esters of sulphuric acid) of vitexin, isovitexin, vitexin 7-glucoside, chrysoeriol- A second lignanoid compound (see also acoradin, galactoside and quercetin occur sporadically in Essential oils), a 1-ethyl-2-methyl-3-aryl-indane (Philodendron ornatum, Culcasia saxatilis, Scindapsus derivative, was isolated from rhizomes of Acorus cala- pictus). As yet not fully characterized sulphates of esters mus (Saxena 1986); it seems to arise by spontaneous of caffeic acid (sulphated caffeoyl glucoses?) are much dimerisation of asarone (Al-Farhan et al. 1992). A series more common, being mainly present in subfamilies of glucosylated lignans was recently isolated from sub- Monsteroideae (67%), Philodendroideae (23%) and terranean parts of Arum italicum, together with ferulic Pothoideae (20%); non-sulphated caffeic acid derivatives acid and glycosides of coniferyl alcohol and 4- with a free carboxylic group are common in subfamilies coumaryl alcohol (Della Greca et al. 1993). Whole Colocasioideae (80%), Lasioideae (38%) and Pothoideae plants yielded traces of 8-O-3’ and 8-O-4’ neolignans (20%) (subfamily circumscriptions in Williams et al. 1981 (Della Greca et al. 1994; see also Gellerstedt et al. follow Bogner 1979a). The results of this investigation 1995). Inflorescences of Zantedeschia aethiopica have been summarized twice by the authors (Harborne yielded the C-glycoflavones swertisin and swertia- 1982; Williams & Harborne 1988). It is highly probable japonin (Sivakumar & Nair 1992). that the two types of acidic caffeic acid derivatives detected by Williams et al. (1981) are involved in the irri- Cyanogenic glucosides tating properties distinguishing most aroids (see Irritants in Araceae). Ellis et al. (1983) showed that the chemistry Cyanogenic glucosides have been known since the of the proanthocyanidins of the fruits of Zantedeschia work of Jorissen, Greshoff and Treub (see Hegnauer aethiopica (two types investigated) and Z. rehmannii 1963, 1977, 1986) to be rather common in the family. varies with the taxon. They contain afzelechin, catechin The distribution and chemistry of cyanogenesis (the and (or) gallocatechin and their epimers as building ability to release HCN after injury) in aroids have been blocks and consist either of pure procyanidins, mixtures thoroughly discussed by Hegnauer (1963, 1973, 1977, of propelargonidins and procyanidins or mixtures of 1986). Additional observations have been published prodelphinidins and procyanidins. Since p-coumaric acid by McBarron (1972: one of three tested samples of Gymnostachys anceps was weakly and two were strongly cyanogenic), Kaplan et al. (1983: two out of 38 T H E G E N E R A O F A R A C E A E