P^a L^-airaa «!««»?*>• /Aarine Biological Laboratory Library Woods Hole, Massachusetts Morphology of the Angiosperms McGRAW-HILL PUBLICATIONS IN THE BOTANICAL SCIENCES Edmund W. Sinnott, Consultinci Editor ARNOLD An Introduction to Paleobotany CURTIS AND CLARK An Introduction to Plant Physiology EAMES Morphology of the Angiosperms EAMES Morphology of Vascular Plants: Lower Groups EAMES AND MACDANiELS An Introduction to Plant Anatomy HAUPT An Introduction to Botany HAUPT Laboratory Manual of Elementary Botany HAUPT Plant Morphology HILL Economic Botany HILL, OVERHOLTS, POPP, AND GROVE Botany JOHANSEN Plant Microtechnique KRAMER Plant and Soil Water Relationships KRAMER AND KOZLOwsKi Phvsiology of Trees LILLY AND BARNETT Physiologv of the Fungi MAHESHWARi An Introduction to the Embryology of the Angiosperms MILLER Plant Phvsiologv POOL Flowers and Flowering Plants SHARP Fundamentals of Cytology SINNOTT Plant Moi-phogenesis SINNOTT, DUNN, AND DOBZHANSKY Principles of Genetics SINNOTT AND WILSON Botany: Principles and Problems SMITH Cryptogamic Botany Vol. I. Algae and Fungi Vol. II. Bryophytes and Pteridophytes SMITH The Fresh-water Algae of the United States SWINGLE Textbook of Systematic Botany WEAVER AND CLEMENTS Plant Ecology There are also the related series of McGraw-Hill Publications in the Zoological Sciences, of which E. J. Boell is Gonsulting Editor, and in the Agricultural Sciences, of which R. A. Brink is Consulting Editor. "... Nee contentum exteriori rerum Naturae conspectu, intro- spicare .... Seneque" (Probably a paraphrase from Seneca's "Questiones Naturales," the preface of Book I.) Used by Van Tieghem on the title page of his "Recherches sur la structure du pistil," Mem. Acad. Sci. France, 1875. Two views of a primitive angiospemi Hower, Eupomatia bennettii: from below, show- ing stamens below the pseudoperianth; and from above, showing pseudoperianth covering the gynoecium and bearing food bodies, clearly seen on edges of blades. MORPHOLOGY of the ANGIOSPERMS Arthur J, Eames PROFESSOR OF BOTANY, EMERITUS CORNELL UNIVERSITY McGRAW-HILL BOOK COMPANY, INC. New York Toronto London 1961 MORPHOLOGY OF THE ANGIOSPERMS Copyright © 1961 by the McGraw-Hill Book Company, Inc. Printed in the United States of America. All rights reserved. This book, or parts thereof, may not be reproduced in anv form without permission of the publisher. Library of Congress Catalog Card Number 60-15757 III 18725 THE MAPLE PRESS COMPANY, YORK, PA. Preface Progress in the understanding of the morphology of the angiosperms has been rapid in the twentieth century; much new factual information has been obtained and many conceptions and interpretations have been changed. These changes have been brought about by the studies of many botanists scattered throughout the world in the several fields of moi-pho- logical study — descriptive, comparative, anatomical, ontogenetic — and these botanists have often brought to bear on their conclusions evi- dence from the allied botanical fields — taxonomy, cytology, paleobotany, serology, plant geography, palynology. The trend toward the use of a broader basis for the drawing of conclusions and for the proposal of new theories is apparent. At the end of the nineteenth century, Celakovsky emphasized the necessity for using "all evidence" in the interpretation of structure and in drawing phylogenetic conclusions; in the twentieth century, the need for a broad base for all interpretations has been re- peated by I. W. Bailey and many others. The importance of this empha- sis is being slowly recognized. This book has been prepared to bring together, in some measure, the results of these many scattered studies for the use especially of advanced students and teachers. It reviews much of the new factual material and many of the theories, old and new, related to the morphol- ogy and phylogeny of the angiosperms. Limitations in size of the book have restricted detailed descriptions and discussions of hypotheses, but the author believes that the important aspects of description and hy- pothesis are covered. In the manuscript, the names of plants that serve as examples are placed at the end of the sentence, set apart by a dash. The taxa cited are not necessarily the only examples. The viewpoint of the treatment is that of comparative rather than descriptive morphology, with emphasis on evolutionary modifications and phyletic implications. Velenovsky placed on the front page of his excel- lent textbook, "Vergleichende Morphologic," (1905-1913), these sen- tences as a maxim: "Zur morphologische Losung werden wir — wie immer — die vergleichende Methode in Anwendung bringen. Auf diese Wage vu Vin PREFACE werden wir zu dem richtigen und einheitlichen Ausschauung gelangen." The author agrees heartily with this point of view. The book assumes, on the part of the reader, an acquaintance with botany equivalent to that obtained from a general elementary course, although some elementary descriptions and discussions are, of necessity, included. During the preparation of this book, the author has been impressed with the importance of the earlier morphological literature, that of the middle of the nineteenth century especially, to the research of the twentieth century. Facts and theories discussed in the earlier period, overlooked or forgotten during the succeeding decades of increasing specialization, have been presented again in the twentieth century. The understanding of the earlier students of morphology is remarkable in the light of the small amount of factual information available to them; in- terpretations made a century ago often appear to be sounder than some later ones; today many botanists, in their highly specialized fields, lose sight of the broader aspects in the maze of details in their own particular area. In Chap. 11, the morphology of a few families is discussed in some detail, as examples of the use of morphology in determining relative advance in evolutionary modification, and to provide examples of struc- ture important in the interpretation of specialized form throughout the angiosperms. The families selected for discussion are those that possess the most primitive characters — the lowest dicotyledons and monocoty- ledons. The inclusion of the Amentiferae and Proteaceae — taxa some- times considered primitive dicotyledons — was planned for this chapter, but they were omitted because of restriction in book size. A bibliography for general consultation follows the last chapter, and a bibliography following each chapter covers the subject matter in the chapter. The bibliographies contain only a small part of the references consulted in the preparation of this book; the citation of all of these would fill another book. Selections were made on the basis of general importance from the viewpoint of the treatment in this book and of the size and excellence of their bibliographies. Also cited are works from which illustrations have been borrowed. The author is indebted to all the many botanists who, over a century and more and throughout the world, have laid the foundation for the treatment in this book. He is grateful to the many botanists from whom he has borrowed published illustrations and to Miss Elfriede Abbe who drew Figs. 28, 37, 38, and 97; to Mr. A. List who drew Fig. 3A to C; to Dr. A. T. Hotchkiss who provided the photographs of Etipomatia hen- nettii for the frontispiece and for Fig. 144; to Dr. L. J. Edgerton for Fig. PREFACE IX 133; to Prof. K. D. Erase who contributed the prohferated apple shown in Fig. 97; and to Dr. Roger Gauthier for Fig. 49D. He is also deeply indebted to his wife, Rita Ballard Fames, for extensive and continuing aid in many ways, especially in the preparation of the manuscript and the illustrations. Arthur J. Eames Contents Preface vu Chapter 1 The Plant Body ^ The Root. The Shoot. The Anatomy of the Phwt Body. The Ontogeny of the Phmt Body. Secondary Vascular Tissues and the Cambium. Bibliography ^■'■ Chapter 2 The Inflorescence "' Classification. Reduced Inflorescences. Phylogeny of the Inflorescence. Bibliography °'* Chapter 3 The Flower o" Basic Structure. The Primitive Flower. Reduction in the Flower. Bibliography 9"* Chapter 4 The Androecium "" Classification by Arrangement on the Receptacle. Reduction in the An- droecium. Fusion in the Androecium. Adnation in the Androecium. The Androecium under Zygomorphy. Ontogeny of the Androecium. The Stamen 108 The Primitive Stamen. The Anther. The Filament. The Sporangia. Anther- sac Wall and "Sporangium Wall." Unusual Forms and Arrangements of Stamens. Anatomy of the Stamen. Ontogeny of the Stamen. Morpho- logical Nature of the Stamen. Staminodia. Nectaries. Bibliography 154 Chapter 5 Pollen 158 Arrangement of Pollen Grains in the Tetrad. Structure of the Exine. Pollen- grain Types. Pollen-grain Development. Sequence within the Sporangium. Palynology. Dehiscence of the Anther 165 Pollination 169 The Male Gametophyte 1'^ The Pollen Tube. The Tube Nucleus. Male Gametes. Terminology of the Male Gametophyte. Bibliography 182 xi Xll CONTENTS Chapter 6 The Gynoecium ^""^ The Carpel ^8^ Form of the Carpel. Closure of the Carpel. The Comple.x Carpel. Modifi- cation of the Follicular Carpel. The Solid Carpel. The Stigma and Trans- mitting Tissue. Placentation. Morphological Nature of the Carpel. Axial Nature of the Carpel. Ontogeny of the Carpel. Anatomy of the Carpel. Syncarpy. The Inferior Ovary. Bibliography -^^'-^ Chapter 7 The Ovule 256 Definition of the Ovule. Form and Orientation of the Ovule. Ovule Number. Ontogeny of the Ovule. Anatomy of the Ovule. Reduction in the Ovule. The Nature of the Angiosperm Ovule. Bibliography - 287 Chapter 8 Archesporium 29(J Megaspore Mother Cell. Megasporogenesis. The Embryo Sac 292 Embryo-sac Types. Theories of the Morphological Nature of the Embryo Sac. The Endosperm 303 Types of Endosperm Formation. Cytological Make-up of the Endosperm. Ruminate Endosperm. Nature of the Endosperm. Bibliography ^^^ Chapter 9 Fertilization •^'^•^ The Embryo 310 Germination. The Angiosperm Embryo at Germination and Early Seedling Stages. Polyembryony 345 Anatomy of the Embryo and Young Seedling 347 The Two Major Types of Angiosperm Embryos 355 The Primitive Angiosperm Embryo 365 Bibliography 365 Chapter 10 The Seed • 369 Size of Seeds. Structure of the Seeds. The Fruit 375 Dehiscence. Pericarp. The Integuments in Fruits. Evolutionary Relation- ship of Fruit Types. Bibliography 380 Chapter 11 Notes on The Morphology of Selected Families 384 Ranales 384 Winteraceae. Lactoridaceae. Magnoliaceae. Annonaceae. Eupomatiaceae. Himantandraceae. Degeneriaceae. Myristicaceae. Schisandraceae. Illiciaceae. • •• CONTENTS Xlll Calycanthaceae. Trochodendraceae and Tetracentraceae. Austrobaileyaceae. Cercidiphyllaceae. Monimiaceae. Amborellaceae. Eupteleaceae. Ceratophyl- laceae. Ranunculaceae. Lardizabalaceae. Sargentodoxaceae. Beiberidaceae. Nytnphaeaceae and Cabombaceae. Lauraceae. Some Families Less Well Known Morphologically, Commonly Placed in the Ranales ( Hemandiaceae, Canellaceae, Trimeniaceae ) . Discussion and Summary of the Ranales. Dilleniales 432 Dilleniaceae. Paeoniaceae. Crossomataceae. Piperales 435 Chloranthaceae. Verticillatae ( Casuarinales ) 435 Casuarinaceae. Helobiales 437 Liliales 440 The Primitive Liliaceae. Palmae 442 Habit. The Leaf. Inflorescence. The Flower. The Fruit. Discussion. Bibliography 450 Chapter 12 Phylogeny of the Angiosperms 462 Relation of Monocotyledons to Dicotyledons. Origin of Angiosperms. Age of Angiosperms. Bibliography 470 General Bibliography 474 Index 477 Chapter 1 The angiosperms are the dominant seed-bearing plants of the present day, a vast and varied assemblage estimated to consist of 300,000 species. They are commonly considered the "modern" seed plants, geologically young. But, structurally, they do not appear to be of recent origin; even those families now considered most primitive have some well-advanced characters. Evidence is accumulating from the fossil record and from critical morphological studies that the angiosperms are an old group in which there was early differentiation along several lines and that the group had already become diverse and complex by the early Cretaceous period, long considered the time of their origin. The angiosperms, in their dominance, have not, however, "crowded the gymnosperms from the face of the earth," as is often stated or implied in elementary text- books; large areas of conifer forest exist in the tropics and in both northern and southern temperate lands; the cycads are dominant in some small areas in Australia and South Africa. In all characters, gross and minute, external and internal, sporophytic and gametophytic, the angiosperms show great diversity of form, a diversity that is clearly the result of adaptive specialization over a very long period of time and under great climatic changes. This specializa- tion leads to both increasing complexity and simplification of structure. Simplicity of form has long been considered evidence of primitiveness; the part played in specialization by reduction, "retrogression" or "sup- pression," has been commonly overlooked. The realization that sim- plicity is often secondary rather than primitive has played a prominent part in the interpretations of comparative morphology and phylogeny in the twentieth century. The results of reduction are prominent in the morphology of every part of the plant — in the body of the sporophyte, embryonic and mature, and in gametophyte origin and form. The angiosperms are commonly set apart from other seed plants by enclosure of the seeds, in contrast with the naked seeds of the gymno- sperms; by the presence of vessels in the wood; and by the possession of a complex reproductive structure, the flower. But these characters do not sharply limit the angiosperms. The carpel is open at the time of pol- lination in some taxa, and some conifers (Araucaria) have the seed enclosed; tlie vessel is present in Selaginella, Equisetum, Pteridium, Ephedra, Welwitschia, and Gnetum and is absent in many angiosperms. 1 MORPHOLOGY OF THE ANGIOSPERMS The carpel is still in process of closing; some stages are to be seen in living taxa. The vessel has arisen independently many times within the angiosperms, and all stages of its development are found in living angiosperms. A supposed basic difference in leaf-trace number be- tween angiosperms and gymnosperms — angiosperms with an odd, gymnosperms with an even number — has been shown to be invalid. But the angiosperms, though not set apart by any single character, clearly represent a well-defined stock, distinct from other seed-bearing plants. The sporophytic body of angiosperms shows greater variety in form and size than that of any other major plant taxon, ranging in habit from tree to herb; in size, from the minute Wolffia to the tallest Euca- lyptus; in form, from the simplicity of a thallus to the complex branch- ing of trees and giant vines; in flower structure, from the simplicity of a single, naked sporophyll to the complexity of organs of four types, with much connation and adnation. The gametophytes, although con- sisting of rather few cells, likewise vary a great deal, especially in cell number and arrangement. Structurally, the plant body consists of an axis, branched or un- branched, with lateral appendages.* The axis is commonly divided on structural and functional grounds into stem, with appendages and endarch primary xylem, and root, without appendages and with exarch primary xylem. Stem and root may form a continuous axis, as in most seedlings, with a structurally transitional section, the hypocotyl; or roots may develop as appendages of the stem. Continuity of root and stem in the embryo and seedling of the higher plants has been, in part, the basis for the theory that these two organs represent the specialized parts of an original primitive axis which constituted the entire sporophyte, as in simpler members of the Psilophytales. The stem alone has been considered to represent the entire primitive axis, with the root a secondarily developed organ. This concept is based on the endogenous origin of the radicle and on the common origin, in some embryos and seedlings, of roots as major appendages of the base of the stem. Evidence in support of this view has been seen in the absence or nonfunctioning of the taproot in some supposedly primitive Liliales, where the entire root system is formed by adventitious roots. (Under this theory, the absence of a taproot has been considered more primitive than its presence.) But absence or nonfunctioning of an em- bryonic taproot seems to be a derived condition. The vascular struc- ture of the hypocotyl, in its symmetry and relations to the primary vascular skeleton of stem and root, supports the view that the primary * Anatomical details, already discussed in Eames and MacDaniels, Introduction to Plant Anatomy, are herein largely omitted. THE PLANT BODY root is a basic part of the axis; adventitious roots, with simple, crude attachment to the mother axis, are secondary organs. Variety in growth habit ranges from tree and shrub to herb and vine, woody and herbaceous. The diverse habital forms represent many types and stages of specialization along parallel and convergent lines. Distinction between trees and shrubs is based chiefly on differences in height, woodiness, and permanence, but all habital types merge with others. Some herbs are taller than some trees. Many herbs, even annuals, have strong, woody stems, and some tropical trees have weak, "fleshy" trunks. Vines are both woody and herbaceous, with major and minor structural adaptations to the climbing habit. The tree habit is ancient among vascular plants. Trees of good size were present in the Devonian period and have been present in major taxa since that time. Because of their large size, trees are prominent among angiosperms: it has been estimated that there are 20,000 to 25,000 living species. Great height — over 100 feet — is attained in many families; 200 feet and over in a few. The Australian gums are, without doubt, the tallest angiosperms; Eucalyptus regnans reaches a maximum height of 326 feet. (Taller individuals reported are probably mythical.**) Herbs greatly outnumber woody plants in genera and species. Herbs are prominent in some areas because of the vast numbers of individuals. Their early evolutionary history is unknown; their softer structure is not so likely to be preserved in the fossil record as that of woody plants. THE PLANT BODY The gross structure of the plant body has been variously interpreted, morphologically. In early days, it was considered to be made up of several "fundamental parts" or organs — root, stem, leaf, floral organs, ovules, trichomes; in recent years, the number of these basic organs has been reduced to three — root, stem, and leaf; in present usage, the stem and its appendages are commonly considered as a unit, the shoot. Recognition of the shoot as a fundamental part has been a gradual process. The concept that stem and leaf together form an entity was probably first proposed at the beginning of the twentieth century. In succeeding years, three interpretations of the make-up of the shoot were proposed: (1) that the stem is the fundamental part, and the leaves appendages; (2) that the leaf is the basic part, and the stem consists wholly or in large part of proximal parts of the leaves; (3) that the shoot consists of units, "segments," called phijtons. The second of these theories was later expanded — without the addition of better " Forest Products Department, Australian Council for Scientific and Industrial Research Organization. 4 MORPHOLOGY OF THE ANGIOSPERMS morphological evidence — into the leaf-skin theory (p. 21). The third theory took several forms, including the phyton theory and "phytonism" (p. 21). (The term phytonism has also been applied to the theory that the stem consists of leaf bases. ) If the origin of the complex sporophyte of the higher vascular plants is to be seen in the simple body of the Psilophytales, the description of the body of seed plants as made up of two basic parts, root and shoot, seems reasonable. In general method of development — development by apical meristems — they are alike, but they differ in basic vascular struc- ture, in method of later increase in length, and in structure of the outer tissues. The intercalary growth of stems and leaves (origin of the petiole) is absent in roots. Histological elaboration of the outer tissues in stems is chiefly in the formation of a complex cortex; in roots, of a specialized epidermis with root hairs. In the angiosperms, the presence of an endodermis has been considered a root character, distinguishing root from stem, but an endodermis is present in stems of many taxa — in rhizomes, in seedlings, in aquatic plants, and in some parts of the stems of the majority of herbs. And there is excellent histological evidence in other taxa that the endodermis — characteristic of the entire plant body in many pteridophytes and probably present in the stems and leaves of ancestral angiosperms — has largely been lost. The endodermis ties to- gether root and stem as parts of a continuous plant body, and its pres- ence is not evidence that an organ is a root. Embryo structure has been seen to demonstrate the unity of the axis in embryos that have, in the hypocotyl, the radial vascular structure of the root and continuity of the endodermis in the hypocotyl and the epicotyl. This continuity of the endodermis is considered a re- capitulation of the evolutionary history of the plant body; the axis of the embryo represents the simple, primitive plant body. In the embryo, in the seedling, and persisting (except in some monocotyledons) in the mature plant, the hypocotyl is the section of the axis transitional anatomically from root to shoot; its origin goes back to the early elabora- tion of the sporophyte as shoot and root. The description of the axis of the angiosperm embryo as, in structure, a continuous organ, root at one end, stem at the other, has been con- sidered morphologically inaccurate; the embryos of most monocotyle- dons and those of some dicotyledons cannot be so interpreted. The plumule of these embryos is commonly described as "lateral," but the lateral position is only apparent; the plumule is morphologically termi- nal, as in the dicotyledons (see Chap. 9). The Root In the angiosperm embryo, root and shoot form a continuous struc- tural axis; in the pteridophytes, the embryonic root is the first of a series THE PLANT BODY O of lateral structures, which are temporary. (In the lepidodendrids, Pleuromeia, and Isoetes, the functional counterpart of the root system of seed plants is the rhizomorph, which, without true root structure, bears adventitious, temporary roots.) Continuity of shoot and root, with a transition region, the hypocotyl, is characteristics of the embryos of dicotyledons and most of the monocotyledons. In some of the arbore- scent monocotyledons, the embryo has no root or has only an abortive structure at the end of the stem. Absence of a radicle has been con- sidered by some morphologists the primitive condition in monocotyle- dons, but the presence of briefly functioning and nonfunctioning pri- mary roots in some monocotyledonous taxa is evidence that the taproot has been lost in these highly specialized taxa and adventitious roots on the hypocotyl and stembase have formed a root system. The arborescent shoot system in monocotyledons is a specialized type and is accompanied by a specialized, "secondary" root system. The Shoot Recent studies of apical meristems and the anatomy of the node have shown that stem and leaf are closely interrelated and have emphasized the concept of the shoot as, morphologically, a fundamental part of the plant body. Anatomically, no structural line, external or internal, sepa- rates stem and leaf. Ontogenetically, also, no separation can be made. There is no constancy in details of meristematic origin of these organs; histological limits of tunica and corpus vary even in the same plant. The region of union of leaf and stem, the nodal region, is one of merging tissues, cauline and foliar. Yet the node is commonly called a part of the stem and the region of attachment of the leaf is termed the leafbase. Tissues of leaf and stem merge in the nodal region, and limits cannot be set; this region is a part of the shoot, neither leaf nor stem. Though stem and leaf cannot be sharply delimited — and the leafbase should not be called one of the basic parts of the leaf — stem and leaf are units of body structure that must be treated as parts of the shoot. The term shoot is an old one, going back as far as the late nineteenth century. Stem and leaf were, in early studies, called "correlative parts" of the shoot, but "shoot" has long been suppressed, because of the supposedly important distinction between leaf and stem. Now this old, largely abandoned term has been revived and is slowly coming into use, because it has been found to be useful and morphologically valid. Although stem and leaf are convenient descriptive terms, neces- sary for use in most fields of botanical study, they should not be con- sidered primary categories of body structure. The flower, a reproductive shoot, is treated in later chapters. The Stem. The stem, as an axis, is described as divided into parts, the internodes, by more or less well-defined areas where appendages of leaf 6 MORPHOLOGY OF THE ANGIOSPERMS rank, the nodes, are attached. Nodes are often defined as levels of at- tachment of leaves, sometimes as levels of departure of leaf traces. Nodes and internodes are continuous parts of the stem; there is no structural delimitation. A description of "nodal anatomy," for example, necessarily covers much of the structure of internodes. Significant mor- phological differences in the shoot are chiefly those of internal structure; they are discussed under The Anatomy of the Plant Body later in this chapter. Ecological modifications of the stem and shoot lie outside the scope of this book, but the short shoot should have brief mention, because, in its extreme form, it involves the morphology of "terminal" leaves. In the gymnosperms, the slow-growing, lateral short shoots of Larix and Ginkgo and the determinate, deciduous short shoots of Pinus are well known. In Pinus monophyUa, the short shoot with a solitary pseudo- terminal leaf resembles a simple leaf. In the angiosperms, similar, but less highly specialized, slow-growing short shoots, with crowded nodes, are characteristic of species of Betula, Pijrus, and other genera. Stems terminating in reduced leaves are occasional in monocotyledons— Po/j/- gonatwn, Streptopus, Disporiim, some bamboos. The species of Uvularia show stages in the loss of the stem apex and origin of a terminal leaf. The Leaf. The leaf, commonly called an appendage of the stem, con- sists of two fundamental and more or less clearly distinct parts, the blade, or lamina, and the leafstalk, or petiole. A third part, the leafbase — "leaf cushion," "leaf buttress," "leaf foundation" — has long been recog- nized by many morphologists. (The term leafbase has also been applied to the basal, sheathing part of many monocotyledon leaves, both de- scriptively, to the winged base, and morphologically, to the part de- veloped from the base of the leaf primordium.) The leafbase, as a basic part of the leaf connecting petiole and stem, is not clearly defined, structurally. Some interpretations of leaf mor- phology call the leafbase a definite part of the leaf, the part that con- nects petiole and stem, "the base on which the leaf stands." In other interpretations, this base is merely the connecting or transitional region between leaf and stem and not a fundamental part of the leaf itself. The limits of the leafbase are not recognizable, ex- ternally or internally. In Europe and Asia, the leafbase has been generally recognized as an important part of the leaf; in America, it has received little attention. Definitions of the leafbase are loose and various, and often seem remarkable as examples of definitions of a morphological entity: the leaf primordium, the basal part of the leaf primordium, the ontogenetic buttress or foundation of the leaf, the hypopodium, the part of the apical meristem of a vegetative shoot where a leaf arises that has no share in the development of the leaf it- THE PLANT BODY self, the connecting tissues of leaf and stem, the region in the stem and leaf where lateral leaf traces change course toward the median trace, the segment of the axis which subtends a leaf initial and surrounds the leaf trace as it differentiates. The failure of American morphologists to recognize the leafbase as an important part of the leaf cannot be under- stood by some Indian morphologists, "because a leaf without a base cannot be conceived." Ontogeny provides the best basis for an understanding of the nature of the leafbase. The leaf primordium arises as a lateral, more or less crescent-shaped mound or ridge on the apical meristem of the shoot. From this early primordium, as a base, arise a median conical or cylindrical projection — from which the blade and petiole develop — and, if stipules are present, lateral lobes. If a leafbase is to be recognized, it is recognizable only in shoot ontogeny as the base of the leaf primor- dium; it cannot be distinguished in the mature leaf or node. The leafbase has been well described as a primordial region in the young node, consisting of two components — a proximal cauline or axial part, and a distal foliar part. Together, these components form the part of the shoot transitional from stem to leaf, a part not limited in- ternally but sometimes limited externally as decurrent areas of epi- dermal and cortical tissues set apart by unusual color or vesture. This description sets no limit internally in the stem for the leafbase; others set as the limit "to the pith" or "sometimes to the pith"; still otliers "to the center of the pith." Interpretations of the radial extent of the leafbase conflict with morphological distinctions within the stem. If the leafbase tissues extend only to the pith, the leafbase, as a part of the leaf, clothes the pith, or "core," as a "mantle," and the shoot axis con- sists of the pith only; if the leafbase extends to the center of the pith, no axis, as such, exists in the nodal region. Structure of the apical meristem is reported to show that the view that the leafbase tissues extend to the center of the axis is incorrect, because "two different zones of the meristem form mantle and core." The leafbase, if such a structure is worthy of distinction, is a segment of the shoot, not of the leaf. (The extent of the leafbase as a unit of structure of the shoot is further discussed under the leaf-skin theory, p. 21.) If the shoot is accepted as one of the two basic parts of the plant body, the leaf can well be called a "partial shoot arising from a parent whole shoot." There is then no necessity to distinguish a part of the shoot, partly stem and partly leaf, as a fundamental part of the leaf. The term partial shoot is good in its implication that the leaf is a lobe of the axis, but not because it has "lost its adaxial part in the develop- ment of the dorsiventral form," as suggested under one theory. The leaf- base is not recognized in this book as a part of the leaf; it is considered 8 MORPHOLOGY OF THE ANGIOSPERMS a part of the shoot, a nodal region transitional between stem and leaf — a transition region. Because the term leafbase is used not only in this way but also, loosely, for the proximal part of leaves— which is morpho- logically various— the term leaf buttress would distinguish the part that forms the nodal transitional region from that which forms petiole and blade. The angiosperm leaf blade, or lamina, is remarkable for its extra- ordinary variety of form. Descriptively, two chief types are distin- guished as simple and compound. From the phylogenetic viewpoint, each of these has been considered primitive, but strongest support has come for the view that the simple leaf is primitive. The leaves of the woody ranalian families — now recognized as show- ing many primitive characters and probably the most primitive living dicotylecions — are simple. The simple form is apparently more common in the fossil leaves of the Cretaceous and Tertiary periods. Comparative study shows many series leading from simple, through increasing dis- section, to compound — in Acer, in Rubtis, in the Vitaceae. These series can hardly be read in the other direction. Chief support for the com- pound leaf as primitive has come from families where reduction of leaflets to one is shown by comparison of many taxa and by the presence of vestigial structure in leaflets, petiole, and rachis— Leguminosae ( Fig. 1), Rosaceae, Rutaceae. But the woody ranalian families, with their highly primitive flowers and wood, have simple leaves. The decom- pound leaves of some genera in families with primitive flowers — espe- cially the fernlike leaves of some genera of the Ranunculaceae and Fumariaceae— have been regarded as possible evidence that these taxa retain in some measure the leaf form of an ancestral fernlike stock. Under the Durian theory of the habit of the primitive angiosperms, the large, pinnately compound leaf is considered the form characteristic of early angiosperms. The basis for this conclusion is that leaves of this type accompany the growth habit believed primitive and that they in- habit tropical forests, where today the most primitive living angiosperms are found. But the woody ranalian families with simple flowers and anatomy also inhabit these forests, and their leaves are simple. Though the simple leaf seems undoubtedly at least one primitive type, the simple leaves of many taxa clearly represent modified compound leaves; the lamina of the compound leaf has been re- duced by loss or fusion, or by both loss and fusion, of leaflets. Un- questioned reduction of these types is present in the Leguminosae, Rosaceae, Oleaceae, Rutaceae, Proteaceae. Evidence of loss and fusion of leaflets comes from comparison of closely related taxa, with structural evidence in venation and in the presence of an articulation between rachis and petiole below the solitary leaflet (Fig. lA). The genus Citrus Fig. 1. Sketches of leguminous leaves showing evidence of reduction in compound leaves. A, in Ccrcis canadensis, survival of a terminal leaflet, articulation prominent; B, in a 5-foliolate leaf of Trifolium arvense, reversion of normal trifoliolate to an- cestral multifoliolate; C to H, in leaves of Bauhinia (C, B. saigonensis; D, B. acu- minata; E, B. reticulata; F, B. mollicella; G, B. malabarica; H, B. retusa), a series in connation of two lateral leaflets to form one pseudoterminal leaf, the tip of the rachis still evident in D to G. 9 10 MORPHOLOGY OF THE ANGIOSPERMS is often cited as an example of this reduction. In the Oleaceae, Syringa has both simple- and compound-leaved species; Forsijthia may have 3- foliolate leaves among the simple ones. Fraxinus, which usually has many leaflets, has two species with only one, and unifoliolate forms oc- cur in other species. The Proteaceae perhaps show the most complete story of lamina reduction. Tropical genera with the most primitive floral characters — for example, Placosperma, Hicksheacliia, Austromuellera — have large compound leaves; the genera of more temperate climates show many types of reduction, as- sociated with xerophily, to small and secondarily simple. The leaves of Baiihinia and re- lated genera show both loss of leaflets and fusion of a surviving distal pair to form a solitary pseu- doterminal leaflet. All stages of the union are seen within the genus Baiiliinia — from two free leaflets to one apparently simple blade (Fig. IC to H). Involved in this union is the free tip of the rachis between the two leaflets. (Similar fusion occurs in inflorescences where two lateral flowers unite to form a pseudoterminal flower — Lonicera spp.) In most stages of leaflet union, the fusion is con- genital, and in the species with compound Teaf of Carija buckleyi \iu. completely fused leaflets, there arhansana, showing acropetal sequence in leaflet formation from lateral primordia. c, cortex of leaf axis; h, hair; lU to IL, primordia of lateral leaflets; tl, terminal leaflet. (After Foster.) Fig. 2. Longitudinal section of developing remains little or no evidence in form or venation of the double nature of the "terminal" blade. The compound leaf, as an ad- vanced type, is believed to have arisen by the evolutionary dissection of the simple leaf. In ontogeny, the leaflets develop as do the lobes of a simple leaf — by the development of lateral primordia on the median axis (Fig. 2). The development of the compound leaf of the palms is wholly different — by an ontogenetic split- ting of the primordium ( Fig. 18 ) . In gross venation, pinnate venation appears to be primitive, the palmate derived. Many transitional forms occur. Evidence that pinnate venation is primitive is found in the anatomy, especially in the THE PLANT BODY 11 leaves of the woody Ranales. The pattern of minor vein arrangement is also of two major types — reticulate or netted, and parallel or striate — but there are intermediate types. There are many exceptions to the characterization of the leaf venation of dicotyledons as netted and that of the monocotyledons as parallel; the leaves of many of the lower monocotyledons have netted veins, and some of the dicotyledons — the Epacridaceae, for example — have parallel veins. The venation of the Epacridaceae contrasts strongly with that of the closely related Ericaceae and has clearly been derived from the reticulate type. (Modification of nodal structure has accompanied the change in venation; a multilacunar node in the Epacridaceae has replaced the unilacunar node in the Ericaceae. ) Perhaps the most important feature in the arrangement of the lesser veins is the presence of free vein endings within the vein eyelets in some taxa and the absence of free endings in other taxa. The view that signif- icant changes in size of the vein eyelets occur with increasing age of the plant was shown to be without foundation. Vein eyelets are usually of irregular form, but, in some tropical genera, tliey are rectangular and remarkably uniform in size. The petiole arises late in the ontogeny of the leaf, by intercalary growth of a region at the base of the lamina. Increase in length is probably chiefly by cell enlargement. The delayed development of the petiole has been considered evidence that the petiole is a recent de- velopment in the specialization of the leaf. But this seems unlikely, because sessile leaves can often be shown, by comparison with the leaves of related taxa, to have lost their petioles. The leaves of the primi- tive woody dicotyledons are petiolate, and the leaflike carpels of many of them are stipitate. (The homology of leaf and carpel has been strongly supported in recent years by evidence from ontogeny and anatomy. ) Stipules are lateral parts or appendages of the leaf, usually borne one on each side at, or near, the base of the leaf. When borne on the stem free from the leaf itself, they appear to be independent organs and have been called "cauline stipules," an unfortunate term which implies that they are part of the stem. But anatomical structure shows that these stipules are a part of the leaf; their vascular supply is derived from the lateral leaf traces. Although this anatomical relationship was shown in 1880, its morphological significance was not recognized until nodal anatomy was critically studied in the early decades of the twentieth century. Stipules have been called "appendages of the leafbase, not of the leaf" because they "arise from the leafbase." But only independently at- tached stipules arise in this manner; other stipules arise congenitally 12 MORPHOLOGY OF THE ANGIOSPERMS fused with the leaf primordium as this meristem develops on the leaf buttress. In some compound leaves, appendages — stipels — occur at the bases of the leaflets. Stipels are surely a part of the leaf, and are evidence that stipules also are a part of the leaf. Stipules rarely occur on cotyledons. They are occasionally present on perianth parts in some of the more primitive monocotyledons — Melunthium, Zijgadenus — and on stamens — Ornithogalum and other liliaceous genera and many of the Bromeliaceae. Allium shows in its many species great variety in the stipules of stamens, from well-developed to vestigial, and some species have none. Stipules have been rather loosely classified on the basis of form and relation to the leaf and stem. They are called lateral when they seem to be lateral parts of the leaf, appearing like lobes of the lamina or wings of the petiole. Lateral stipules that are borne on the stem inde- pendently of the leaf are termed free (Fig. 3A, B, C)— Begonia, Vitis, Hydrocotyle, Liriodendron; they are termed adnate where fused along one margin to the petiole for all or part of their length— many Rosaceae (Fig. 3D), Leguminosae. The term adnate is sometimes restricted to stipules that are fused to the body of the leaf throughout their length, as in the sheathing leaves of many monocotyledons; this type is also called "vaginal." Ventral stipules are those whose margins meet on the ventral side of the petiole — Artocarpus, Magnolia, some species of Begonia. In axillary stipules, the margins are fused above the petiole, forming a single leaflike pseudoaxillary structure. Axillary stipules have also been called intrapetiolar. Intrapetiolar stipules are formed by the union of stipules of two different leaves at the same node — Fuchsia, Elatine. In the whorls of leaves of Galium ( Fig. 3C ) and related genera, there are only two true leaves; the other "leaves" are intrapetiolar stipules or, in part, individual simple stipules. Free stipules arise early on the leaf buttress, close to the primordium of the blade and petiole, but are isolated from the rest of the leaf during the enlargement and maturation of the buttress. Stipules, common in most of the less advanced angiosperms, are absent in most of the Sympetalae; they are present in some of the Rubiaceae. Some families in the Archichlamydeae — Rosaceae, Legumi- nosae — have stipulate and estipulate genera. In the woody Ranales, now considered very primitive, there are estipulate families — Winteraceae, Degeneriaceae, Eupomatiaceae, Annonaceae, Himantandraceae, Moni- miaceae, Calycanthaceae, Lauraceae; and stipulate famflies — Magno- liaceae, Chloranthaceae. Stipules commonly accompany the woody habit; 40 per cent of woody dicotyledons are stipulate, as contrasted with 20 per cent of herbs. In the monocotyledons, stipules, as paired, free appendages of the THE PLANT BODY 13 leaf, are rare; they occur in the Hydrocharitaceae, Butomaceae, Najadaceae, perhaps some of the Dioscoreaceae. The monocotyledons have sometimes been described as lacking stipules, but there is little agreement in the interpretation of the sheathing base of the monocotyle- don leaf. The thin margins are commonly interpreted as representing Fig. 3. Sketches of stipules showing variations in type. A, Hamamelis, simple, free; B, Viola, deeply dissected, free; C, Galium, two leaflike stipules with each leaf, free; D, Rosa cinnamomea, leaflike, adnate to petiole. {A-C, drawn by A. List.) adnate stipules that have lost identity as such in adaptation to the gen- eral monocotyledon habit of "telescoped shoot." The leaf sheath is also sometimes considered an elaboration of the primordial leafbase, without distinction of blade and stipules. The loose use of the term ligule for the wings of the sheathing leaf adds to the difficulty of interpreting the monocotyledon leaf as a whole. Evidence in support of the concept that the stipules form the marginal parts of the sheathing base has been found in those monocotyledons that show paired adnate stipules, especially Totamogeton. In this genus. 14 MORPHOLOGY OF THE ANGIOSPERMS the leaves of seedlings show stages in the development of the sheathing base by the merging of adnate stipules with the base of the petiole (Fig. 4). In many families, there is evidence of reduction of the stipules, asso- ciated with the loss of their function as protective structures for meri- stems and young leaves; in the Leguminosae, Droseraceae, Onagraceae, Cornaceae, and many others, there are taxa with poorly developed or vestigial stipules and other taxa without stipules. In some species, stipules are present in the upper leaves, absent in the lower; in other E F Fig. 4. Sketches of stipules of Potamogeton showing stages in connation of a pair of stipules to form a stipular sheath. A, P. perfoliatus; B, C, P. lucens; D, E, P. natans; F, P. crispus. (After Monoijer.) species, present in the lower, absent in the upper leaves. In Tropaeolum, they are present only in seedlings. Stipules appear to be a disappearing feature of the angiosperm leaf. In dicotyledons, they have apparently been lost in most of the higher families; in the monocotyledons, they may survive as a part of the sheathing leaf, an adaptation to the domi- nant monocotyledon habit. Anatomical evidence for the primitiveness of the stipulate condition seems contradictory. Stipulate taxa commonly have trilacunar nodes; they rarely have unilacunar nodes with an odd number of traces. If the unilacunar node with two traces is primitive (Fig. 6A), as now seems THE PLANT BODY 15 probable, the primitive angiosperm leaf should be stipulate. The primi- tive Helobiales fit into this picture, as do the Magnoliaceae and a few other woody-ranahan taxa, but other woody-ranalian hues are without stipules. Stipules have been interpreted as basal leaflets of a compound leaf and as remnants of an ancestral whorl of leaves, but neither ontogeny nor anatomy supports these views, and, in angiosperms, whorled leaves represent advanced phyllotaxy. The stipules of the gymnosperms and ferns (except perhaps the Eusporangiatae) are not nodal appendages. Prolongations of the stipulate margins beyond the median part of the sheathing base of monocotyledon leaves form a more or less free structure, the ligitle. The ligule may be a two-lobed or two-toothed structure, or a simple structure where the two parts are united as a ventral "collar." (This collar may become greatly enlarged and form an important part of the leaf. ) Formed in this way, a ligule is obviously stipular in nature, but the interpretation of the ligule has been expanded to cover the basal sheath, with which the ligule is continuous, and the entire sheath called the ligule. This understanding of the ligule has brought about great confusion in the interpretation of the monocotyle- donous embryo (Chap. 9). The ligule is not morphologically distinct from the stipules, of which it usually represents the distal part. (A descriptive distinction becomes necessary only in the interpretation of the coleoptile of the grasses and some other families where the coleoptile is called "ligular." The coleoptile is a part of the sheathing leafbase, not of the ligule.) Ligules are present not only in many monocotyledons but also in some dicotyledonous families — Droseraceae, Saxifragaceae, Araliaceae, Piperaceae. The ligule, or hastula, of the palms represents a part of the telescoped rachis (Fig. 148). Whether or not the stipules are recognized as parts of the primitive angiosperm leaf, they show great variety in form and relation to other parts of the leaf. Evolutionary modification is obvious, and there are two very different concepts of the direction of the series. According to one concept, paired stipules, free or partly adnate, represent the primitive leaf structure. With specialization, adnation became complete, with the stipule tips becoming auriculate at the base of the blade. Re- duction and loss of the auricles left a simple sheathing leafbase, con- sisting of the united "leafbase" and the stipules, an apparenUy simple but really complex structure. According to the other concept, the primitive angiosperm leaf had a prominent sheathing base and a blade, sometimes separated by a petiole. In specialization, this sheath became auriculate and, with shortening of the sheath, became auricles. Under this concept, free stipules are the higher type; under the other concept, free stipules are the most primitive. Under one theory, the monocotyledons, under the 16 MORPHOLOGY OF THE ANGIOSPERMS other, the dicotyledons have the most primitive stipules. Evidence from nodal structure, from seedlings, and from comparison in related taxa indicates that the free stipules are the more primitive form. The many traces of the monocotyledon-sheath margins suggest advanced structure. (The multilacunar node is generally recognized as an advanced type.) Seedlings in some monocotyledons show progressive stages in their early leaves from paired free stipules to sheathing leafbases — Potamogei^on. Free paired stipules are present in Potamogeton densus, in species of Ruppia, Althenia, Najas, Hydrocharis, Smilax. Increasing adnation of free stipules is seen in many dicotyledonous families — Rosaceae, Leguminosae. Fused adnate stipules and sheathing leafbases are fre- quent in dicotyledons, rare in monocotyledons — Zannichellia palustris. The theory that free basal stipules are the primitive type seems best supported by evidence of all types. In stipular type, the leaves of the dicotyledons are more primitive than those of the monocotyledons. On the basis of the theory that paired lateral stipules are primitive, two views have been held as to the relationship of the free and the adnate. It is claimed that the free stipules stood at first at the base of the petiole; that a change to a free (cauline) position came later as the leaf buttress matured and displaced them. The other view is that the position of the free stipules is the primitive one in the shoot, that a closer relation to the leaf is an advanced character. Ontogenetic evi- dence can be found to support each of these views; to determine the ancestral position of the stipules, evidence from ontogeny alone is not sufficient. Small, reduced leaves borne close to the base of lateral branches have long been termed prophyUs. They have also often been called hracteoles, though this term is usually restricted to inflorescences and flowers. Commonly, there are two lateral prophylls in the dicotyledons and one in the monocotyledons. Where there are two prophylls, they may stand "opposite," forming a pair, or one may be distal to the other. Definitions of prophylls emphasize their position in relation to leaves on the branch; they do not fit into the phyllotactic spiral of the branch. It is apparent that they do not when the pairs are opposite but when one prophyll is above the other, they commonly seem to fit into the spiral. The numerical distribution of the prophylls — two in the dicotyledons and one in monocotyledons — has sometimes been considered significant, because it seems to parallel cotyledon number, but there are many ex- ceptions. In inflorescences, prophylls may be prominent organs, espe- cially on the ultimate branchlets, and even on the pedicles, as in the Cyperaceae. (Here they are commonly called bracteoles.) The solitary prophyll is usually superaxillary, with its ventral side THE PLANT BODY 17 toward the lateral branch. The form and anatomy of some solitary prophylls suggest that the solitary organ represents two fused prophylls. The solitary prophyll may have two apices and is frequently bvo- keeled. Two to seven vascular bundles are present. Where there are two bundles, one lies in each keel and one is larger than the other; there is no median bundle. The older interpretation of this asymmetry is that compression of the organ in ontogeny has distorted it, so that the mid- dle part has become lateral. Form, position, and structure support the view that the solitary prophyll clearly represents a fused pair; the asymmetry results from closeness to the mother branch and con- sequent difference in development. Location and number of buds axillant to the prophylls should be further evidence of number of organs present, but it is claimed that in the prophylls of some monocotyledons there is only one bud. This is perhaps the condition to be expected if prophylls are the reduced first leaves of a shoot; the distal node is nor- mally the better developed and the only one to have an axillary bud. Anatomy demonstrates that the prophylls are not stipules of the sub- tending leaf; their vascular supply is derived from the lateral branch, not from the mother stem. Prophylls seem to have no significance as unique appendages in the shoot; they are merely leaves of reduced form, sometimes in apparently peculiar positions. They seem to represent the first appendages of a lateral shoot, weakly developed and sometimes displaced by closeness to the mother shoot. The prophyll of monocotyledons is undoubtedly a pair of leaves. The terminal leaf has been described as present in some reduced and highly specialized shoots, as in some grass inflorescences. In these shoots, the apical meristem has been transformed into a leaf primordium, and the leaf is, ontogenetically, strictly terminal. Stages in the assumption of a pseudoterminal leaf position in determinate shoots are seen in both dicotyledons and monocotyledons. In sympodial growth of woody twigs, the stem apex aborts and the dead tip may be abscised, as is a leaf, either in a terminal position or after crowding to one side into an apparently lateral position (Tilia, Cladrastis). In the monocotyledons, evidence of the true position of an apparently terminal leaf may be obscure or lacking. The genera Streptopus, Disporum, PolygorMtum, and Uvularia show various stages in the loss of the stem tip. A vestigial apex may be enclosed in the sheath of the distal leaf. In Uvularia, one species shows no trace of a stem tip; another, a vestigial cone; another, an obvious stem tip. The "terminal" position of leaves in herbaceous monocotyledons is obviously secondary; the shoot apex is abortive or lost, and a leaf assumes its position. Determinate leaves have somewhat different origins in monocotyledons and dicotyledons. Similarly, a soli- 18 MORPHOLOGY OF THE ANGIOSPERMS tary carpel or stamen and a basal ovule (in a syncarpous ovary) may acquire a terminal position in a floral shoot. The "terminal" cotyledon of many monocotyledon embryos is also secondary (Chap. 9). Change of position of these organs, congenitally established, does not make such a leaf, carpel, ovule, or cotyledon morphologically cauline, as has been sometimes claimed. The error of interpretation of basal ovules as cauline because of their apparently terminal position has brought about suggested changes in the classification of major angiosperm taxa that are morphologically unsound: the division of angiosperms into Phyllo- sporeae and Stachyosporeae, and the removal of the Casuarinaceae from the angiosperms. The monocotyledon leaf is characteristically simple; compound forms are rare — Dioscoreaceae, some of the Liliaceae and Araceae. Parallel venation is dominant; the compound leaves and some simple leaves have parallel-reticulate, palmate-reticulate, or pinnate-reticulate vena- tion — Trillium, Smilax, Colocasia, Arisaema, Butomaceae, Pontederiaceae, Alismataceae. The simple leaf is elongate, commonly linear, and con- sists of a sheathing base — the leaf sheath or leafbase — and a distal "limb"; these parts may merge, or be more or less clearly delimited. The resemblance of the simple, linear monocotyledon leaf to the phyllodes of some dicotyledons suggested the plu/Uode theory, the conception that this leaf represents morphologically the basal sheath or petiole of an ancestral leaf. The sheathing base is looked upon as an adaptation in the development of the characteristic habit in mono- cotyledons — leaves close-packed on a greatly shortened stem. In the parallel venation of the "limb" is seen the vascular pattern of a petiole. Morphological support for this theory is supposedly given by the close resemblance of the monocotyledon leaf to the sheathing bases of some dicotyledon leaves — Umbelliferae — and of the resemblance of the terete "limb" to petioles in such genera as Triglochin and species of Allium, Sagittaria, Sisyrinchium. Vascular anatomy is considered to support the interpretation of the monocotyledonous leaf as homologous with the proximal part of the dicotyledonous leaf. Parallel venation is characteristic of sheathing leafbases and petioles. Species of Sagittaria are believed to show stages in the flattening of terete leaves. The flattened leaves show two series, dorsal and ventral, of vascular bundles oriented with phloem toward the dorsal and ventral surfaces, respec- tively. Bifacial leaves, so formed, show stages in the merging of these series, with, in the completely flattened leaves of some taxa, all the vascular bundles in one plane and alternating irregularly in orientation. This vascular structure is present also in the leaves of the Ponte- deriaceae, Hydrocharitaceae, and other families. An apparently dorsi- ventral leaf is shown by its anatomy to be a modified cylindrical leaf. THE PLANT BODY 19 It has been argued that the concept of the monocotyledon leaf as a phyllode cannot be supported by the presence of inverted bundles, because similar bundles occur in petioles of dicotyledon leaves, but the leaves cited — Erijngium, Ranunculus, Plantago — are also phyllodes — flattened petioles, rachises, or midribs. As interpreted under the phyllode theory, the modification of the monocotyledon leaf by reduction has gone beyond the loss of the blade. The leaves of many genera of the Liliaceae and Amaryllidaceae — Hyacinthus, Doryanthes, Dracaena — have petioles reduced to vestigial, "solid" apices (the leafbase forms the photosynthetic structure). In Hemerocallis and other taxa, even this tip is considered lost, and the leafbase forms the entire leaf. The phyllode theory also holds that, in contrast with this reduction, specialization has brought about the development of a new leaf part, a secondary blade, by the expansion of the tip of the phyllode to form a distal lamina. The palms, Scitamineae, Alismataceae, Smilax, Eichornia, Pontederiu, are cited as examples. In such blades, parallel venation is considered evidence that the "blade" is petiolar in nature, but the vena- tion is commonly intermediate between parallel and reticulate. The theory that the simplicity of the typical monocotyledon leaf is the result of reduction was discussed at the beginning of the twentieth century as a part of the idea that the monocotyledons arose by the "self-adaptation" of dicotyledons "to a moist or aquatic habitat." The blade-bearing monocotyledon leaf has been considered by some to be the primitive, by others, the highly developed type. The simple, ligulate leaf has been called "rudimentary," but it is associated with highly specialized habit, as in the grasses. A reticulate lamina in mono- cotyledon leaves may represent survival of an ancestral character or a secondary structure developed as an elaboration of a part of a parallel- veined leaf. The evidence that the monocotyledon leaf is, in large measure, a phyllode is convincing, and there seems no doubt that the "blade" of such leaves as those of Eichornia and Pontederia represents the modified tip of a phyllodelike leaf. The interpretation of the blade of the palm leaf as an example of a lamina secondarily acquired by modification of the tip of the petiole is not supported by its structure or its ontogeny. Early stages of its ontogeny show that this leaf is a "complete leaf" — sheath, petiole, and blade. The development of petiole and blade follows the pattern of a typical dicotyledon leaf; the primordium of the blade is formed on the leaf buttress. This is followed by development of the petiole from an intercalary meristem at its base. If the blade is to be interpreted as petiolar in nature, the presence of an unusually long petiole and a prominent, well-defined leafbase must be explained. 20 MORPHOLOGY OF THE ANGIOSPERMS There is little evidence to support the selection of an ancestral type for the monocotyledon leaf. On the assumption that the monocotyledons arose from dicotyledonous stock, the primitive monocotyledon leaf prob- ably resembled that of the dicotyledon in gross structure. But, if the angiosperms are polyphyletic, the ancestral monocotyledon leaf may have differed greatly from that of the dicotyledon and comparisons with modern types would therefore be valueless. In the morphological study of the angiosperm leaf, recognition of shoot and root as the basic structural categories in the plant body raises the question of the fundamental nature of the leaf and its relation to the stem. The lower vascular plants suggest that there are two morphologically distinct types of leaves: microphijUs, which are lobes or outgrowths of the stem; and megaphijlls, which are modified minor branch systems. Leaf gaps are as- sociated with the traces of megaphylls but not with those of microphylls. The angiosperms are mega- phyllous. As a part of the development of the telome theory of the nature of the plant body, considerable at- tention has been given to the interpretation of the make-up of the angiosperm leaf and sporophylls. These appendages have been interpreted as systems of dichotomizing axis tips, united laterally. But nodal structure — number, position, and relation of traces and gaps — and ontogeny of the appendages, especially development by apical and marginal meristems, do not support a telomic make-up. All appendages in the angiosperms have long been interpreted as having basically an odd number of traces, but it has been shown recently that, in the primitive node, the appendage has two traces asso- ciated with one gap (Fig. 5). Branches of the shoot also have two traces with a single gap; leaf and branch are alike in the origin of their vascular supply. The nature of the leaf as a basic part of the shoot sys- tem is apparent, and the term "partial shoot" well expresses its funda- mental nature. A two-part trace system throughout the shoot may suggest ancestral dichotomy. Evidence in support of this view has been seen in veinlet branching, cotyledonary lobing, and the forking of some simple stigmas, but phyllotaxy and branching habit do not show dichotomy. (The Fig. 5 showing tion of leaf of letja. Diagram vasculariza- node and Austwbai- Unilacunar nodal structure with two traces which continue as two bundles through petiole and leaf blade. {After Bailey and Swatny.) THE PLANT BODY 21 "dichotomous" branching of a palm is probably not true dichotomy. See Chap. 11.) It has been pointed out in studies of apical meristems that leaf and branch primordia have diflFerent origins; leaf initials arise less deeply in the meristem than do branch initials but this difference is apparently not constant. (Little attention has been given to the origin of branch primordia.) The various interpretations of the make-up of the shoot fall into three morphological categories: that in which the stem is regarded as an axis, with lateral lobes or appendages, tlie leaves; that in which the shoot is considered a multiple structure, made up of segments or units called phytons; and that in which the stem is considered a secondary structure, built up partly or wholly of leaves, which are the fundamental units. On the basis of the second and third interpretations, the phyton theory and the leaf-skin theory of the nature of shoot have been pro- posed. Akin to the interpretation that the leafbase is, in part, cauline, is the theory that the outer tissues — "skin" — of the shoot axis consist of de- current leaf bases. This "skin" has been described for some taxa, both dicotyledons and monocotyledons, and regarded as present in most, probably in all, angiosperms. Formation of the "leaf skin" is considered to be by downward continuation and development of the leafbase, as the internode elongates. Limitation of the "skin," both externally and internally, has been variously interpreted; according to one descrip- tion, this is "a question solely of definition." Obviously, the theory meets major morphological difficulties where structural limits are not set, and it seems of little importance. Under the phyton theory, the shoot consists of "units of growth" that are renewed by a type of terminal "budding." The make-up of these units has been variously defined: as an internode with its attached leaf; a leaf with a root (the internode being the base of the leaf); a segment of the stem, limited by nodes, with or without a leaf; a leaf primordium with its base incorporated in the axis. (A root, as a part of the phyton, is lacking in angiosperms.) Concepts of the shoot as consisting of a series of structural units are old and have been ob- scured by the dominance of the stem-and-leaf theory. Anatomically, units like those described under the leaf-skin and phyton theories do not exist; the shoot is the basic unit. The Anatomy of the Plant Body* Structurally, the basic plant body is that built up by the embryo and its apical meristems. This is the primary bodi/. All fundamental body parts are represented in the primary body. To this primary body is * Anatomy of the flower in detail is discussed in Chap. 3. 22 MORPHOLOGY OF THE ANGIOSPERMS commonly added a secondary body, by the activity of secondary meri- stems of the cambium type. The secondary tissues so formed may ob- scure, distort, or destroy parts of the primary body. The various organs may consist entirely of primary tissues (formed by primary meristems) or partly of primary and partly of secondary tissues. The primary body is complete in itself in so far as the presence of all basic organs is con- cerned — root and shoot (stem and leaves, sterile and fertile). In all these parts, vascular tissues — forming the vascular skeleton — are promi- nent structural features. In the axis (root and shoot), a central core, the stele or central cylinder, is more or less clearly set apart from the surrounding cortical and epidermal tissues. Limiting the stele is the endodermis, a uniseriate layer of specialized cells. Morphologically, the endodermis is probably best considered the outermost layer of the stele. In the angiosperms, as compared with lower vascular plants, the endo- dermis is greatly reduced in distribution and in histological structure. The anatomical structure of the primary body of the angiosperms, especially that of the shoot, is complex. The modifications of high specialization, with reduction, accompanying the herbaceous, aquatic, epiphytic, and parasitic habits, have brought about vascular structure that may be difficult to interpret. A basis for the interpretation of the vascular structure of the stele was provided by the stelar theory, late in the nineteenth century. This theory has been somewhat modified and new terms added, but satis- factory definitions are difficult to make and there has been much loose- ness in use of the terms applied to the stele and its vascular prolonga- tions. The terms protostele and siphonostele were applied to two basic types: protostele, to a central cylinder with a solid (pithless) vascular core; and siphonostele, to a cylinder with a tubular vascular core and central pith. The siphonostele, when broken by openings (gaps) where traces pass out to lateral appendages, is phyllosiphonic; where gaps are formed by branch traces alone, it is cladosiphonic. The term solenostele has been applied to siphonosteles in which the leaf gaps in the inter- node above are closed before the gaps associated with traces next above appear, and dictyostele where the vascular tube is dissected into a meshwork by overlapping leaf gaps. But, in general use, the term dictyostele has been incorrectly applied to angiosperm steles; by origi- nal definition, the vascular bundles of a dictyostele are concentric, and this is a type unknown in angiosperms. The newer term eustele is now in general use for steles with a vascular skeleton of collateral or bi- collateral bundles which anastomose more or less freely. This is the common type in angiosperm shoots. Steles in which phloem is present only external to the xylem are ectophloic; those in which there is also phloem internal to the xylem are amphiphlvic. THE PLANT BODY 23 The ectophloic stele is the common type, but amphiphloic steles are frequent — Apocynaceae, Asclepiadaceae, Campanulaceae Compositae, Convolvulaceae, Cucurbitaceae, Gentianaceae, Myrtaceae. Internal phloem is especially well developed in the Solanaceae and Cucurbita- ceae, where, in herbaceous genera, it may play a prominent part in conduction. The phylogenetic relations of these stelar types in the angiosperms are uncertain. The sometimes apparently vestigial or de- generate condition of the internal phloem and its late ontogenetic de- velopment suggest that the amphiphloic stele may be more primitive than the ectophloic, but the absence of internal phloem in primitive families and its presence in some advanced families does not support this concept. If the angiosperms are polyphyletic, both stelar types may have been present in the stock from which they arose. Anomalous vascular structvire is frequent in angiosperm roots and stems. The peculiarities are various and are chiefly in the secondary tissues and in their method of origin and development. The original cambium of the young stem or root may cease to function and may be replaced by other secondary cambia which arise outside of the phloem. A cambium of normal type and persistence may form xylem and phloem of unusual distribution and arrangement. Excessive multiplication of parenchyma in restricted regions may break the original cylinder into strips. A cambium that develops tissue only centripetally may form complete vascular bundles embedded in "interfascicular tissue." Anoma- lous steles are doubtless in large part adaptations to unusual function, habit, or habitat; they occur in many woody vines and parasites, and in food-storage organs. They may characterize families — Chenopodiaceae, Amaranthaceae. Anatomy of the Root. The root is typically protostelic; in specialized types, a pith may be present. Angiosperm root steles range from mon- arch to polyarch, but are most commonly diarch and tetrarch. The basic types are clearly the diarch and tetrarch, and each of these has been claimed to be the most primitive. The diarch type is characteristic of large groups of herbaceous genera — members of the Ranales, Rhoead- ales, and Urticales, for example. Woody taxa commonly have tetrarch roots. The less common stelar types seem more likely to have been derived, phylogenetically, from the tetrarch than from the diarch. Diarch and monarch are probably reduction types; monarch steles are found in greatly reduced roots. The roots of the monocotyledons are typically polyarch and usually have a pith. It seems probable that the tetrarch stele is the primitive type; it best fits, structurally, into hypo- cotyls and into the pattern of the two-trace node; it is characteristic of most of the Archichlamydeae and is rare in the Metachlamydeae. The polyarch stele with a pith seems to be highly specialized. 24 MORPHOLOGY OF THE ANGIOSPERMS Anatomy of the Shoot. The vascular structure of the shoot axis in the angiosperms is complex and has a pattern that is repeated from node to node. The nodal sections are more complicated than the inter- nodes, because of the presence of divergent vascular strands — leaf traces — which extend from the internodal system to the leaf. Branch traces, always two, arise just above the leaf traces and similarly connect the vascular stele of the branch with the stele of the mother axis. The num- ber of leaf traces ranges from one to many and is characteristic of taxa; it is phylogenetically significant. The term leaf trace was first applied to the traces of a leaf, collec- tively, but is now applied to the individual strand, because number of strands per leaf is important. A leaf trace cannot be delimited rigidly, because it is continuous from stele to leaf. Distally, it is considered to end where it passes into the cortex. In the cortex, it may fork or unite with other traces. (The base of the petiole may show a number differ- ent from that of the stele. ) Proximally, the trace unites with the stelar cylinder or with bundles of the internodal steles at various levels, ac- cording to patterns related to trace number and phyllotaxy. The stele may be siphonostelic, with an essentially unbroken vascular cylinder, or eustelic, with "free" bundles of various sizes. The eustelic type has been incorrectly considered characteristic of herbs, but many herbaceous genera — Nicotiana, Salvia, Aster, Hypericum — have siphonostelic stems. Woody angiosperms have both types of steles. The eustele, with re- duced vascular tissue, is perhaps the advanced stelar form. Areas where cortex and pith are continuous are termed leaf and branch gaps, respectively. Where the primary vascular stele has the form of a more or less unbroken cylinder, breaks are present above and often lateral to the leaf and branch traces where they depart from the cylinder. Each trace may have its own gap, or two or more traces may have a common gap. Nodes are termed unilacunar, trilacunar, and multilacunar, where the number of gaps is one, three, and many, with- out regard to number of traces per gap. In a unilacunar node, more than one trace may be associated with the gap. The primary vascular cylinder is often weakly developed, and the gaps may not be obvious, becoming prominent only after secondary thickening has begun. In typical eusteles, where the cylinder consists of more or less isolated sti'ands, gaps are usually not apparent; they have been considered as merged with the spaces between the bundles. In the floral shoot, nodal anatomy is like that in the vegetative shoot. This similarity is strong evidence that floral appendages are of leaf rank, that they are not organs sui generis, not mere lobes of the axis, not mere spore-bearing areas (Chap. 3). The trilacunar node, with three traces (Fig. 6C), has long been looked upon as the primitive type in angiosperms, a type associated THE PLANT BODY 25 with the leaf type considered primitive— simple and pinnately veined. Support for the primitiveness of the trilacunar node was found in the dominance of this nodal sti-ucture in the Archichlamydeae, with elabora- tion to multilacunar types (Fig. 6D) in the Epacridaceae, Platanaceae, Araliaceae, Umbelliferae, Polygonaceae, and Meliaceae, and to reduc- tion types in the Centrospermae, Myrtiflorae, and most of the Meta- chlamydeae. Some Archichlamydeae show reduction in trace number Fig. 6. Diagrams showing evolutionary development in nodal angiosperm anatomy from a primitive type. A, unilacunar with two traces; B, unilacunar with two traces fused; C, trilacunar, with three traces, the median a double trace; D, multila- cunar with seven traces; E, unilacunar with three traces from one gap; F, five traces from one gap; G, unilacunar with one massive trace of five to seven fused traces. {After Canright.) within the family — Leguminosae, Anacardiaceae; and some Metachla- mydeae, multiplication of traces— Epacridaceae. Further evidence sup- porting the primitiveness of the three-trace node was seen in the pres- ence of these nodes in seedlings of dicotyledons that have many traces in the mature plant. The monocotyledons are commonly multilacunar, but seedlings frequently have three-trace leaves, and their carpels, usu- ally multitrace, may have, as reduction forms, three traces or only one. The two-trace, unilacunar node (Fig. 6A)— a "fourth type" of node —was long overlooked. Certain two-trace organs — cotyledons, stamens in a few taxa, carpels with double midribs, set aside as abnormalities 26 MORPHOLOGY OF THE ANGIOSPERMS or examples of fused organs — are now seen to be persisting examples of early nodal structure in angiosperms. The odd-numbered types, so char- acteristic of angiosperms in general, have been derived by the union of the tv^'O traces of the primitive node, as a part of the specialization of stem and leaf. Fusion of the two traces of a unilacunar node gives a one-trace unilacunar node; the addition of lateral traces in pairs, with their individual gaps gives tri- and multilacunar nodes with odd-num- bered traces (as under the older theory of nodal specialization) (Fig. 6C, D ) . Advanced types of unilacunar nodes with a single trace have devel- oped by the loss of lateral traces and by the lateral union of three or more traces and their gaps (Fig. 6E, F, G). In these nodes, the "compound" trace may be three-parted or massive — Asclepias. The lateral union of three or more traces — the median trace of double nature — is more com- mon in floral shoots, where the ap- pendages are crowded, than in vegetative shoots. The carpels of some dicotyledons may show stages in the fusion of three traces from three gaps to form one trace from one gap — Ruhus spp. Ranunculus Ficaria, Anem- one canadensis. (In these examples, the reduction and simplification of nodal structure accompanies the evo- lutionary modification of a follicle in achene development. The carpels appear to be one-trace organs until the origin of their vascular supply is determined. ) Two-trace unilacunar nodes are known in the stems of the Austro- baileyaceae (Fig. 5), Chloranthaceae, Lactoridaceae, Trimeniaceae, and some Verbenaceae, Labiatae, and Solanaceae. In most of these taxa, the traces are independent well down in the eustele. In cotyledons and sporophylls also, there are frequent examples of paired or fused median traces. Paired traces are probably more numerous in cotyledonary than in vegetative nodes (Figs. 7 and 8). In reproductive nodes, they are doubtless more frequent than is known; tiieir presence cannot be so readily determined as in leafy stems, but two-trace floral organs are occasional. Double median bundles are frequently found in the carpels Fig. 7. Forms of cotyledonary vena- tion showing double traces and ex- tent of independence of the two vas- cular systems within the cotyledon. {After Bailey.) THE PLANT BODY 27 of taxa of widely scattered families — Ericaceae (Clethra, Pyrola, Epigaea), Caryophyllaceae (Anagallis). In some free carpels, the stigmas are two-lobed and the vascular supply forked. The double- bundle was described in many taxa in demonstration of the carpel polymorphism theory. Two-ti-ace stamens are uncommon but are present in both advanced and primitive taxa: Austrohaileija, Sarcandra, Victoria, NupJiar, Casuarina, Cijrtandra, Eranthe- mum, Donjanthes (two pairs), several genera in the Betulaceae, Fagaceae, and Proteaceae. Origin of the single trace in stamens is seen in Hakea, where, in a single species, even a single Hower, some stamens have two independent traces and others have obviously double traces, which may fork distally. Sporophylls with double vascular supplies have doubtless not been recorded because they were con- sidered abnormal. The frequency of occurrence of these sporophylls with two traces cannot be com- pared with that of similar leaves, because trace origins in the flower are often concealed by fusion. The presence of two traces in cotyledons has long been known and at one time aroused interest as possible evidence of ancestral dichotomy persisting in the em- bryo. This formed part of the basis for a "theory of the double leaf trace"— a theory that two-trace appendages characterize most major vascular taxa. This theory received little attention, because of interest in the problem of the phylogenetic relation of the cotyledons of the dicotyledons and the monocotyledons— does the single cotyledon of the monocotyledons, with two traces, represent two fused cotyledons, or one of a pair, the other lost (Chap. 9)? The two-trace cotyledon was considered primitive for the Liliaceae, but from the viewpoint that it represents retention of the vascular supply of the two cotyledons of dicotyledonous ancestors, rather than as basic angiosperm structure. B Fig. 8. Diagrams of cotyledonary nodes and cotyledon. A, B, of Degeneria viti- ensis showing four traces in petiole and venation in A and a trilacunar node with three cotyledons with four traces each from three gaps each; C, of Mag- nolia grandiflora showing a trilacunar node with two cotyledons with four traces each. The four lateral traces "de- rived by the bifurcation of two traces which arise from two gaps in the stele." {After Swamy.) 28 MORPHOLOGY OF THE ANGIOSPERMS Reduction in the cotyledon of the two traces to one — as in stems — has been described in Scilla (Chap. 9). Recognition of the two-trace node as primitive for angiosperms con- stitutes a major forward step in the use of anatomy in interpreting the phylogeny of the angiosperms (although odd-numbered traces will doubtless continue to be described as characterizing angiosperms). In the possession of a nodal trace system based on two units, the angio- sperms join the other megaphyllous taxa. In angiosperms, an cndodermis is present in roots and in the stems of many herbaceous taxa and of seedlings. In these locations it is often apparently vestigial. It consists of a layer of cells resembling endo- dermal cells in form and arrangement but lacking the critical character of cutinized wall areas. The stems of woody plants lack an cndodermis. The presence in the angiosperms of a perici/cle, a sheath of tissue be- tween the vascular tissue and the cndodermis, has been questioned. It has been shown by critical histological studies that, in at least some genera, the fibers between the phloem and the cndodermis, commonly called "pericyclic fibers," belong to the primary phloem; protophloem elements formed among them soon degenerate and disappear as the stele matures. The morphological problem here relates to paren- chyma cells that often lie between these fibers and the cndodermis. These may be part of the protophloem or of the more or less distinct pericycle. The absence of an cndodermis in most angiosperm stems makes the delimiting of the primary phloem, with its parenchyma and fibers, uncertain. A pericycle is characteristic of vascular cryptogams and of the roots of seed plants. In the angiosperm stem, the pericycle, together with the cndodermis, is in process of reduction and loss. The primitive angiosperm leaf, under the interpretation of the primi- tive nodal structure as trilacunar, was seen as palmately veined, per- haps three-lobed, with three major veins united at the base of the blade. The present recognition of the unilacunar, two-trace node as apparently the basic nodal structure also supports the concept of the palmately veined leaf as primitive. Leaves with two traces continued independently through petiole and blade (as a double midvein) are rare — Austrobailcya (Fig. 5), Chloranfhus, Sarcandra, Ascarina. Pairs of traces that continue through the organ are more frequent in cotyle- dons and sporophylls than in leaves (Fig. 7). The more advanced three-trace system is formed by the addition of two lateral traces with separate gaps (Fig. 6C). The three traces — the median double in nature fundamentally — unite at various points: in the cortex, at the base of the petiole, or in the petiole. From this three- trace system, a common type, have been derived other still more specialized types. THE PLANT BODY 29 There is strong correlation between the simple, pinnate leaf and the unilacunar node, but there are also extreme exceptions, as in Eupomatia, where there are 7 to 11 traces from as many gaps. Leaves with sheath- ing bases usually have several traces and gaps. Though the fundamental ; '// %%9f>] *; Fig. 9. Two series of diagrams showing vascularization of node and successively hio-her levels of the petiole of Dcgeneria viticnsi^: series 1 from a seedling leaf and series 2 from a mature leaf. {After Swamy.) vascular relationship of node and leaf now seems clear, variations re- lated in part to leaf form are many. For example, highly complex nodal structure accompanies interpetiolar stipules and other stipules of un- usual form and function. Pinnate venation may have been derived in two ways: directly from a simple, two-trace leaf with a double midvein, by a strengthening of lateral veinlets; from a three-trace leaf with three veins by a weakening 30 MORPHOLOGY OF THE ANGIOSPERMS or loss of the lateral traces and veins. Among leaves of the woody Ranales, which, by association in this order with so many other highly primitive characters, have been looked upon as simple, there are both palmate and pinnate types — palmate in Aiistrobaileya and Tetracentron, pinnate in others. The vascular structure of the petiole varies greatly. The variations are in part related to form and function of the leaf, but the basic plan is dependent upon the number and arrangement of the traces and their freedom or fusion as they enter the petiole (Fig. 9). Within the petiole, the vascular bundles may continue undivided or divide and unite in their course to the blade. Their orientation may remain as it was at entrance to the petiole or may change greatly. The arrangement is in part an adaptation to mechanical support, U-shaped, I-shaped, and hollow-cylinder types are common. The distribution of the bundles may be fairly constant in a family — Ericaceae, Rhamnaceae — but is more commonly characteristic of genera, as in the Proteaceae and Umbelliferae, where it has aided in taxonomic studies. The vascular structure of the petiole, when better known, will be of much help in taxonomy and perhaps in phylogeny. The Ontogeny of the Plant Body Increase in length of the plant body is brought about in the shoot by apical and intercalary meristems; in the root, by apical meristems only. Intercalary meristems are parts of apical meristems separated from apical meristems by areas of mature or maturing tissues. They are in- ternodal in stems and basal, in part, in leaves and floral organs. In angiosperms, the meristems of the plant body are more complex in structure and development than those of lower vascular plants. Between 1920 and 1950, topographical and histological aspects of the apical meristem of the shoot received much attention. Details of structure and development of these meristems were studied in a large number of taxa, and an apparently sound basis was obtained for generalizations on the morphological value and significance of their structure. Ontogeny of the Shoot Apex. The apical-cell theory, which arose in the earlier days of interest in the ontogeny of the plant body, assumed de- velopment from a solitary apical cell or group of apical initials. This theory was replaced in the 1860s by the histogen theory, which holds that development is by meristems that build up individually the various tissues or parts of the axis. The histogen theory dominated in- terpretation of apical meristems for more than fifty years, but in the early decades of the twentieth century, with increasing interest in the anatomy of the shoot, it was found to have little morphological value THE PLANT BODY 31 and was replaced by the tunica-corpus theory (Fig. 10). This new theory stimulated research in shoot apices of many taxa during the following decades. According to the tunica-corpus theory, the shoot apex consists of meristematic tissues arranged more or less distinctly in two major parts 7-K/»^ r TrJ=>^ ^-^ •ry^£^ TJX T-yper iv Type v Txpfr iri T-ype v/r Fi of ig. 10. Diagrams of longitudinal sections of shoot apices showing types of organization in vascular plants. S, surface meristem; M, mantle; MO, central mother cells; C, cambium-like zone; SA, subapical initials; CM, central meristem; P, peripheral meristem. {After Popham.) — a central core, or corpus, sheathed by one or more external layers, the tunica (Fig. 11). In descriptions of the earlier years under this theory, interpretations were made rather rigidly. But it soon became evident that limitation of zones is not always clear and was sometimes "found" in different places by different investigators; that the number of layers in the tunica, considered characteristic of a species, varies with 32 MORPHOLOGY OF THE ANGIOSPERMS position in the plant, vigor of the shoot, and season. To improve in- terpretations, it was suggested that the terms tunica and corpus be replaced by mantle and core — so defined as to avoid the rigidity of the earlier use and to allow considerable variation in some characters. Mantle is defined as consisting of all the distinct peripheral layers of the apex in which anticlinal cell divisions maintain the layers. The core B Fig. 11. Diagrams to illustrate two concepts of growth in dicotyledonous shool apex. A, the tunica corpus theory: zone 1 and 1', initials of tunica layers; zone 2, corpus initials; zone 3, peripheral zone; zone 4, rib meristem. B, the mantle-core concept: zone 1, mantle layers; zone 2, central-mother-cell zone; zone 3, cambium- like zone; zone 4, rib meristem; zone 5, peripheral zone. {After Gifford.) THE PLANT BODY 33 consists of the remaining central tissues in which cell divisions are in various planes and little or no peripheral layering is present. The mantle may add to the core. In somewhat modified definitions, the corpus is the central core of the apex, consisting of larger cells, varying in form, without definite peripheral layering, and with cell divisions in many planes; the tunica consists of one to several outer uniseriate layers of smaller cells, uni- form in size, with divisions wholly or largely anticlinal (Fig. 12). In the corpus, cell divisions are infrequent, and the protoplasts stain lightly; in the tunica, divisions are more frequent and the cytoplasm stains deeply. Tunica and corpus, though not always sharply delimited, are independent, self-perpetuating meristems. The chief variations are in number of layers in the tunica — at first considered constant for a species — and in the clearness of separation of tunica and corpus. The major difficulty in de- limiting tunica and corpus has been the interpretation of the outer layer of the corpus, a layer, with frequent anticlinal divisions, which may be more or less dis- tinct as a uniseriate sheath and appears transitional between corpus and tunica. Below the tunica-corpus zone in the apex, there may be distinguishable a region of transition between the highly meristematic distal cells and the maturing cells, in which enlarging cells of like type stand in long rows or files. This zone has been called a third meristem, a. rib meristem (Fig. 12). A rather broad survey of shoot-apex structure in various tribes of the Rosaceae seems to show that clear tunica-coipus zonation is not a characteristic feature of tliis family. Absence of definite zonation here has been considered a surprising departure from the apex structure described for taxa throughout angiosperms and as perhaps the result of differences in interpretation. But dehmitation of tunica and corpus may be weak or obscure. In tlie monocotyledons, rather few shoot apices have been studied, perhaps because of greater complexity of structure in these plants. The tunica is probably most commonly two-layered. A one-layered tunica is common in monocotyledons and occasional in dicotyledons. A few major taxa have mostly several-layered tunicas — Compositae, Caprifoliaceae, Rosaceae, Guttiferae. Variability in number of layers is more frequent in dicotyledons than in monocotyledons. According to Fig, 12. Zonation diagram of a hypothetical shoot apex showing direction of cell divisions. 1, tu- nica; 2, corpus; 3, rib meristem; 4, flank meristem. (After Stant.) 34 MORPHOLOGY OF THE ANGIOSPERMS the information now available, number of tunica layers seems to have no phylogenetic significance. A comparison of doubtful importance has been made with the reduction in number of layers in the nucellus, where the larger number of cell layers seems to represent primitive structure (Chap. 7). Morphological significance has been seen in manner and place of origin of leaf initials, but a leaf primordium may arise wholly from cells of the tunica or from initials in both tunica and corpus. Angiosperm shoot apices seem to have two types of organization: one, in which the cells of the tunica divide anticlinally except in the center of the apex, where some periclinal divisions occur; the other, in which no periclinal divisions occur. Evolutionary progress in specializa- tion in the zonation of the apex has apparently been from a tunica with layers in which periclinal divisions are general throughout, to those with periclinal divisions restricted to the apex, and to those with anticlinal divisions only. Recognition of timica and corpus as descriptive units has been of much value in studies of growth activities, especially in cytohistology, but use of the terms must be considered primarily topographical. The presence of a fairly well-defined zonation — tunica and corpus — despite inconstancy in number of layers in the timica and perhaps absence of such zonation (as reported for some rosaceous taxa), seems to characterize the angiosperms. The zonation of apical meristems should not be interpreted too rigidly, as it was in the earlier years of its study; the apices must be recognized as dynamic structures, responding within limits to various growth conditions. The shoot apex from which a flower will develop (the reproductive or floral apex) has been claimed to differ fundamentally from the vegetative apex, to be a completely different structure and, like the sporophylls borne upon it, a structure sui generis. But this concept has been shown by several critical studies to be without foundation. In the transformation of a vegetative into a floral apex, the vegetative axis undergoes, as it elongates, a gradual change in zonal pattern. The change is in proportion of zones and is, in large part, the result of the bringing close together of the nodes and the many appendages of the flower. The interpretation of the floral apex as morphologically distinct from the vegetative apex was based largely on the following claims. First, the vegetative apex has a tunica and a corpus; the flower axis has a "parenchymatous core" — not highly meristematic — and a thick, "heavy," meristematic "mantle." But differences in gross structure between the two types of apices are in degree only and are associated with basic structure. In the floral apex, the nodes are crowded together, and the THE PLANT BODY 35 core is "parenchymatous" (not meristematic), because the shoot is de- terminate and cell divisions have largely ceased in the central part. The mantle is wide and complex in structure, because in it are developing the primordia of many appendages. The mantle corresponds to the timica and the outer part of the corpus. Secondly, the floral appendages are claimed to have arisen from more superficial layers than do leaves. This difference does not exist; the wider mantle and the contrast be- tween mantle and core — greater than in vegetative apices, because of more abundant primordia in the outer part and few cell divisions in the center — makes the origin appear to be more superficial. The third claim was that there are no foliar buttresses in the floral apex. But the buttresses develop later and are smaller, because the nodes are crowded. A fourth claim was that development of procambial strands is strictly acropetal in the floral apex, whereas in the vegetative axis development is in both directions from the base of an appendage. But development of the procambium is alike in both kinds of apices. The small size of the appendages at this stage and their crowded arrangement make direction of development difficult to determine. And, finally, it was claimed that carpellary primordia are not crescent-shaped and do not embrace the shoot axis as leaf primordia do. But the carpel primordia of many taxa, those in which carpel closure is postgenital, are crescent- shaped; those in which closure is congenital are not crescent-shaped. The floral apex is morphologically like the vegetative apex; differences are of degree only, associated with differences in function. It is a vegetative axis gradually transformed, not a new structure. Under some growth conditions, the floral axis may be transformed into a vegetative axis, as in terminally proliferated flowers. Ontogeny of the Root Apex. The anatomy of the root apex was well known long before critical studies of the shoot apex were made, doubt- less because of the greater simplicity of the root apex and better dis- tinction of histogenetic zones. In the root apex was found, in large part, the basis for the histogen theory, but greater interest in the more com- plex shoot apex has placed acquaintance with this growing point ahead of that of the root. Early descriptions of the apical meristem of the root set apart a central core, the plerome; an outer sheath, the periblem; and a uniseriate, outer layer, the dermatogen. Distinction of these layers, with implica- tion of restriction in function and morphology in the parts developed by each, formed the histogen theory. Though better acquaintance with anatomy showed that the terms do not have the morphological value assigned to them, they have continued in use as of topographical value. Tunica and corpus are terms not applicable to the root apex, because of its markedly different morphology. The apical meristem of the root is 36 MORPHOLOGY OF THE ANGIOSPERMS shorter than that of the shoot and differs from it in its clearer and more constant zonation, in the definition of layers close to the apex, in the presence of a root cap, and in the absence of primordia of ap- pendages. The meristematic layers of the root apex of angiosperms are usually formed by three, rarely four, groups of initiating cells (Fig. 13). In the dicotyledons, the distal group forms the cap and the dermatogen; the median group, the periblem; the innermost, the plerome. In the mono- cotyledons, the distal group forms the cap; the median, the dermatogen Fig. 13. Diagrams of root-apex types in angiosperms. A, initials in three groups, cap not distinct, formed by same initials as dermatogen; B, initials in three groups, cap distinct in structure and independ- ent in origin. {After Eamcs and Mac- Daniels. ) Fig. 14. Median longitudinal section of the basal end of a mature embryo of Tropacolum majus, showing initial cell (Z) and suspensors attached at the base. (After von Guttenhcrg.) and periblem; the innermost, the plerome. The outstanding characteristic of the apex of the dicotyledon root is the common origin of cap and dermatogen, a resemblance to the ancient type of root-apex origin, where both cap and epidermis are formed by a solitary apical cell (Fig. 14). The root apex, lacking the complexities of appendage development, is simple, though the cap adds a different "appendage." Its simplicity of structure suggests that it is more primitive than the shoot apex. The simplicity of the root apex adds to the evidence (lack of appendages, protostelic structure, and exarch xylem ) that the root is a more primitive organ than the shoot. THE PLANT BODY 37 The root cap is an important morphological feature of the root. It is absent in only a few monocotyledons, where it seems to have been lost as a part of adaptation to an aquatic habitat. Ontogeny of the Leaf. The development of the leaf follows patterns related as a whole to the type of leaf that is formed. Variety in details of meristem activity and tissue specialization is great, and only general features of form and anatomy are discussed here. In the dicotyledons, the leaf primordium is initiated close to the shoot apex in die tunica or in an area involving cells of both tunica m n nr Fig. 15. Diagrams showing types of marginal growth in the leaf blade. E, epidermal initial; S, subepidermal initial; P, procambium. {After Hara.) and outer corpus. A lateral shoulder, or "buttress," increases in size by apical growth and by lateral expansion around the apical meristem. The lateral extent of the more or less crescent-shaped mound so formed varies with form of leaf and with stipule form and position. Median growth builds up a somewhat flattened, fingerlike projection, on which marginal meristems soon appear. Apical growth continues, but increase in size is soon largely a result of the activity of the marginal meristems (Figs. 15 and 16), which early lay down the foundation of the blade, often outlining its general form very early. Below the marginal meristems a basal region, which shows less activity, later becomes an "intercalary" or "basal" meristem, which builds the petiole. If the leaf is stipulate, stipule meristems arise below this meristem on the shoulders of the buttress, or, where adnate to the petiole, on the petiole meristem when 38 MORPHOLOGY OF THE ANGIOSPERMS A-- it becomes active. The petiolar meristem may remain inactive for a long period, or may continue cell division, with little cell enlargement. The petiole may develop either from a transverse meristematic plate, or, more commonly, by the rapid enlargement of small cells. The midrib region of the leaf is formed by the central part of the early elongate primordium, the remainder of the blade by specializa- tion of the tissues formed by the marginal meristems. The history of the various tissue layers of the blade is complex but follows a general pattern. The morphological nature of the basal shoulder of the primordium — leaf buttress — on which the leaf arises is discussed earlier in this chapter. Ontogeny shows that it exists as a topographical feature of the shoot development, that it is a part of the shoot meristem not referable to the mature leaf or axis. In the mature shoot, it is a region where leaf and stem merge. Fig. 16. Transverse sections of edges of Early stages in the development young laminae in progressive stages of r ^ i ^^ iiir development, showing subepidermal cell ^^ ^^e compound dicotyledon leaf divisions that bring about marginal are the same as those of the simple growth in the young lamina A, initial leaf. Evidence that the leaf will be cell; B, daughter cell of A by vertical -, . i r. i-i division; A\ A\ daughter cells of A by compound appears m the fingerlike horizontal division; B\ B', daughter cells stage of the primordium, with the of B by vertical division; u ep, upper development of leaflet primordia epidermis; ti m, upper mesophyll; m m, ■, \, • i . i r. middle mesophyll; I m, lower mesophyll; ^^^^g ^^^^ marginal ridges. Se- l ep, lower epidermis. {After Avery.) quence in development in the leaflet primordia varies with the taxon— acropetal, basipetal, or "divergent." Apparently little is known of the origin of these new apical and marginal meristems from the margins of the mother primordium— whether they appear before or after the mother marginal meristems are established. Leaflet develop- ment follows the same course as that in the simple leaf. Ontogenetic fusion between leaflet primordia apparently may occur, as between floral-organ primordia, but is infrequent or rare and little is known about it. (Leaflet fusion in Bauhmia and related genera is apparently largely congenital. ) The history of development of the leaf of the monocotyledons differs only in detail from that of the dicotyledons. The leaf buttress is less prominent than in dicotyledons and may be absent. (In many mono- THE PLANT BODY 39 cotyledons, the leaves are borne more closely than in dicotyledons, and buttresses are inconspicuous or late in development, as in floral axes where the organs are crowded.) Buttress and primordium may be in- distinguishable. The primordium has a broader base than in the dicot- yledons; the crescent margins extend far, or completely (in some sheathing leaves), around the axis — an obvious relation to the sheathing bases of so many monocotyledons. Marginal meristems are absent. The activities of an apical meristem are brief, and the leaf is built up chiefly B C Fig. 17. Diagrammatic sketches, showing position and relationship of leaf parts in primordia of the two leaf types in palms. A, B, pinnate leaf, B, showing detail of furrows and ridges of pinnae primordia; C, palmate leaf; h, hook; n, rein; p, petiole; r, rachis; vc, ventral crest (hastula). {From Eames, 1953.) by an intercalary meristem. Correlations can perhaps be found here with the supposed petiolar or leafbase nature of this leaf (Chap. 12). Compound leaves are rare in the monocotyledons. Details of develop- ment of compound monocotyledon leaves are little known. The ontogeny of the compound palm leaf is described here because it is probably a type unique among angiosperms. In the palms, the simple leaf, present in seedlings of most taxa and mature plants of a few, is primitive; the compound leaf has been derived from the simple. The derivation has been described as ontogenetic — by a folding of the blade brought about by alternate dorsal and ventral invaginations and a later splitting along the lines of the folds. The dis- section is indeed ontogenetic, but the process is much more complex than this. The leaf is a "complete leaf," with sheathing base, petiole, and blade. The primordium is similar to that of other monocotyledonous 40 MORPHOLOGY OF THE ANGIOSPERMS leaves. But very early, while the primordium is only 1 or 2 mm long, a pattern of dissection is laid down which later determines the lines of folding and splitting of the blade into "leaflets" (Fig. 17). The pattern of folding is laid down on both dorsal and ventral surfaces of the blade primordium by a series of shallow ridges and furrows ( Fig. 18). Simultaneously, minute, needle-shaped openings, formed by sepa- ration of cells, appear in close rows in the tissue below each ridge, dorsal and ventral (Fig. ISA). These openings, at first only 10 to 20 jx in diameter, do not extend to the surfaces of the blade; later, they increase in length, extending to the furrows opposite them but not to the ridges Fig. 18. Diagrams of cross sections of parts of leaf-blade primordium of palms, based on Roijstonea, showing successive stages in origin of "folds" from which pinnae arise. A, early stage of differentiation, acicular slits below each ridge; B, somewhat advanced stage, the slits extended to the furrows opposite them; C, later stage, blade increased in thickness and in area, procambium of midveins of pinnae developed, beginnings of abscission-tissue development in dorsal ridges which sep- arate pinnae, d, dorsal; p, pinna; v, ventral. {From Eames, 1953.) (Fig. 18B). Rows of these needlelike perforations then unite, splitting the ridges longitudinally from below, and forming a series of low, com- pressed "folds" (Fig. ISC). Increase in tissues of the blade in area and in thickness builds up the ridges (folds), which alternate on the upper and lower sides and form the typical plicate structure (Fig. 19) of the immature and, in many palmate types, the mature leaf. Lateral separa- tion into leaflets comes about as the leaf unfolds from the bud, through division by abscission layers or by "disorganization." The folding is in- itiated by dissection, and the folds are built up by differential growth — "invagination," in one sense. The folds are not formed by compression within the sheaths of outer leaves, though, in some genera, there may be a crinkling in other planes. This is only the first part of the story of the formation of the com- THE PLANT BODY 41 pound leaf in palms. The major veins of the blade extend pinnately or palmately from a midrib or basal area, distally running parallel with the margin, and the lowermost veins form a narrow, strongly vascular band along the margin of the blade (Fig. 20). The early ridges and later folds do not extend into this band. When the folds are separated, Fig. 19. Sketches of transverse sections of young leaves showing pinnate and palmate development, reins R at margins of blades. A, Roijstonia regki, a pinnate type, rachis below developing pinnae; B, Livistona chincmis, a palmate type, hastula above united with dorsal crest (lobe of rachis) below. {From Eames, 1953.) Fig. 20. Diagrams showing details of reins and hook in palm leaves. A, showing lines of abscission and relation of reins and hook to blade; B, showing course of vascular bundles and lines of abscission between pinnae and between pinnae and rein, h, hook; r, rein. (From Eames, 1953.) 42 MORPHOLOGY OF THE ANGIOSPERMS as the leaf opens, the leaflets or pinnae so formed are at first held to- gether at their tips by the marginal band but are soon freed by the abscission of the band, together with the apex of the blade where the veins meet. From each leaf hang two straplike structures, the reins, with the apex, the hook, attached to one of them. These structures are, morphologically, a part of the blade, and their presence is related to the dissection of a lamina in which outermost veins of a pinnate-parallel series enclose the inner, distal veins. By the formation of the reins and hook and their abscission, leaflets are cut out of the central part of the blade. The leaflets have no normal margins or apices and differ in this way from other leaflets. The reins and hook vary greatly in structure, often persisting as tough, fibrous strips (sometimes green), suspended from the base of the blade for months; in palmate-leaved genera, they may be threadlike and ephemeral. In the palms, the palmate leaf has clearly been derived from the pin- nate (see Palmae, Chap. 11). That the compound leaf of the palms was derived from the simple was accepted by morphologists in the middle of the nineteenth century, but its method of origin was not understood for one hundred years. There are two wholly different ontogenetic origins for the compound leaf. The palm type is known only in the palms and their close relatives, the Cyclanthaceae. (Plicate leaves resembling those of palms are present in Curculigo and species of Pennisetum, but the folding in these genera is the result of differential growth, with no dissection in- volved.) In the ontogeny of the compound leaf in all other angiosperms, so far as now known, the leaflets arise as lateral structures of the elon- gating rachis primordium, much as the simple leaf arises (Fig. 2). The existence of two types of compound-leaf origin strengthens the theory that the basic lines of angiosperm stock had diverged far from one another in early angiosperm history. The ontogeny of modified leaves — btid scales, cataphijlls — has received some attention as a part of the investigations of apical meristems and leaf development. The studies have been largely morphogenetic and have raised the question whether bud scales represent modified or transformed foliage leaves or are, ancestrally, appendages of different rank. Bud scales are like leaves in vascular supply and major features of ontogeny — apical and marginal meristems, axillary buds — and many taxa show stages transitional to the foliage leaf; they are surely of leaf rank and are homologous with the entire leaf or a part of the leaf. The ontogeny of floral appendages is like that of leaves at many and at all important stages. Differences believed to exist, which have been used to demonstrate that the floral shoot apex is mor- phologically unlike that of the vegetative apex, are those of degree THE PLANT BODY 43 only, or do not exist. (These diflFerences have already been discussed earlier in this chapter.) It has been reported that floral organs are initiated in the inner tunica layers only, but carpel initials in some taxa arise in part in the outer corpus, as do some leaf initials. In stamens, marginal meristems are present in those that have laminar form and in some of those with broad filaments. Ontogeny strongly supports, as does vascular anatomy, the homology of all floral appendages with leaves. Secondary Vascular Tissues and the Cambium Xylem. The xylem of angiosperms ranges widely in structure from simple to highly complex and to a secondary simplicity de- rived by reduction. The evolutionary story of the two cell systems in xylem — vertical and transverse — and their cell types has been worked out in detail. In the most primitive woods, the vertical system con- sists of tracheids and a small amount of wood parenchijrrm; the trans- verse system consists of parenchyma cells in the form of wood rays. In specialization, the tracheids gave rise to vessel elements and various types of fibers, and the wood parenchyma was increased in amount and changed in distribution. In the transverse system, specialization changes were in form, size, and structure of the rays. The primitive angiosperm tracheid is well shown in form and structure by vesselless genera and those with very primitive vessels. This tracheid is very long, with long, tapering, overlapping ends and numerous pits. The pits of these tracheids are all scalariform — Trochodendron, Eupo- matla. Such tracheids are remarkably fernlike. In other primitive genera, some of the pits of the tracheids are round. Where both forms of bordered pit are present, as in the Winteraceae, the scalariform type occurs in the early-formed part of the annual ring. Scalariform pitting, the undoubtedly primitive type of tracheary pitting in angiosperms, is widely distributed among the less specialized families but is absent in the higher families. The most primitive type of vessel in the angiosperms differs from the primitive angiosperm tracheid only in the absence of closing membranes in some of the scalariform pits of the end walls. In these vessel ele- ments, the number of perforated pits is high — in Etipomatia, from twenty to one hundred. They are arranged in ladderlike series along the flattened, oblique end walls (Fig. 21A). In evolutionary modification, this primitive vessel element became progressively shortened; accom- panying the shortening, the advanced vessel element became rounded in cross section, with the wall usually thinner than that of the primitive element and often irregularly thick. The characteristic long-tapering ends became less acute, until the end walls stood at right angles to the 0O <^ 0© 0^0 « © e\ 9 «e '^^ 90 ® e «e ''el Fig. 21. Vessel elements in side view and cross section. A, B, Bctitla alba; C, D, Liriodcndron; E, F, Lobelia cardinalis; G, H, Quercus alba; I, J, Pijriis Malus; K, Acer Ncgiindo; L, M, Quercus alba; N, end of vessel element from Lobelia showing perforation indicating derivation of porous vessel from scalariform; O, ends of vessel elements from Lobeliu showing method of union of elements in a series. {From Fames and MacDaniels.) 44 THE PLANT BODY 45 side walls (Fig. 21L). The perforations were reduced in number, en- larged, and fused, by loss of the ladderlike bars between them, to form a single opening. Vessel elements with one large pore in each end wall are termed simply perforate (Fig. 21G to L); elements of less spe- cialized type, with two or more perforations on the end walls, are termed scalariform or midtiperforafe (Fig. 21A to C). Modification in form of vessel element is phylogenetic, but perforation of the wall is ontogenetic. In primitive, vessel-bearing woods, the vessels are solitary or chiefly so (Fig. 22); in advanced types, they tend to be aggregated. Where the vessels, solitary or clustered, are scattered through the annual ring — as seen in cross section — the wood is disuse-porous; where the vessels, especially the largest ones, are chiefly in the early wood of the annual rings, the wood is ring-porous. Ring porosity is an advanced character associated usually with simply perforate vessels. Scalariform vessels occur in about 110 families — exclusively in some, dominantly in others, infrequently or sporadically in others. Except for the Paeoniaceae, whicli includes both herbaceous and woody species, families in which scalariform vessels occur exclusively are all woody — Eupomatiaceae, Himantandraceae, Annonaceae, for example. Families in which part of the vessels are scalariform are scattered tlirough the dicotyledons — Cornaceae, Betulaceae, Ericaceae, Bixaceae, Magnolia- ceae, Theaceae. Scalariform vessels are most common in families recog- nized as primitive in other characters, especially those that seem to be primitive members of more or less isolated lines — Magnoliaceae, Betula- ceae, Ericaceae. They sometimes accompany advanced flower structure — Betulaceae, Caprifoliaceae. Simple vessels are also broadly distributed among angiosperms. The vessels of herbs, even in primitive families like the Ranunculaceae, are simply porous. (Paeonia, with scalariform ves- sels, has been removed from the Ranunculaceae.) Vessels are absent in some aquatics — Ceratophyllaceae, Nymphae- aceae, Podostemonaceae — parasites, saprophytes — Monotropa; in these taxa, absence clearly represents phylogenetic loss. Throughout angio- sperms, in highly specialized secondary wood, vessel number tends to be greatly reduced. Some annual dicotyledonous herbs have few or no vessels, as do some larger plants of unusual habit, such as the Cacta- ceae and larger Crassulaceae, which possess strong vascular cylinders. The vessel has arisen independently in most of the major lines of vascular plants: Selaginellales — SelagineUa; Equisetales — Equisetum; Gnetales (sensu stricto) — Gnetum; Welwitschiales — Welwitschia; Ephe- drales — Ephedra; Dicotyledoneae — apparently at least several times; Monocotyledoneae — many times. The vessel cannot be considered a distinguishing character of the angiosperms. That some angiosperms are vesselless was known as early as 1842; in 1900, the "Homoxyleae," a Fig. 22. Secondary xylem of Cercidiphtilhim japonicum. A, transverse section of mature wood; B, part of a vessel member showing the spiral thickenings in the tapering end; C, tangential section through late-summer wood; D, tangential sec- tion through early-summer wood. {From Swamy and Bailey.) 46 THE PLANT BODY 47 taxon consisting of the family Winteraceae and the genera Tetracentron and Trochodendron, were described. Other vesselless genera — Sarcandra in the Chloranthaceae (Fig. 23) and Aynborella (Fig. 24), perhaps con- stituting a family close to the Monimiaceae — have been added to the Fig. 23. Transverse section of part of a two-year-old stem of Sarcandra showing vesselless xylem with multiseriate and uniseriate rays in the first annual ring. B, a highly magnified part of A to show detail. ( From Swamy and Bailey. ) vesselless group. But the group is not phylogenetically related; they constitute surviving remnants of several ancestral stocks. Critical and extensive study of the distribution of vessels in the plant body shows that in the dicotyledons vessels apparently appeared first 48 MORPHOLOGY OF THE ANGIOSPERMS in the secondary xylem, then, progressively, in metaxylem and pro- toxylem. Advance in speciaUzation in wood has "worked backward" from later to earlier and earlier formed tissue. In the monocotyledons, vessels clearly arose first in the later-formed metaxylem, then in earlier- formed metaxylem and protoxylem. Vessels did not appear simul- taneously throughout the plant body. In the monocotyledons at least, they arose first in the roots and then developed successively in higher Fig. 24. Transverse (A) and tangential (B) sections of part of a stem of Amhorella trichopoda showing vesselless xylem with uniseriate, biseriate, and triseriate rays. (From Bailey and Swamij.) and higher parts of the plant. All steps in this phylogenetic "spread" of the vessel through the plant body are found. In the primitive Alismata- ceae, they are present only in the roots; in the highly advanced Grami- neae, they are present throughout the plant. The origin of the vessel in the monocotyledons is perhaps comparatively recent. Vessels in angiosperms arose without question from scalariformly THE PLANT BODY 49 pitted tracheids. Such an origin makes impossible the derivation of angio- sperms from the Gnetales (sensu lato) or from other higher gymno- sperms that have tracheids with only round-bordered pits. The evolution of the vessel provides strong evidence that the angiosperms arose from fernlike stock. (The resemblance of the vessels of Eupomatia to those of Ptcridium is remarkable.) Evidence from vessels supports the theory that dicotyledonous herbs arose from woody ancestors; the vessels of herbs are more highly spe- cialized than those of trees. The obviously recent origin of vessels in monocotyledonous herbs makes impossible the origin of monocotyledons from dicotyledonous herbs, as expressed in some theories of the rela- tionship between monocotyledons and dicotyledons. The primitive pitting in intertracheary walls is scalariform-bordered. From this primitive pitting has been derived, by reduction in size, by division, and by change in form, the circular-bordered pit. Division of the elongate pit formed a row of shorter rectangular pits; these later became rounded. In a few of the woody dicot)'ledons, as in Eiipomatia, all the interhacheary pits are scalariform; in others, as in the Win- teraceae, scalariform pits are present only in the early wood, with transi- tional types and round pits in the late wood. The round pits, at first arranged in transverse rows — opposite — become, in higher types, spirally or irregularly arranged — alternate. In the various types of fibers — which are derivatives of tracheids — the pits are modifications of the round type, arranged irregularly. In wood fibers of various types — structural modifications of the primi- tive tracheids — the ti'acheid has been reduced in diameter and in num- ber and size of its pits, and its wall has been much thickened. No evidence of the primitive scalariform pits persists; all the pits are reduced forms of the circular pit. In length, there is little change. (The fibers of advanced types of wood appear unusually long in contrast with the shortened vessel elements and the great increase in length in development from the short cambium cell of this wood.) Several types of wood fibers are recognized, distinguished on the basis of degree of modification from the ancestral tracheid. Cells in which the wall is greatly thickened and the lumen nearly occluded, with pits small and reduced in number, are typical fibers. Cells inter- mediate between tracheids and fibers in thickness of wall and size and number of pits — the pits with narrower borders and slitlike apertures — are fiber tracheids. Libriform fibers are fibers in which the walls are very thick and the pits very small, often essentially simple. No line separates these types, which represent stages in the elaboration of tracheids as supporting cells, with the conducting function lost. Sub- stitute fibers are fiberlike cells with protoplasts. That they are tracheids 50 MORPHOLOGY OF THE ANGIOSPERMS morphologically is doubtful. The term fusiform wood-parenchyma cells perhaps best suggests their nature. The more highly specialized types of fibers occur in wood that is in other ways highly specialized — often much simplified — especially that of woody herbs, subshrubs, and shrubs. In herbaceous stems with strong cylinders of secondary wood, as in many Compositae, the wood consists largely of fibers. The cells of wood parenchyma stand in vertical rows that extend for long distances in the secondary xylem. The rows are basically uniseriate, consisting of elongate, mostly rectangular cells placed end to end, but two or more rows may stand together, forming clusters or bands of various sizes. Wood-parenchyma cells are formed by the transverse di- vision of daughter cells of fusiform cambium initials, each daughter cell forming a vertical series of a few cells. Each series unites, end to end, with a similar series above and below to form long, continuous columns. Elongate fiberlike parenchyma cells, formed directly by fusi- form initials, are called fusiform wood-parenchyma cells. These living cells, with fiberlike shape and wall, were formerly called "substitute fibers." Wood-parenchyma distribution is termed di§use where single columns are isolated and scattered among the other cells of vertical series; terminal where they are present as isolated strands or tangential clusters at the end of a season's growth; metatracheal where the strands are aggregated in clusters or bands but not including scattered, isolated strands and are wholly or largely free of contact with vessels; paratra- cheal where aggregates of columns are associated, wholly or largely, with vessels and typical tracheids; vasicentric where the aggregates are restricted to sheaths about the vessels. No sharp distinction can be made between the diffuse and metatracheal types, and the term apotracheal has been proposed to include paratracheal and vasicentric types. In wood-parenchyma distribution, the diffuse type seems to be the most primitive, and the vasicentric the most advanced. Terminal distri- bution appears to be a reduction type derived from diffuse and, there- fore, an advanced type. Wood parenchyma is usually absent in stems of herbs that have discrete vascular bundles and in some vines and subshrubs with anoma- lous vascular steles where the bundles are separated by wide primary rays. In some shrubs, wood parenchyma may increase greatly in amount where the rays are reduced in the later-formed wood. In wood where rays are reduced and disappearing, the ray cells are upright, and the disappearing ray is apparently transformed into wood parenchyma by the elongation of the ray initials in the cambium. In highly specialized woods, shorter and wider wood-parenchyma THE PLANT BODY 51 cells accompany shorter fusiform cambium cells. This shortening is closely correlated with a similar shortening in the vessel element. The types of wood-parenchyma distribution merge to some extent and may differ in seedlings and mature plants. The presence of both apotracheal and paratracheal parenchyma in the same wood is rare; it is known in only six families — among them Tiliaceae, Myrtaceae, Bombacaceae. Within a family, the amount of wood parenchyma may be fairly con- stant or may range from little to much, but type of distribution is fairly constant except in families with considerable range in floral struc- ture. There is little information about the occurrence of fusiform wood- parenchyma cells ("substitute fibers"). They are found in the first annual rings of some genera, especially in tropical families, in woody vines, and in suffrutescent herbs. They are, perhaps, most abundant in woody plants that appear to have been derived secondarily from herbs, as in the Berberidaceae; in these plants, strong-walled, elongate paren- chyma cells may furnish the support given by fibers in typical woody plants. In contrast with the vertical system, the transverse system of cells in the secondary wood of angiosperms — xijlem rays or wood rays — is made up of a single basic cell type, the parenchyma cell. But the rays, radial sheets of these cells, though simple in cellular make-up, show much diversity in form and cell arrangement. As diey elongate, they may increase or decrease in size; they may divide or fuse with other rays; they may have different form close to the primary body than farther out in tlie secondary wood. In tangential section, rays have various shapes — linear, oblong, fusiform, elliptical, oval; the larger types — as seen in section — may have parallel sides with extended, uniseriate wings at top and bottom. Two major types of wood rays are distinguished: uniseriate and multiseriate, made up, respectively, of one and of more than one series of cells (Figs. 22 and 25). Uniseriate rays consist of one type of cell only; multiseriate rays may be homogeneous, consisting of one type of cells only, or heterogeneous, consisting of two types of cells: upright, vertically elongate cells, which form an outer, limiting, enclosing layer, and prostrate inner cells, with their longest diameter usually radial. These two kinds of cells seem to differ cytologically also. The term aggregate is applied to clusters of rays that lie more closely together than other rays in the same wood; "aggregate rays" are not a kind of ray. Clustering of rays is characteristic of genera in widely scattered families. The term primary ray is applied to rays that are connected with the pith. Fig. 25. Transverse and tangential sections of wood showing uniseriate and multi- 3, 6, Belliolum haplopus. ( From Bailey. ) 52 seriate rays in the Winteraceae. 1, 4, Zygogynum Vieillardi; 2, 5, Drimys Winteri; 53 54 MORPHOLOGY OF THE ANGIOSPERMS Ray systems commonly show no definite pattern or uniformity of arrangement. Nodal regions may be exceptions where primary rays are related to leaf traces and gaps and, in some highly specialized woods, where the rays, of fairly uniform size and shape, lying in transverse tiers, are storied. Storied rays occur in scattered families: Bignoniaceae, Compositae, Leguminosae, Meliaceae, Tiliaceae, Ulmaceae, and a few others. They may characterize an entire family or only part of the genera. The early history of the ray in vascular plants is unknown. Multi- seriate form goes back to Devonian taxa. In angiosperms, the primitive ray system consists of both uniseriate and multiseriate types, extending from the margin of the primary xylem. The multiseriate are heteroge- neous, and the uniseriate, high celled. Although rays frequently increase in size as they elongate, there seems to be no evidence that, phylo- genetically, the multiseriate ray is an enlarged uniseriate ray. Change in size and form of the ray as secondary growth continues is common in some taxa, rare or absent in others. Phylogenetic specialization of the ray system seems to be by simpli- fication; a system made up of only one type of ray is developed by suppression of the other type. In this way, systems consisting of only uniseriate or only multiseriate rays have been developed. In multi- seriate rays, the trend in modification may be from homogeneous to heterogeneous or vice versa. Within the angiosperms, a system con- sisting wholly of uniseriate rays seems to be a highly specialized type, which has developed independently in a few taxa — Salicales, Sapindales, Castanea. Rayless secondary xylem doubtless represents the greatest modifica- tion in the transverse conducting system. No rays are present in the secondary wood of many herbs, even among those with well-developed woody cylinders. Little is known in detail of the histology of the wood of herbs and subshrubs, but it is apparent that, in these plants, rays may be lost by ti-ansformation into wood parenchyma. (The rays, usu- ally of upright cells, become merged with surrounding wood paren- chyma.) But rays are apparently lost also by dropping out, onto- genetically or phylogenetically; some herbs that have secondary wood made up chiefly of fibers are rayless. Rayless secondary wood occurs in woody herbs and subshrubs in widely scattered families — Geraniaceae, Tremandraceae, Crassulaceae, Empetraceae, Caprifoliaceae. The loss of wood rays in herbs is associated with reduction in cambial activity and the consequent drop in demand for radial conduction. Histologically, distal termination of rays comes about by transformation in the cambium of ray initials into fusiform initials. As the axis increases THE PLANT BODY 55 in diameter, new cells are added by change in function of cambium initials. The ray cells of vascular plants were probably derived originally from tracheids— the conifers show the origin of prostrate ray tracheids from erect tracheids— but the origin of ray cells must go far back in the history of vascular plants. The high, upright cells of uniseriate rays appear to be specialized types, which approach wood-parenchyma cells in form and perhaps in function. The xylem of angiosperms, both as a whole and by its constituent cell types, provides excellent series in evolutionary modification. Ad- vance to greater complexity and reduction to simpler structure are clearly shown. Specialization in wood structure is correlated with ad- vance in flower structure; the wood of sympetalous families is more highly specialized than that of polypetalous families. Broadly considered, the various types of wood cells advance in specialization together: the higher types of vessels usually accompany the highest types of fibers; scalariform vessels commonly accompany liber tracheids and are not present with libriform fibers. But there are conspicuous exceptions; for example, fiber tracheids accompany simply perforate vessels in Quercus. SpeciaHzation in the ray appears to be chiefly in simplification and reduction. The vesselless woods have both uniseriate and multiseriate rays, and many advanced genera have only one type. Heterogeneous rays, present in some of the more primitive genera, are present in lower seed plants — Pteridospermae, Bennettitales, Cycadales. In wood structure, generic differences are usually evident; specific differences, rarely; well-marked differences may exist between sub- genera or sections, as in Quercus. (Some anatomists believe that, in structure, the wood may be as conservative, or even more conservative, than the flower.) Within families, wood structure ranges from remark- ably uniform to only fairly uniform and even diverse; in orders, there may be uniformity or an almost complete lack of resemblance (prob- ably evidence that the order is unnatural under the present interpreta- tion). Comparisons of advance in wood structure with specialization in flowers frequently show high correlation in groups of genera commonly assumed to be closely related, but there are examples of major dis- crepancies in large families and orders. Broadly considered, the struc- ture of wood is one of the most important characters in the determina- tion of natural relationships. Phloem. The phloem of angiosperms has not been studied as thoroughly as the xylem, but the story of its evolutionary modifica- tion is known. It closely parallels that of the xylem. The fundamental 56 MORPHOLOGY OF THE ANGIOSPERMS conducting cell, the sieve cell, is elongate, and has walls with minute perforations, through which extend, in restricted areas, cytoplasmic strands connecting the protoplast with the protoplasts of adjacent simi- lar cells. With specialization, the sieve cell structurally parallels the vessel element; it becomes progressively shorter, with end walls more and more oblique, until they stand finally at right angles to the side walls. Where sharper delimitation sets apart a group of sieve areas, sieve plates are formed. These plates indicate lines or courses of con- duction in the sieve cell, a first step in the formation of a linear group of these cells, a sieve tube. ( Sieve cells that have sieve plates and unite to form a sieve tube are termed sieve-tube elements.) With increased specialization, the sieve plates become more and more prominent, their pores larger and fewer, and the plates become restricted to the end walls of the cells. Where the end wall is transverse, the plate occupies most of the wall and the pores are few and large. A cylindrical cell of this type is the highest form of sieve-tube element, resembling in form the most highly specialized vessel element. With the elaboration of sieve plates, the other sieve areas remain weakly developed or vestigial — "ghost plates" or lattices. Advanced woody families, herbs, vines, and plants of unusual habit usually have sieve-tube elements of the highest type. The companion cell, a new cell type in the phloem of vascular plants, is characteristic of angiosperms. No companion cells have been found in Austrobaileija (Fig. 26). Companion cells are parenchyma cells of spe- cial structure and function, closely related in position and function to a sieve element. They lie alongside the sieve elements and are pitted only with those cells — evidence that their function is closely related to that of the sieve elements. Companion cells are characteristic of angio- sperms but absent in other seed plants. Usually one or more companion cells accompany each sieve element; rarely, a sieve-tube element has no companion cells. The number of companion cells appears to increase with increase in specialization of the phloem, and the phloem of herbs and many woody taxa may have numerous companion cells. In general, highly specialized phloem ac- companies highly specialized xylem. The correlation in advancement between xylem and phloem is often closer than that between flower and vascular tissues. Phloem is usually a complex tissue, including phloem parenchyma and sclerenchyma of one or more types, in addition to sieve tubes and companion cells. In the more primitive woody angiosperms, the phloem tends to be simpler than in the more advanced families and has fewer companion cells and much less sclerenchyma. The phloem of some of the woody Ranales is "soft," not "stratified" with alternate bands of THE PLANT BODY 57 sclerenchyma and sieve tubes. Sclerenchyma, where present in phloem of simpler structure, consists chiefly of sclereids; fibers are few or absent. The phloem of Drimijs is entirely "soft"; that of Zijgogi/num, in the same family, has a few isolated fibers. Amborella has sclereids but Fig. 26. Transverse section of part of stem of Austrobailt'i/a scandens showing secondary phloem with no companion cells. {From Bailey and Sivamy.) no fibers. Calycanfhus, on the other hand, has stratified phloem and sieve tubes of high type. In herbs and subshrubs where there is little phloem, the phloem is simple, usually without sclerenchyma — except as a sheath or cap — and consists sometimes of sieve tubes and companion cells only. The story of evolutionary modification seems to be one of increase in complexity from a simple type, consisting wholly or largely 58 MORPHOLOGY OF THE ANGIOSPERMS of "soft cells," to one that has also sclerenchyma cells of one or more types variously arranged, and to a secondarily simple type, consisting largely or wholly of sieve tubes and companion cells. Internal (or intraxijlanj) phloem, as well as that external to the pri- mary xylem, is present in a considerable number of the more highly specialized angiosperm families — the Myrtaceae, Solanaceae, Gentiana- ceae, Asclepiadaceae, Onagraceae, Convolvulaceae, Campanulaceae, Apocynaceae, Cucurbitaceae. This phloem, although always scanty, dif- fers considerably in amount, and it lies within the primary xylem. Although shown to function effectively in girdled tomato plants, it has been called vestigial, perhaps because the sieve tubes are in small isolated clusters. Phylogenetically, the presence of internal phloem is puzzling. In ferns, amphiphloic structure is often prominent, and the internal tissue is as well developed as the external. The absence of internal phloem from the more primitive angiosperm families — some has been reported for the Rosaceae — seems to make unlikely a derivation from amphi- phloic ancestors. Its strong development in herbaceous genera in fami- lies with many woody genera can perhaps be related to the presence of the small amount of secondary phloem. Its absence in the most primi- tive woody families seems unlikely to be the result of loss. Its wide- spread presence must be evidence that it existed in ancestral taxa, and the suggestion has been made that this is evidence that the angiosperms are polyphyletic and one of the ancestral taxa was amphiphloic. But the families with internal phloem do not seem to be related, even distantly. The presence of internal phloem — a major anatomical character — must play a prominent part in the search for the ancestors of the angiosperms. Strands of secondary phloem embedded in secondary xylem are termed interxijlanj . The nature of this phloem is chiefly of ontogenetic and histological interest, but its presence within the xylem has been critically studied in only a few taxa. There are two methods by which phloem becomes interxylary. In some genera, like Entada and Com- bretiim, some strips of the cambium cylinder form phloem cells to the inside for a brief period, then return to normal activity; the strips of phloem are thus embedded in xylem, which was formed normally. By the other method, sti-ands of phloem are normally formed by the cambium, but the initials that formed them cease to function, and new arcs of cambium develop external to the phloem strands and form xylem, which encloses the phloem strands, as in Sfnjchnos. Phloem of peculiar origin is present in the complex tissues formed inwardly by secondary or accessory cambia, as in the Chenopodiaceae. In this family, the cambium forms both xylem and phloem to the in- side, mostly in cell aggregates that suggest vascular bundles. THE PLANT BODY 59 The Cambium. In the axis of typical woody plants, the cambium forms a continuous cylinder, broken only by branch and leaf gaps; in some herbs and subshrubs, the secondary vascular tissue of the stem is small in amount; it consists of a cylindrical network of more or less discrete strands — the stele is a eustele (Fig. 27B). Complete cylinders of secondary vascular tissues are present in many herbs (Fig. 27A), both annual and perennial; the description of herbs in general as char- acterized by dissected steles is erroneous. The network of bundles rep- resents a reduced and dissected vascular cylinder. Proof of this in- terpretation lies in part in the cambium. In the areas between the B Fig. 27. Transverse sections of herbaceous stems showing two dicotyledonous types. A, Digitalis, stele continuous; B, Artemisia, stele dissected. (From Eames and MacDaniels after Sinnott and Bailey. ) vascular bundles, lines of tangential divisions in parenchyma cells con- nect the strips of cambium in the bundles. All stages in the loss of cambium between the bundles are present in related species and even in the same shoot, which may have the cambium cylinder complete at the base and vascular bundles free at the top, with, in between, strips of vestigial cambium, which are progressively weaker toward the top. The evolutionary history of the cambium in the angiosperms is one of reduction in activity and in area. All stages of loss are present in dicotyledons; in monocotyledons, the vascular bundles typically have no cambium, but, in many genera, some of the bundles have a weak 60 MORPHOLOGY OF THE ANGIOSPERMS cambium, which functions only briefly. In some genera, "secondary" cambia — cambia arising late outside the primary vascular tissue — may build up the vascular cylinder, as in many arborescent monocots and some families of herbaceous and subshrubby dicotyledons, such as the Chenopodiaceae. These accessory cambia may continue to function permanently or may soon be replaced by similar, later-formed meristems that arise successively in outer tissues. Many woody vines — lianas espe- cially — increase in diameter in this way. Histologically, the angiosperm cambium is a sheet of initiating cells of two shapes, fusiform and rectangular or rounded. The fusiform cells build the vertical system; the rectangular, the radial systems of the xylem and phloem. The length of the fusiform cambium initials is closely correlated with the length of the cells formed by them; cells derived from long initials generally increase little in length as they mature. Cells derived from short initials also increase little and, in the higher types of vessel element, the "body" of the cell may be even shorter than the initial from which it arose, with a "tail" showing the length of the initial. Fibers may increase greatly in length beyond the length of the mother initials. In the early decades of the twentieth century, the possibility of the penetration of the ends of maturing vascular cells between the walls of adjacent cells as they elongate — "intrusive growth," "gliding growth" — received considerable attention. This growth was considered impossible by some anatomists, partly on the basis that protoplasmic connections would be ruptured; but intensive studies of the cambium and of cam- bium-cell length have shown that intrusive growth is a common feature of wood ontogeny. The basic pattern of cellular sti'ucture in xylem and phloem is present in the cambium, and change in structure of these tissues is initiated there. Changes in type, number, and arrangement of cells formed as secondary growth continues are based on changes in the cambium initials. New cells are added with increase in area of the cambium. Fusiform initials form additional similar cells by radial division or by a transverse or oblique division followed by elongation. New ray initials are formed by transverse divisions of fusiform initials. The fusiform cells of the more primitive dicotyledons are long and long-tapering, as are the tracheids, scalariform vessels, and sieve elements formed by them. Specialization in the cambium has been a progressive shortening of the fusiform initials and the development of uniformity in ray-initial clusters. The cambium initials of the vesselless taxa are very long and long-tapering; a general shortening of the initials accompanies the presence of vessels. Within an individual plant, as secondary growth continues, the fusiform initials tend to increase slightly in length for BIBLIOGRAPHY 61 some years, although vessel elements and wood parenchyma cells may become shorter. BIBLIOGRAPHY Root and Stem Andrews, E. C; The development of the natural order Myrtaceae, Proc. Linn. Soc. N.S.W., 38: 529-568, 1913. : The development and distribution of the natural order of Leguminosae, Jour. Roy. Soc. N.S.W., 48: 333-407, 1914. Arber, A.: The tree habit in angiosperms: its origin and meaning. New Fhijt., 27: 69-84, 1928. ■ — : Root and shoot in the angiosperms: a study of morphological categories, New Phijt., 29: 297-315, 1930. The interpretation of root and shoot in the angiosperms, Biol. Rev., 16: 81-105, 1941. Bailey, I. W.: Nodal anatomy in retrospect. Jour. Arnold Arb., 37: 269-287, 1956. Ball, E.: The development of the shoot apex and of the primary thickening meristem in Phoenix canariensis Chaub., with comparisons with Washingtonia fHifera Wats, and Trachycarpus excelsa Wendl., Am. Jour. Bot., 28: 820-832, 1941. Bancroft, H. The arborescent habit in angiosperms: A review, New Phyt., 29: 153- 169, 227-275, 1930. Bews, J. W.: "Plant Forms and Their Evolution in South Africa," London, 1925. Boke, N. H.: Development of the adult shoot apex and floral initiation in Vinca rosea L., Am. Jour. Bot., 34: 433-439, 1947. Bower, F. O.: Size, a neglected factor in stelar morphology, Proc. Soc. Edinburgh, 41: 1-25, 1921. Canright, J. E.: The comparative morphology and relationships of the Magnoliaceae. IV. Wood and nodal anatomy. Jour. Arnold Arb., 36: 119-140, 1955. Catesson, A.-M.: Structure, evolution et fonctionnement du point vegetatif d'une monocotyledone: Luzula pcdcmontana Boiss. et Reut. (Joncacees), Ann. Sci. Nat. Bot., 11 ser., 14: 253-291, 1953. Celakovsky, L. J.: Die Gliederung der Kaulome, Bot. Zeit., 59: 77-114, 1901. Comer, E. J. H.: The durian theory or the origin of the modern tree, Ann. Bot., 13: 367-414, 1949. : The durian theory extended. I. Phytomorph., 3: 46.5-476, 1953; II. Phy- tomorph., 4: 263-274, 1954. Cotton, A. D.: The megaphytic habit in the tree Senecios and other genera, Proc. Linn. Soc. London, 156: 158-168, 1943. Davy, J. B.: The sulfrutescent habit as an adaptation to environment. Jour. EcoL, 10: 211-219, 1922. Fames, A. J., and L. H. MacDaniels: "Introduction to Plant Anatomy," 2d ed.. New York, 1947. Engard, J. A.: Organogenesis in Rubus, Univ. Hawaii Res. Publ, 21: 1-234, 1944. Esau, K.: Vascular differentiation in the pear root, Hilgardia, 15: 299-324, 1934. — : Origin and development of the primary vascular tissues, Bot. Rev., 9: 125- 206, 1943. Fahn, A., and I. W. Bailey: The nodal anatomy and the primary vascular cylinder of the Calycanthaceae, Jour. Arnold Arb., 38: 107-117, 1957. Foster, A. S.: The shoot apex in angiosperms, Bot. Rev., 20: 477-529, 1954. Gilford, E. M., Jr.: The structure and development of the shoot apex in certain woody Ranales, Am. Jour. Bot., 37: 595-611, 1950. 62 MORPHOLOGY OF THE ANGIOSPERMS -: The shoot apex in angiosperms, Bot. Rev., 20: 477-529, 1954. Griffith, A. M., and M. E. MaUius: The unit of shoot growth in dicotyledons, Proc. Leeds Phil. Soc, 2: 125-139, 1930. Guttenberg, H. von: Studien iiber die Entwickking des Wurzelvegetationspunktes der Dicotyledonen, Planta, 35: 366-396, 1947. JefiFrey, E. C., and R. E. Torrey: Transitional herbaceous dicotyledons, Ann. Bot., 35: 227-248, 1921. Kumazawa, M.: The medullary system in the Ranunculaceae, Lardizabalaceae, and Berberidaceae, Jap. Jour. Bot., 8: 19-46, 1936. Majumdar, G. P.: The shoot of higher plants: Its morphology and phylogeny. Jour. Asiatic Soc. Letters and Set., 23: 39-62, 1957. Markfeldt, O.: Ueber das Verhalten der Blattspurstrange immergriiner Pflanzen beim Dickenwachstum des Stammes oder Zweiges, Flora, 68: 3.3—39, 81-90, 99-113, 1885. Marsden, M. P. F., and I. W. Bailey: A fourth type of nodal anatomy in dicotyledons, illustrated by Clerodendron trichotomum Thunb., Jour. Arnold Arb., 36: 1-51, 1955. Metcalfe, C. R.: The systematic anatomy of the vegetative organs of the angiosperms, Biol. Rev., 21: 159-172, 1946. Nast, C. G.: The comparative morphology of the Winteraceae. VI. Vascular anatomy of the flowering shoot. Jour. Arnold Arb., 25: 454^66, 1944. Parkin, J.: The durian theory: a criticism, Phijtomorph., 3: 80-88, 1953. Perry, B. A.: Chromosome number and phylogeny of relationship in the Euphor- biaceae. Am. Jour. Bot., 30: 527-543, 1943. Petersen, H. E.: Some preliminary remarks on the origin of isolated vascular bundles in herbaceous dicotyledonous plants, Dansk Bot. Tidskr., 37: 136-147, 1920. Philipson, W. R.: The ontogeny of the shoot apex in dicotyledons, Biol. Rev., 24: 21-50, 1949. : Organization of the shoot apex in dicotyledons, Phytomorph., 4: 70-76, 1954. Popham, R. A.: Principal types of vegetative shoot apex organization in vascular plants, Ohio Jour. Sci., 51: 249-270, 1951. : "Developmental Plant Anatomy," Columbus, Ohio, 1952. Posthumus, O.: On some principles of stelar morphology, Rec. Trav. Bot. Neerl, 21: 111-295, 1924. Reeve, R. M.: The "tunica-corpus" concept and development of shoot apices in cer- tain dicotyledons. Am. Jour. Bot., 35: 65-75, 1948. Riidiger, W.: Die Sprossvegetationspunkte einiger Monokotylen, Beitr. Biol. Pfl., 26: 401-443, 1939. Sargant, E.: A theory of the origin of monocotyledons, founded on the structure of their seedlings, Ann. Bot., 17: 1-92, 1903. Saunders, E. R.: The leaf-skin theory of the stem, Ann. Bot., 36: 135-165, 1922. Schade, C, and H. von Guttenberg: Uber die Entwicklung des Wurzelvegetation- spunktes der Monocotyledonen, Planta, 40: 170-198, 1951. Schoute, J. C: Die Stammesbildung der Monokotylen, Flora, 92: 32-48, 1903. : On phytonism, Rcc. Trav. Bot. Necrl, 28: 82-96, 1931. Sifton, H. B.: Developmental morphology of vascular plants. New Phyt., 43: 87-129, 1944. Sinnott, E. W.: Investigations on the phylogeny of the angiosperms. I. The anatomy of the node as an aid in the classification of angiosperms, Am. Jour. Bot., 1: 303-322, 1914. : The evolution of herbs. Science, 44: 291-298, 1916. BIBLIOGRAPHY 63 and I. W. Bailey: Investigations on the phylogeny of the angiosperms. IV. The origin and dispersal of herbaceous angiosperms, Ann. Bot., 28: 547-600, 1914. Stant, M. Y.: The shoot apex of some monocotyledons. I. Structure and development, Ann. Bot., n.s. 16: 115-128, 1952. Stebbins, G. L., Jr.: Cytological characteristics associated with growth habits in the dicotyledons. Am. Jour. Bot., 25: 189-198, 1938. Thomas, E. N.: A theory of the double leaf-traces founded on seedhng structure, New Phyt., 6: 77-91, 1907. Tippo, O.: The role of wood anatomy in phylogeny. Am. Midi. Nat., 36: 362-372, 1946. Van Tieghem, P.: Sur les diverses sortes de meristeles corticales de le tige, Ann. Sci. Nat. Bot., 9 ser., 1: 33-44, 1905. Wagner, N.: Uber die Entwicklungsmechanik der Wurzelhaube und des Wurzelrip- penmeristems, Phnta, 30: 21-66, 1939. Wardlaw, C. W.: Comparative observations on the shoot ape.x of vascular plants, New Phyt., 52: 195-209, 1953. : On the organization and reactivity of the shoot apex in vascular plants. Am. Jour. Bot., 44: 176-185, 1957. Williams, W. C: The structure of the meristematic root tip and origin of the primary tissues in the roots of vascular plants. Am. Jour. Bot., 34: 45.5-462, 1947. Worsdell, W. C: The origin and meaning of medullary ( intraxylary ) phloem in the stem of dicotyledons. I. Cucurbitaceae, Ann. Bot., 29: 567-.590, 1915. Zimmerman, W.: Der Baum in seinem phylogenetischen Werden, Ber. Deutsch. Bot. Ges., 48: 34-49, 1930. Leaf Artschwager, E.: Anatomy of the vegetative organs of sugar cane. Jour. Agr. Res., 30: 197-221, 1925. Avery, G. S.: Structure and development of the tobacco leaf, Am. Jour. Bot., 20: 565-592, 1933. Bailey, I. W., and E. W. Sinnott: Nodal anatomy and the morphology of stipules. Am. Jour. Bot., 1: 441-453, 1914. and B. G. L. Swamy: The morphology and relationships of Austrobaileya, Jour. Arnold Arh., 30: 211-226, 1949. Blaser, H. W.: Studies in the morphology of the Cyperaceae. II. The prophyll, Am. Jour. Bot., 31: 53-64, 1944. Bugnon, P.: Sur les homologies de la feuille chez les Graminees, Bull. Soc. Bot. France, 4 ser.. No. 24, 71: 246-251, 1924. Corner, E. J. H.: The durian theory extended. II. The arillate fruit and the com- pound leaf, Phytomorph., 4: 152-165, 1954. Cross, G. L.: The origin and development of the foliage leaves and stipules of Morns alba. Bull. Torrey Bot. Club, 64: 145-163, 1937. Domin, K.: Ein Beitrag zur Morphologie des Dikotylenblattes, Ceskd Akad. Umene, Prague (Bull. Internat. Compt. Rend. Sci.), 16: 145-170, 1911. : Morpbologische und phylogenetische Studien iiber die Stipularbildungen, Ann. Jard. Bot. Buitenzorg, 24: 117-320, 1911. -: Phylogenetic evolution of the phyllomc, Am. Jour. Bot., 18: 237-242, 1931. Eames, A. J.: Neglected morphology of the palm leaf, Phytomorph., 3: 172-189, 1953. Foster, A. S.: Leaf differentiation in angiosperms, Bot. Rev., 2: 349-372, 1936. 64 MORPHOLOGY OF THE ANGIOSPERMS : A histogenetic study of foliar determination in Carya buckletji var. arkansana. Am. Jour. Bot., 22: 88-147, 1935. Gaisberg, E. von: Zur Deutung der Monokotylenblatter als Phyllodien, unter be- sonderer Beriicksichtigung der Arbeit von A. Arber, Flora, N.F., 15: 177-190, 1922. Gliick, H. : "Die Stipulargebilde der Monokotyledonen," Heidelberg, 1901. : "Blatt-und Bliitenmorphologische Studien," Jena, 1919. Hara, N.: On the types of the marginal growth in dicotyledonous foliage leaves, Bot. Mag. Tokyo, 70: 110-114, 1957. Hare, C. L.: On the taxonomic value of the anatomical structure of the vegetative organs of the dicotyledons. H. The anatomy of the petiole and its taxonomic value, Proc. Linn. Soc. London, 155: 223-229, 1942. Krauss, B.: Anatomy of the vegetative organs of the pineapple, Bot. Gaz., 110: 159-217, 333-404, 550-587, 1948. Lubbock, J.: On the stipules, their form and function. II. Jour. Linn. Soc. Bot., 30: 465-532, 1894. MacDaniels, L. H., and F. F. Cowart: The development and structure of the apple leaf, Cornell Univ. Agr. Exp. Sta. Mem., 258, 1944. Majumdar, G. P.: The complete foliage leaf, Proc. Indian Acad. Sci., B42: 65-72, 1955. Marsden, M. P. F., and I. W. Bailey: See under Root and Stem Bibliography. Mitra, G. C., and G. P. Majumdar: The leaf-base and the internode: Their true morphology, Paleobotanist, 1: 351-367, 1952. Money, L. L., I. W. Bailey, and B. G. L. Swamy: The morphology and relationships of the Monimiaceae, Jour. Arnold Arb., 31: 372-404, 1950. Monoyer, A.: Sur les stipules des Potamogeton, Rpt. Assoc. Fran. Adv. Sci., 1926: 1-3. Petit, L.: Nouvelles recherches sur le petiole des phancrogames, Actes Soc. Linn. Bordeaux, 43: 11-60, 1889. Philipson, W. R.: Development and morphology of the ligule in grasses. New Phyt., 34: 310-325, 1935. Pijl, L. van der: The leaf of Bauhinia, Acta Bot. Neerl, 1: 287-309, 1952. Priestley, J. H.: Gell growth and cell division in the shoot of flowering plants. New Phyt., 28: 54-81, 1929. and L. I. Scott: Leaf venation and leaf trace in the monocotyledon, Proc. Leeds Phil. Soc, 3: 305-324, 1937. Queva, C.: Contributions a I'anatomie des Monocotyledonees. II. Les Uvulariees rhizomatenses, Beih. Bot. Centralbl, 22: 30-77, 1907. Riiter, E.: Uber Vorblattbildung bei Monokotylen, Flora, N.F., 10: 193-261, 1918. Schrodinger, R.: Phylogenetische Ansichten iiber Scheiden- und Stipularbildungen, Verhandl. Zool.-Bot. Ges. Wien, 69: 162-193, 1931. Sinnott, E. W., and I. W. Bailey: Investigations on the phylogeny of angiosperms. V. Foliar evidence as to the ancestry and early climatic environment of the angiosperms. Am. Jour. Bot., 2: 1-22, 1915. Swamy, B. G. L.: Further contributions to the morphology of the Degeneriaceae, Jour. Arnold Arb., 30: 10-38, 1949. and I. W. Bailey: Sarcandra, a vesselless genus of the Chloranthaceae, Jour. Arnold Arb., 31: 117-129, 1950. Troll, W.: Morphologie der schildformigen Blatter, Planta, 17: 231-314, 1932. : Vergleichende Morphologie der Fiederblatter, Nova Acta Leopold., N.F., 2: 315-455, 1935. BIBLIOGRAPHY 65 Tyler, A. A.: The nature and origin of stipules, Ann. N. Y. Acad. ScL, 10: 1-49, 1897. Yarborough, J. A.: Arachis hypogaea: The seedling, its cotyledon, hypocotyl, and roots, Am. Jour. Bot., 36: 722-758, 1949. Xylem, Phloem, and Cambium Arber, A.: Studies on intrafascicular cambium in monocotyledons, Ann. Bot., 33: 459-465, 1919; 36: 251-256, 1922. Bailey, 1. W.: The cambium and its derivative tissues. II. Size variations of cambial initials in gymnosperms and angiosperms. Am. Jour. Bot., 7: 355-367, 1920. : The problem of differentiating and classifying tracheids, fiber-tracheids, and libriform fibers, Trop. Woods, 45: 18-23, 1936. : The comparative morpholog)- of the Winteraceae. III. Wood, Jour. Arnold Arb., 25: 97-103, 1944. •: The potentialities and limitations of wood anatomy in the study of the phylogeny and classification of angiosperms. Jour. Arnold Arb., 38: 211-254, 1957. and C. G. Nast: The comparative morphology of the Winteraceae. VII. Sum- mary and conclusions. Jour. Arnold Arb., 26: 37-47, 1945. and B. G. L. Swamy: Amborclla trichopoda Baill.: A new morphological type of vesselless dicotyledon. Jour. Arnold Arb., 29: 245-254, 1948. and W. W. Tupper: Size variation in tracheary cells. I. A comparison between secondary xylem of vascular cryptogams, gymnosperms, and angiosperms, Proc. Am. Assoc. Arts and ScL, 54: 149-204, 1918. Bancroft, H.: The arborescent habit in angiosperms: A review. New Phtjt., 29: 153- 169, 227-275, 1930. Barghoom, E. S., Jr.: The ontogenetic development and phylogenetic significance of rays in the xylem of dicotyledons. I. The primitive ray structure, Am. Jour. Bot., 27: 918-928, 1940; II. Modifications of the multiseriate and uniseriate rays, Am. Jour. Bot., 28: 273-282, 1941; III. The elimination of rays. Bull. Torrey Bot. Club, 68: 317-325, 1941. Barkley, G.: Secondary stelar structures of Yucca, Bot. Gaz. 78: 43.3-442, 1924. Boureau, E.: Sur certaines especes homoxylees, vivantes et fossiles, a ponctuations areolees scalariformes de la Nouvelle-Caledonie, 8° Congr. Bot. Rapp., 2: 231- 232, 1954. Chalk, L.: The phylogenetic value of certain anatomical features of the dicotyledon- ous woods, Ann. Bot., n.s., 1: 409-428, 1937. Cheadle, V. I.: Secondary growth by means of a thickening ring in certain mono- cotyledons, Bot. Gaz., 98: 535-555, 1937. : The occurrence of types of vessels in the various organs of the plant in the Monocotyledoneae, Am. Jour. Bot., 29: 441-450, 1942. : The origin and certain trends of specialization of the vessel in the Mono- cotyledoneae, Am. Jour. Bot., 30: 11-17, 1943. : Vessel specialization in the late metaxylem of the various organs in the Monocotyledoneae, Am. Jour. Bot., 30: 484-490, 1943. : Specialization of vessels within the xylem of each organ in the Mono- cotyledoneae, Am. Jour. Bot., 31: 81-92, 1944. : Independent origin of vessels in the monocotyledons and dicotyledons, Phytomorph., 3: 23-44, 1953. : Research on xylem and phloem: Progress in fifty years. Am. Jour. Bot., 43: 719-731, 1956. 66 MORPHOLOGY OF THE ANGIOSPERMS and K. Esau: Secondary phloem of the Calycanthaceae, Univ. Calif. Fuhl. Bot., 20: 397-510, 1958. and N. B. Whitford: Observations on the phloem in the Monocotyledoneae. I. The occurrence and phylogenetic specialization in structure of the sieve tubes in the metaphloem. Am. Jour. Bot., 28: 623-627, 1941. Church, A. H.: Thalassiophyta and the subaerial transmigration, Bot. Mem. Oxford Univ. Press, 1919. Eames, A. J., and L. H. MacDaniels: See under Root and Stem Bibliography. Eckardt, T.: Kritische Untersuchungen iiber das primare Dickenwachstum bei Monocotylen, mit Ausblick auf dessen Verhaltnis zur sekundaren Verdickung, Bot. Archiv, 42: 289-334, 1941. Esau, K., V. I. Cheadle, and E. M. Gilford: Comparative structure and possible trends of specialization of the phloem, Arii. Jour. Bot., 40: 9-19, 1953. Frost, F. H.: Specialization in secondary xylem of dicotyledons. I. Origin of vessels, Bot. Gaz., 89: 67-94, 1930. Gilbert, S. G.: Evolutionary significance of ring porosity in woody angiosperms, Bot. Gaz., 102: 105-120, 1940. Gupta, K. M.: On the wood anatomy and theoretical significance of homoxylous angiosperms, Jour. Indian Bot. Soc, 13: 71-101, 1934. Helm, J.: Das Erstarkungswachstum der Palmen und einiger anderer Monokotylen, zugleich ein Beitrag zur Frage des Erstarkungswachstums der Monokotylen iiberhaupt, Planta, 26: 319-364, 1936. Huber, B.: Die Siebrohrensystem unscrer Biiume vmd seine jahreszeitlichen Veriin- derungen, Jahr. Wiss. Bot., 88: 176-242, 1939. and K. Miigdefrau: Zur Phylogenie des heterogenen Markstrahlbaues, Ber. Deutsch. Bot. Ges., 66: 117-124, 1953. Kraus, G.: Uber Dickenwachstum der Palmenstamme in den Tropen, Ann. Jard. Bot. Buitenzorg, 24: 34-44, 1911. Kribs, D. A.: Salient lines of structural specialization in the wood rays of dicotyle- dons, Bot. Gaz., 96: 547-557, 1935. : Salient lines of structural specialization in the wood parenchyma of dicotyledons. Bull. Torreij Bot. Club, 64: 145-163, 1937. Lemesle, R.: Les divers types de tracheids et leurs principaux modes d'association chez les dicotyledones heteroxylees, Rev. Gen. Bot., 54: (643): 326-340, 1947. Metcalfe, C. R.: The systematic anatomy of the vegetative organs of the angiosperms, Biol. Rev., 21: 159-172, 1946. Mohl, H. von: Uber den Bau des Palmenstammes in "Vermischte Schriften Botan- isches Inhalts," Tiibingen, 1845. Schoute, J. C: Die Stammesbildung der Monokotylen, Flora, 92: 32-48, 1903. : tjber die Verdickungsweisc des Stammes von Tandanus, Ann. Jard. Bot. Buitenzorg, 2 ser., 6: 115^137, 1907. Sifton, H. B.: Developmental morphology of vascular plants, TSIew Phtjt., 43: 111- 129, 1944. Stern, W. L.: Comparative anatomy of xylem and phylogeny of Lauraceae, Prop. Woods, 100: 1-72, 1954. Swamy, B. G. L., and I. W. Bailey: The morphology and relationships of Cercidi- phtjUum, Jour. Arnold Arh., 30: 187-210, 1949. Tippo, O.: See under Root and Stem Bibliography. See also references for families. Chap. 11. Chapter 2 THE INFLORESCENCE Flowers are borne singly and in clusters. Those in clusters, together with the stems and bracts associated with them, form inflorescences; those borne singly are termed solitary flowers. Solitary flowers may terminate shoots of major or of minor rank, or may be axillary. In the terminal position, some solitary flowers may represent a simple and primitive condition; for example, in primitive woody genera such as those of Magnolia, Annona, Eupomatia, Calij- canthus. Others, the great majority, are doubtless surviving members of reduced inflorescences, some derived from terminal, some from axillary clusters. SoHtary axillary flowers are commonly surviving units of axillary clusters. It is difficult to draw a line between a group of solitary flowers and an inflorescence. The limits of inflorescences are often poorly defined; specialized inflorescences of various types pass downward into flowers — solitary or in small clusters — in the axils of foliage leaves — for ex- ample, in Lijsimachia, Campanula. Obviously, a definition for inflores- cence that will satisfactorily describe and limit all flower clusters called inflorescences cannot be made. The inflorescence is not a morphological unit; it is a part of the branching system of the stem with more or less definitely segregated flowering tips. In the long course of specialization of the angiosperms, the inflorescence has been morphologically modified in many ways, with the production of similar form in unrelated lines. Under extreme specialization, some inflorescences no longer resemble branch systems but flowers and fruits — Compositae (Figs. 28 and 37), Proteaceae, Urticaceae. Inflorescences may be grouped in larger in- florescences, and these may be reduced, as in the smaller herbaceous species of Euphorbia. Inflorescences range in size from minute to huge, and in flower num- ber up to millions. The multiflowered panicles of the grasses and the heads (often compound) of some Compositae are well known examples. The terminal inflorescence of Corypha wnhraculifera, a monocarpic palm, is a pyramidal panicle* about 10 m high and about 1 m in diameter at the base, with a flower number estimated at 6,000,000. The " The term panicle is here used in the broader sense, for any large, compound inflorescence, determinate, indeterminate, or mixed in type. 67 68 MORPHOLOGY OF THE ANGIOSPERMS grass trees, Xanthorrhoea (Fig. 29), similarly have millions of flowers in a contracted panicle of cylindrical form, which may be 2 m long. The inflorescence of Tijpha has been estimated to bear 300,000 flowers. Most inflorescences die as soon as the fruit is mature, and many are soon abscised; rarely, they are biennial or perennial, with annual or continuous flowering. Those of some Bromeliaceae produce flowers for two or three seasons, and those of Couroupida, the cannonball tree, A ^ B Fig. 28. Sketch of inflorescence of Galinsoga simulating a flower. A, face view show- ing ray and disc flowers; B, lateral view showing subtending bracts. {Drawing by Elfriede Abbe. ) persist on the tree trunk for many years, leafless but woody, and in- crease in length and diameter like vegetative branches. Classification Many descriptive terms have been applied to the varieties of in- florescences, terms the use and meaning of which have often been loose; transitional types are common and rigid definition cannot be made. Inflorescences have been classified on two bases: that of position on the stem system, and that of sequence in flowering within the cluster. On the basis of position on the stem, they are grouped as terminal, axillary, and intercalary. Strictly terminal clusters terminate branches; axillary clusters are terminal on short axillary branches or represent foliaceous axillary branches which have been reduced to inflorescences; intercalary inflorescences are terminal clusters that have been "left be- hind" by continuing apical growth of the main axis, which forms, alter- nately, fertile and sterile sections, or by sympodial growth. Myrtaceous genera, such as Callistemon, the bottle brush, and Melaleuca, the paper- bark tree, are excellent examples of the intercalary type. Inflorescences THE INFLORESCENCE 69 borne on internodes have sometimes been called intercalary, sometimes adventitious, because believed developed from adventitious buds. Though some adventitious inflorescences may be truly so, the mor- phological position of the clusters is commonly misinterpreted. The Fig. 29. Photograph of part of an inflorescence of Xanthorrhoea hastilis, a condensed cymose panicle, with branches spirally arranged. Left, young stage, with only the first median flower of each cyme open; right, later stage, with secondary flowers also mature. pseudolateral position may be the result of sympodial growth, as in Asclepias, or of adnation of the peduncle to the internode. In cauliflonj, where inflorescences develop on older branches or tree trunks, various relations of the flower cluster to the mother stem may be present. Long- 70 MORPHOLOGY OF THE ANGIOSPERMS dormant axillary buds of dwarfed lateral branches may develop flowers, as in some legumes — Cercis. Cauliflory is frequent in tropical forests but rare in temperate regions. It has been considered, under the durian theory, a primitive character, but the morphological relation of the in- florescences to the trunk or larger branches — often developed from adventitious buds — does not support this view. Cauliflory has been described as related to pollination or seed dissemination by bats and birds. Evidence that pollination in the primitive angiosperms was by insects seems well supported. On the basis of order of development of flowers within the cluster, inflorescences are classified as determinate (ctjmose) and indeterminate (racemose). In determinate inflorescences, development of the first flower limits apical growth of the main axis of the inflorescence, and sequence in development of the other flowers is basipetal; in indetermi- nate inflorescences, the first flower is the lowest, and order of develop- ment of the others is acropetal — the number of flowers is theoretically unlimited. Basipetal order of development is often called centrifugal, and acropetal development, centripetal. It is unsatisfactory to use the order of development of flowers as a basis for distinguishing the major types of inflorescences, and it is mor- phologically inaccurate. Many inflorescences are intermediate in type, especially racemose forms that have a terminal flower, as in Clethra, ]uglans, Digitalis, Convallwia, Campanula, Pyrola. That the line be- tween determinate and indeterminate inflorescences is weak is shown by the large, branched inflorescences, which are often mixed, partly de- terminate, partly indeterminate. Some families — the Campanulaceae, Violaceae — have a great variety of inflorescences. A single genus, for example, Hybanthus, may have both determinate and indeterminate inflorescences. Descriptions and morphological interpretations of all the many varia- tions and modifications of the two major types of inflorescences are not considered here; they are described in elementary textbooks. Some of the variations and modifications that are not readily interpretable or are partly obscured by connation or adnation are discussed below. In the determinate inflorescences, the simple basic type, the dichasium, consists of a terminal flower, with other flowers terminal on two lateral axes borne in the axfls of bracts below the terminal flower. A similar inflorescence, with more than two lateral branches, is a pleiochasium; one with only one lateral branch, a monochasium. All these are termed cymes, although "cyme" is often used as synonynous with dichasium. Different types of monochasia are formed by differences in the position and sequence of the successive branchings; among these are the helicoid and scorpioid cyme and the cincinnus. The morphological nature of THE INFLORESCENCE 71 some examples of these has been misinterpreted, because of complexities of nodal structure. Variations in pedicel length and in orientation of flowers produce forms of inflorescences difficult to place in the usual categories. In all the determinate types, branching may continue suc- cessively, as in the first cluster, and large, compound, cymose in- florescences be built up. In these clusters, the ultimate three-flower unit is often called a cymuJe. This term is applied also to the few units sur- viving in a small cluster reduced from a large one, as in the Betulaceae. The term panicle is applied to large, branched, racemose inflores- cences, but it is more commonly used for large, loose clusters of any type, determinate, indeterminate, or mixed. The inflorescence of the lilac, Sijringa, is racemose, but all its branchlets have a terminal flower, and those that have only three flowers form a typical cyme. The large inflorescences of many grasses — for example, the rice, Oryza — are cymose in major, racemose in minor branching. In large inflorescences, such as those of many rosaceous genera — Crataegus and Ruhus, for example — the order of development may be irregular and inconstant. The umbel may be determinate or indeterminate. The importance of sequence in development has been overemphasized. Some, perhaps many, of the apparently simple inflorescences — cer- tain types of monochasia, even some racemes — are not morphologically simple but represent reduced branched types, as is shown by anatomy and critical comparative studies. Some dichasia become simple by the abortion of the terminal flower; the branching then appears dichoto- mous; compound dichasial inflorescences reduced in this way have been called dichotomous. Branched inflorescences of any type are called compound, as con- trasted with simple, but this is, morphologically, a valueless distinction. Many apparently simple clusters have been derived from branched ones by reduction, but the origin is more or less obscure. For example, the apparently simple "raceme" of Claijfonia is not simple but highly complex in nature. The raceme and the dichasium are commonly called the basic simple types, but die dichasium is a branched, not simple, inflorescence. Reduced Inflorescences In the evolutionary modification of inflorescences, reduction and con- densation have played prominent parts (Fig. 30). Reduction is pri- marily in flower number, but the stem system is commonly also in- volved; internodes are shortened or suppressed, and, in reduction to one flower, nearly all the stem system may be lost. The bracts of the reduced stem system may similarly be greatly reduced, but they may be retained where the internodes are lost and form an involucre about 72 MORPHOLOGY OF THE ANGIOSPERMS the remaining flowers or flower. Involucres of this type may be petaloid and simulate a corolla or perianth — Actinotis ( Umbellif erae ) , Phnelea ( Thymelaeaceae ) , Leucademlron and Protea ( Proteaceae ) , Cornus. Reduction in flower number has taken place in many genera through- out angiosperms. Examples of reduction to one flower from racemose clusters are seen in Droscra, TiiUpa, Plantago, Cypripcdkim, Moncses, Vicia, Vaccinium, Scutellaria, Gentiana, Monotropa, Ornithogalum; from cymose clusters in Narcissus, Viola, Pofentilla, Silcnc, Daliharda, Rosa, Anemone, Philadelphus, Rubus, Vinca. A solitary flower may represent one surviving member of a raceme, as in Vicia (Fig. 31); of an umbel, as in Xanthosia ( Umbelliferae ) ; and of a head, as in Corynibium (Com- 1^ ^ 1 i B i 1 I I Fig. 30. Diagrams showing progressive reduction of lateral branclilets of the inflorescences of coryphoid palms. A, all axes more or less shortened; B, ultimate branchlets (pedicels) shortened, flowers sessile; C, flower-bearing branchlets re- duced to hemispherical bases; D, lateral branchlets wholly reduced, the base elongated, flowers borne in a row along main axis. Reduction in flower number continuous from many to three (triads) or two, sometimes sunken in the fleshy main axis. {After Bosch.) positae). Each flower of a head may be the surviving flower from one of a compacted cluster of heads, as in Echinops, Oedera (Compositae). Where reduction is to a few-flowered cluster, no satisfactory descriptive term may be applicable, and such phrases as "umbellate cluster" (in Rosaceae) and "contracted cymose cluster" (in Labiatae) are used. The morphological nature of the cluster may be obscure, unless com- parisons are made with inflorescences of related taxa. Solitary flowers like that of the quince bush (Cydonia) have been cited as examples of primitive solitary flowers. But the evidence of re- duction through a racemose-umbellate series exists in the related genera, Pyrus, Malus, Chaenomeles. Chaenomeles shows all tlie stages in this reduction, THE INFLORESCENCE 73 Reduction is usually superficially evident from comparison with re- lated taxa and, in extreme reduction, when externally obscure, can often be detected anatomically, as in Acer, where vestiges of the in- florescence axis are present in the clusters of flowers of the soft maples, and in Yicia, where the axis of an axillary raceme is buried in the cortex of the mother stem beyond the attachment of the trace of a Fig. 31. Sketches of flowers and inflorescences of Vicia spp. and diagrams of vascularization of the nodes. A, B, E, showing stages in flower reduction in raceme: E, the surviving flower apparently borne on the main stem; C, D, and F, G, dia- grams in longitudinal and transverse sections showing the anatomy of the nodes of B and E, respectively. In F, the peduncle of the raceme has disappeared ex- ternally, vestiges of its traces are seen buried in the main stem, and traces of the flower are apparently derived independently of the inflorescence stele. proximal flower (Fig. 31), so that the flower of a secondary axis seems to be borne on a primary axis. Similarly, in the well-known proteaceous genera GreviUea and Baaksia, the flowers that appear to be borne on the main (primary) axis of the inflorescence are borne on greatly re- duced, hidden, secondary axes. A different type of reduction is that where the number of flowers re- mains high but the axis system is greatly reduced. In Platanus, the numerous flowers are crowded into heads; those of each major branch 74 MORPHOLOGY OF THE ANGIOSPERMS of the inflorescence form a spherical cluster, with its many flowers com- pacted on the fused branchlets (Fig. 32C). Within the genus, the pistillate heads are reduced from a racemose series to one (Fig. 32E). The surviving head appears terminal but is the proximal member of the series. The phyllotactic spiral of the branchlets making up a head can be seen indistinctly on the proximal surface of the head and is clear anatomically. (The staminate heads are not reduced in number but are ephemeral. ) The oaks {Quercus) show, in the acorn cups and acorns, a similar reduction of the compacted pistillate clusters. The primitive pistillate D Fig. 32. Diagrams to show the evolution of the inflorescence of Platanus. A, B, showing the fundamental racemose character of the inflorescence; C, inflorescence with each lateral branch condensed, forming a head, the lateral branchlets retaining in the head their phyllotactic position; C, D, E, a series showing reduction in number of heads from five to one. (After Boothroijd.) inflorescence of this genus is a large complex panicle, one still present in some tropical species. The ultimate cymose branchlet systems have been reduced to fused woody "cups," surrounding a terminal flower (Fig. 33A). The bracts of the fused branchlets form the scales of the cup. In primitive living species, many acorns are borne laterally on elongate branching axes; in temperate-climate oaks, the cups are re- duced to a few, which are borne on shortened, rarely branching axes (white oaks), and to two or one on a very short axis (black oaks). The family Fagaceae shows also a series in reduction in number of flowers within the surviving fertile cluster — Costanea has five or three flowers; Fagiis, two (Fig. 33B); Quercus, one (Fig. 33A). The lepidocaryoid palms show a similar reduction of the pistillate inflorescence, with the union of the bracts of the sterile branchlets forming an "armored" sheath, the lorica (Fig. 34). The lorica surrounds a surviving terminal THE INFLORESCENCE 75 flower of a lesser branch system (see Palmae). (The scales that make up the lorica have been called emergences of the ovary wall, not bracts, because believed to be borne directly on the wall, but sections of the flowers show the ovary free from the encasing lorica.) Fig. 33. Sketches of longitudinal sections of young female flowers and inflorescence with involucre prominent. A, Qucrcus rubra; B, Fagus americana. {After Langdon.) The multiflowered catkins of the Amentiferae are often described as racemes and spikes, but probably none of them is of so simple a type. The catkins of the Betulaceae are fine examples of reduction from a complex cymose inflorescence. The "florets" represent, in various genera, a cymule of three flowers, with their subtending bracts in various stages of connation and reduction (Fig. 35). In the elms, Ulmtis americana has a compacted panicle; U. thomasi, a raceme; U. piimila, a head — an obvious reduction series. In Acer, A. pseud o- Platomis has a panicle; A. spicatum and A. pennsijlvaniciim, a simple raceme or a raceme with a few weak basal branches; A. saccharum, an umbellate cluster; A. ruhruni and A. saccharinum, few flowered heads with vestiges of the rachis. The solitary flower of Viola is the surviving median flower of a dichasium. V. betonicifolia of Australia and Asia ^.^ ^^ loricate fruit of Metroxy- has flowers in the axils of the pair of /on. (After L. H. Bailey.) % Fig. 35. Diagrams and dorsal views of "flowers," cymules, of Betulaceae showing varied make-up of male flowers in a cymule. A, B, G, H, Carpinus japonica: A, B, H, floral diagrams from terminal portion of ament, from transition region between ament and twig, and from central portion of ament; G, dorsal view. C, Ostrtja virginiana, floral diagram. D, Carpinus laxiflora, floral diagram. E, F, C. caroliniana, dorsal view and floral diagram. I, J, Alnus firma var. hirtcllci, dorsal view, (Bi, Bz, Ba, bracts of inflorescence) and floral diagram. Horizontal bar in E, G, and 1 represents one mm. In dorsal views, all anthers are removed, stamens represented by filaments only. ( After Abbe. ) 76 THE INFLORESCENCE 77 bracts ( "prophylls" ) on the peduncle. Species of the section Nosphinium of Viola — woody shrubs of Hawaii — have a cyme with two to four flowers. In other sections cleistogamous or vestigial flowers are described as present in the axils of the bracts. The related genus, Hybanthus, with various forms of inflorescence, shows similar reduction to solitary flowers. The inflorescence of prominent genera of the Proteaceae — Banksia, Grevillea — shows flowers arranged in pairs in longitudinal rows, each pair subtended by a single bract. Comparative and anatomical study shows that each pair represents two surviving flowers of a lateral racemose cluster which has been shortened until the axis is buried within the cortex of the main axis. Stages in this reduction are present in the more primitive genera, Macadamia and HicksbeacJiia, which have compound racemose inflorescences with two flowers on each lateral axis. Extreme inflorescence reduction accompanying vegetative reduction is seen in annual and perennial herbs with solitary terminal flowers and radical leaves — Cypripcdium, Tulipa. Inflorescences Resembling Flowers. Many greatly reduced inflores- cences resemble flowers; some have passed as flowers in taxonomic treatments. The flowerlike inflorescences of such families as the Com- Fig. 36. Reduced inflorescences. A, Euphorbia piilcherrima, median, longitudinal section of one half an inflorescence, pistillate flower not shown, a, gland; b, stamen; c, bracteole; d, articulation between pedicel and flower consisting of one stamen; e, left secondary branch of dichasium; /, pedicel of flower; g, right secondary branch of dichasium; h, two of three traces in stele of pistillate flower. B, Heliconia bihai L., portion of inflorescence with side of single bract cut away to show, from right to left, two unopened buds, flower in anthesis, and immature and mature fruits on elongated pedicel. Each structure subtended by bract which subtends a raceme of many flowers not shown. Water level indicated by broken lines. (A, after Haber; B, after Skutch.) 78 MORPHOLOGY OF THE ANGIOSPERMS positae, Cornaceae, and Euphorbiaceae with petaloid involucres or modified outer flowers are well known. Similar "false flowers" occur in many other families, as in the genera Leucadendron, Actinotis, Pimelea. In Euphorbia, extreme reduction in flower number and flower structure forms a highly specialized flowerlike inflorescence (Fig. 36A); a naked pistillate flower is surrounded by staminate flowers, each reduced to a single stamen; sterile, fused branches of the inflorescence form glands. The outer, petaloid, ray flowers of the Compositae form a pseudocorolla. Where these flowers are connate, as in the "cup Cosmos" the re- semblance to a true corolla is close (Fig. 37). Fig. 37. Sketches of inflorescence of Cosmos sp., from left to right, a normal in- florescence and two with corollas of ray flowers connate by margins, simulating a sympetalous flower. ( Drawing by Elfricde Abbe. ) Groups of flowers — usually staminate — may be so closely associated that it is difficult to determine the limits of a single flower, and the "flower" description is made up of two or more flowers — Leitneria and several genera of the Betulaceae. In these taxa, the staminate "flower" is made up of a cluster of flowers, each consisting of several stamens, the cluster subtended by an adnate bract. In CercidiphijIIum ( Fig. 146 ) , the staminate flowers, each a cluster of stamens subtended by a petaloid bract, form a compact flowerlike inflorescence. The pistillate inflores- cence consists of carpels arranged like the staminate flowers. That each carpel, subtended by its bract, constitutes a flower is evident from the orientation of the carpels: the ventral margins appear abaxial. Inflores- cences commonly described as flowers are those of Triglochin and Potamogeton. In Triglochin, anatomy demonstrates that the "flower" consists of a whorl of staminate flowers, separated by a whorl of bracts from a whorl of pistillate flowers. The presence of bracts (not stami- nodes) between the stamens and the carpels is in itself sufficient evi- dence that this is not a true flower. The "flower" of Potamogeton, long considered by a few students an inflorescence, is shown bv anatomy to THE INFLORESCENCE 79 consist of four staminate flowers adnate to four calyxlike bracts, which subtend them, and four distal pistillate flowers. Extensive Fusion in Inflorescences. The great variety of form in in- florescences is increased and complicated by connation and adnation in the branch system and among parts of the flowers. Simple adnation of inflorescences to adjacent organs is frequent and often apparent, as in Fig. 38. Sketches of Streptopus showing reduction of inflorescence and adnation of peduncle to stem. A, S. simplex, inflorescence reduced to solitary flower, peduncle adnate to stem for only a short distance above axil of leaf; B, S. roseus, inflorescence as in A, peduncle adnate almost up to leaf above the one in whose axil it is borne; C, S. amplcxifolius, lowest inflorescence showing two flowers with bract indicating a third, and those above showing one flower and a bract indicating another, peduncle adnate to stem up to leaf above, making it appear to be borne below a leaf. ( Drawing by Elfriede Abbe. ) the adnation of the inflorescence of Tilia to its subtending bract and that of the spathe of some aroid inflorescences to the flower cluster — Phijllocarpiis. More obscure relationships involve fusion of the peduncle to the mother axis, as in Streptopus, Sparganhim, and other genera, where the flowers appear to be borne below a leaf rather than in its axil (Fig. 38).- 80 MORPHOLOGY OF THE ANGIOSPERMS The flowers of an inflorescence may be connate laterally either by their basal parts or throughout their length, forming "compound" flowers whose morphological structure is recognizable only anatomically — ^some of the Betulaceae. Among the species of Lonicera with two- flowered inflorescences, there are found stages in the lateral connation of the pairs of flowers (Fig. 39). Some species have the flowers wholly Fig. 39. Sketch and diagrams of Lonicera spp. showing an inflorescence of two flowers fused by their ovaries, and, in the diagrams, the fusion of vascular bundles under connation. A, L. syringantha; B, L. canadensis; C, L. tatarica; D, L. Standishii; E, L. oblongifolia; F, L. caerulea. (A to E after Wilkinson.) free; other species show various degrees of lateral connation of the flowers by their pedicels and ovaries up to complete union of the flowers to the base of the perianth. The various species show well not merely external fusion but fusion between major elements of the vascular skeleton. The free-flowered species show derivation from a cymose inflorescence like that of the related Diewilln. Similar lateral fusion of the several flowers of an inflorescence by their ovaries is seen in Sijncarpea (Myrtaceae); the united, inferior ovaries form a globose THE INFLORESCENCE 81 base, on which are attached the perianths and androecia of all the flowers. In Mitchclla (Rubiaceae), fusion similar to that in Lonicera occurs, but the flowers are solitary and belong far apart, morpho- logically, in the axils of leaves on opposite sides of the stem. The erect flowers are fused laterally to the axis above the node at which they are borne and to one another. The fusion extends to the base of the perianth (sometimes to the top of the corolla) and involves the tip of the modier axis; a falsely terminal, double flower is thus formed. The abortive tip of the stem can sometimes be seen on the double ovary between the perianth tubes. Two solitary axillarv flowers become a falsely terminal inflorescence of two flowers. (Similar conditions are seen in Ephedra, a gymnosperm, in which ovules on opposite sides of a cone axis become fused, forming a falsely terminal ovule.) Critical study of other inflorescences reduced to a single flower shows their apparent position to be false. The solitary axillarv flower of species of Vicia is a lateral flower of a reduced axillary raceme (Fig. 31). The pistillate flower of Carex, apparently terminal on a spikelet, is a lateral flower, which, in the reduction of the spikelet, has become adnate to the vestigial tip of the spikelet. The axis tip is seen in related genera and, within the genus, in all stages of reduction and fusion with the ovary. Fleshiness, associated with fusion in inflorescences, may obscure, in various degrees, form and structure. The entire inflorescence or only parts may be involved in the fleshy transformation; leaflike or platelike structures are formed in this way. Absence of bracts or distortion in their position may increase difficulties of interpretation. In the Urtica- ceae and Moraceae, fleshy inflorescences are common; the "fruit" of the fig tree (Ficus) is a highly specialized fleshy inflorescence. In this genus, the many inflorescence branches ate erect and closely approxi- mated, fleshy, and fused to one another to form a hollow structure which is open distally. The major bracts may enter into the fusion or remain free on the outer surface of the "fruit." The greatly reduced flowers cover the inner surface of the hollow structure. Other genera of the Moraceae and Urticaceae (Dorstcniu, Elatostemon) show phylo- genetic steps in the development of the fig type of inflorescence. Fusion, involving all the parts of an inflorescence and associated with fleshiness, is seen in the fruit of the pineapple plant {Ananas), where the inflorescence axis, bracts, and ovaries of all the flowers be- come intimately fused with one another. Fusion of sterile branches of the inflorescence, with the development of much woody tissue, and the transformation of bracts into scales or spines form the burrs of such genera as Castanea and Fagus and the acorn cup of Quercus (Fig. 33). 82 morphology of the angiosperms Phylogeny of the Inflorescence Different views have been held as to the basic phylogeny of the in- florescence: that the solitary flower is primitive and the inflorescences are buflt up by the addition of flowers from stems below the original flower; that the large compound inflorescence is primitive and that the simpler types and the solitary flower have been derived from this by reduction; and that both these conditions were present in the ancestral stock. The theory that the solitary flower represents the primitive condition is largely based on the presence of the solitary flower in certain woody genera now known to possess many other primitive features — Magnolia, Eupomatia, Degeneria, Cahjcanthiis, Dillenia, Hhnantandra. In support of the theory that the panicle is the primitive flower-bearing form are the dominance of paniculate inflorescences in woody plants in general; the presence of panicles in herbaceous families considered fairly primitive — Ranunculaceae, Rosaceae, Liliaceae; the prevalence of reduction series that lead to solitary flowers throughout angiosperms; and evidence that at least some of the flowers considered examples of primitive flower position are really surviving members of an in- florescence. Within inflorescences, the racemose type has frequently been con- sidered the more primitive, doubtless because development in this type is chiefly acropetal, like that of vegetative stems. But flower develop- ment limits apical growth, and the determinate condition in itself would seem to be primitive. The solitary flower — such as that of Magnolia — which seems to be primitive is determinate, and the majority of large much-branched inflorescences are at least in part determinate. No clear line can be drawn between determinate and indeterminate types; in many genera scattered throughout angiosperms, the racemes have a terminal flower and the acropetal sequence in development may be upset. The racemose arrangement may well be secondary. Changes in the normal ontogenetic sequence that are obviously secondary are the centrifugal development within Hie androecium and the basipetal de- velopment of inflorescences in some palms. The determinate inflores- cence seems to be the primitive type. Evidence is strong that the solitary flowers, which some consider rep- resent the primitive type of flower arrangement, are, in reality, examples of reduction from inflorescences. Part of the basis for the citation of the Magnolia flower as primitively solitary is the general acceptance of this genus as one of the most primitive living angiosperms. But the closely related genus, Michelia, has clusters of large flowers, and Raimondea, in the related Annonaceae, has some flowers in cymes (Fig. THE INFLORESCENCE 83 40), others solitary. The sohtary flowers in Annona have bracts along the peduncle, where abortive flowers may be present. Degeneria and Eu-pomatia laurina also have bracts along the peduncle, and, in Euponiatia, one of these may subtend a second flower. Whatever the flower position in the first angiosperms, it is apparent that reduction in flower number and in inflorescence complexity has taken place along many, often parallel and convergent, lines. This re- duction is made evident by comparative study in many families and is especially clear when found within generic limits, a common condi- tion. If the solitary flower is primitive, there must have been extensive building up to the paniculate inflorescences, followed frequently by Fig. 40. Sketch of Raimondea, Annonaceae, showing inflorescences of large flowers in bud, extra-axillary. {After Safford.) reduction to simple clusters and solitary flowers, which represent a highly specialized condition. Proof of a multiplication of flowers is dif- ficult to obtain. It has been suggested that this multiplication occurred as an accompaniment to the development of microflory, such as that seen in several families in the Australian flora. But the abundance of small flowers in these famifies is doubtless correlated with xerophily, and the many flowers are not borne typically in inflorescences. Simplicity in flower position may represent either a primitive or an advanced condition; it probably represents an advanced condition, the result of reduction. The phylogenetic position of the simple inflorescence is not clear; many of them obviously represent modified compound types. If the large compound type is primitive, as it seems to be, no inflorescence is fundamentally simple; all simple inflorescences are re- duction types. The apparently simple "catkins" of the Amentiferae, the 84 MORPHOLOGY OF THE ANGIOSPERMS umbellate clusters of the Rosaceae, the spikes of the Polygonaceae, and the racemes of the Proteaceae are basically complex. BIBLIOGRAPHY Abbe, E. C: Studies in the phylogeny of the Betulaceae. I. Floral and inflorescence anatomy and morphology, Bot. Gaz., 97: 1-67, 1935. Bailey, I. W., and C. G. Nast: The comparative morphology of the Winteraceae. VII. Summary and conclusions. Jour. Arnold Arh., 26: 2n-Al, 1945. Bailey, L. H.: Palms and their characteristics, Gentes Herb., 3: 1-29, 1933. Becker, W.: Violae asiaticae et australienses, Beih. Bot. Centralhl, 34 (II): 209- 216, 1916. Bergdolt, E.: Morphologische und physiologische Untersuchungen iiber Viola, Bot. Abhandl, 20: 1-120, 1932. Bernbeck, F.r Vergleichende Morphologic der Urticaceen — und Moraceen — Inflores- zenzen, Bot. Ahliandl, 19: 1-100, 1932. Blaser, J. LeC.: The morphology of the flower and inflorescence of Mitchclla repens, Am. Jour. Bot., 41: 53.3-539, 1954. Boothroyd, L. E.: The morphology and anatomy of the inflorescence and flower of the Platanaceae, Am. Jour. Bot., 17: 678-693, 1930. Bosch, E.: Bliitenmorphologische und zytologische Untersuchungen an Palmen, Ber. Schweiz. Bot. Ges., 57: 37-100, 1947. Croizat, L.: The concept of inflorescence. Bull. Torretj Bot. Club, 70: 496-509 1943. Haber, J. M.: The anatomy and morphology of the flower of Euphorbia, Anr\. Bot., 156: 657-707, 1925. : The comparative anatomy and morphology of the flowers and inflorescences of the Proteaceae. I. Some Australian taxa, Phtjtomorph., 9: 325-358, 1959. Hjelmquist, H.: Studies in the morphology and phylogeny of the Amentiferae, Bot. Not., 2 (Supp.): 5-171, 1948. Langdon, L. M.: Ontogenetic and anatomical studies of the flower and fruit of the Fagaceae and Juglandaceae, Bot. Gaz., lOI: 301—327, 1939. Manning, W. E.: The morphology of the flowers of the Juglandaceae. II. The pistil- late flowers and fruits, Am. Jour. Bot., 27: 839-852, 1940. Melchior, H.: Bliitenstandbildung bei der Gattung Viola; ein Beitrag zur Phylogenie der Violaceen, Ber. Detitsch. Bot. Ges., 50: 198-204, 1932. Miiller, F.: "The Fertilization of Flowers by Insects," London, 1883. Nast, C. G., and I. W. Bailey: The comparative morphology of the Winteraceae. VI. Vascular anatomy of the flowering shoot, Jour. Arnold Arb., 25: 454-466, 1944. Parkin, J.: The evolution of the inflorescence, Jour. Linn. Soc, 42: 511-562, 1914. Pauchet, L.: Recherches sur les Cupuliferes, These. Fac. Sci., Paris, 1949. Rickett, H. W.: The classification of inflorescences, Bot. Rev., 10: 187-231, 1944. Safford, W. E.: Raimondia, a new genus of Annonaceae from Colombia, Contr. U.S. Nat. Herb., 16: 217-219, 1913. : Annona sericea and its alHes, Contr. U.S. Nat. Herb., 16: 263-275, 1913. Skutch, A. F.: The aquatic flowers of a terrestrial plant, Heliconia bihai L., Am. Jour. Bot., 20: 535-543, 1933. Smith, A. C.: Taxonomic notes on the old world species of Winteraceae, Jour. Arnold Arb., 24: 119-164, 1943. Snell, R.: Anatomy of the spikelets and flowers of Carex, Kobresia, and Uncinia, Bull. Torretj Bot. Club, 66: 277-295. 1936. BIBLIOGRAPHY 85 Swamy, B. G. L., and I. W. Bailey: The morphology and relationships of the Cercidiphyllaceae, Jour. Arnold Arb., 30: 187-210, 1949. Turpin, P. J. F. : Memoire sur I'inflorescence des Graminees et des Gyperees com- paree avec celle des autres vegetaux sexiferes, Mem. Museum Hist. Nat. Paris, 5: 426-492, 1819. Wilkinson, A. M.: Floral anatomy and morphology of some species of the tribe Lonicereae of the Gaprifoliaceae, Am. Jour. Bot., 35: 261-271, 1948. Zimmermann, W.: Die Phylogenie der Angiospermen-Bliitenstande, Beih. Bot. Cen- tralbl, 53A (I): 95-121, 1935. Chapter 3 THE FLOWER The flower is often considered characteristic of angiosperms; yet no one character can be used to set these plants apart from other seed- bearing plants. Moreover, the flower cannot be so defined as to separate it from similar reproductive structures in the gymnosperms. The mor- phological basis of the flower is rarely emphasized in definitions; often the flower is "the reproductive structure of the angiosperms." Mor- Fig. 41. Diagrams showing primitive structure of flowers of Magnolia acuminata, A to C, and Liriodcndron tulipifcra, D to F. C and F, longitudinal sections; A and E, cross sections through peduncle; B, longitudinal section showing vascular supply to appendages; D, cross section through ^mdroecial region of receptacle showing origin of stamen traces. /;' phologically, it is a determinate stem tip bearing sporophylls and, com- monly, other appendages that are sterile (Fig. 41). But this definition applies equally well to many cones, those of the gymnosperms and even of some of the lower vascular plants. The cone is a characteristic reproductive structure in most vascular plants. If the term flower is to be restricted to the fertile stem tip of the angiosperms and its ap- pendages, it is necessary to compare these structures and their sporangia with those of other groups. 86 THE FLOWER 87 The looseness of use of the term flower by botanists is, in part, re- sponsible for the difficulty in defining the structure. By "the flowering plants" is commonly meant the angiosperms, but such phrases as "the flowering of the conifers" (referring to the period of pollination), "the flowers of the gymnosperms," and "the flowers of the seed plants" are frequently seen. Basic Structure The Receptacle. The flower is, first of all, a stem tip, the receptacle, resembling in ontogeny and fundamental structure a vegetative tip. It consists of nodes and internodes and bears appendages. The nodes are usually closely crowded by shortening and often brought together by suppression of internodes. Apical growth is limited early in develop- ment, but other growth may continue until the fruit is mature. That the flower is only in part a mature structure is commonly overlooked; the receptacle and carpels are still in early stages when the rest of the flower is mature. (Failure to recognize that the vascular supply of the carpels within the receptacle is incomplete at flowering time has led to incorrect interpretations of the nature of the flower and its parts.) The receptacle is often greatly modified and unstemlike in form, size, and structure; and, as it matures with the fruit, it may become still less stemlike. On the receptacle are borne, typically, both fertile and sterile ap- pendages. The shortening and suppression of internodes bring the ap- pendages close together, either in spirals or whorls. Whorls repre- sent the more specialized arrangement, as in leaf arrangement; a whorl represents one or more "turns" of a flattened spiral. Parts of highly compressed spirals often pass as whorls, as they appear to be from naked-eye and hand-lens inspection (and are, for practical pur- poses, in taxonomic description), but they can be seen to be spiral by anatomical study. Each organ stands at a level microscopically different from the others, as in the corolla of Ranunciihis. The appendages may all stand in spirals or all in whorls, or part may be spirally placed and part in whorls. Where all types are spiral, the spiral may run con- tinuously throughout the flower, as in Poeonia; more often, there are discontinuities iDctween the different kinds of organs. Where the ap- pendages are all in whorls, the members of successive whorls usually alternate in position with those of the whorls directly above and be- low; where there are several or many whorls of organs of one kind, as in the androecium of Aqidlegia, the organs may stand in longitudinal rows. Typically, the different kinds of organs — though separated by breaks in phyllotactic continuity — stand close to one another longi- tudinally on the receptacle, but there may be prominent, naked inter- 88 MORPHOLOGY OF THE ANGIOSPERMS nodal areas on the lower part of receptacle, as in Liriodendron (Fig. 41F). Where the arrangement of organs is spiral, the phyllotactic spiral is often the same as that of the leaves but may be different, especially where the floral organs are numerous. The Sterile Appendages. The sterile appendages are typically of two kinds: sepals, which together form the calyx; and petals, which make up the corolla. These appendages are below the fertile appendages, the calyx below the corolla. (In Eupomatia, petaloid staminodia form a pseudocorolla between the stamens and the carpels.) Sepals and petals commonly differ in form, size, and other characters. In some families, they may be closely alike, as in most of the Liliaceae; in others, transi- tional forms occur, as in the Magnoliaceae. Interpretation of the peri- anth as calyx or corolla may be difficult and unimportant, as in Driniijs and Wintera, where the organs, ranging from two to several, are spirally placed; and where the perianth is represented by only one or few appendages, which serve as a bud-scale-like cap. Commonly, sepals are more or less leaflike or bractlike in form and structure, especially in their vascular relations to the stem. Mor- phologically, they are modified leaves. Typically, they stand in whorls, but, in some primitive families, are spirally arranged — Dilleniaceae, Paeoniaceae. They may be petaloid, but this condition usually accom- panies reduction or loss of the petals, as in the Proteaceae. Where greatly reduced, they have the form of minute teeth, scales, bristles, or mere ridges. Cornus shows all stages of calyx reduction in both ex- ternal and internal structure, and some species have no vestige of a former calyx. (Connation and adnation involving sepals are discussed in later pages.) Petals are typically laminar and larger than the sepals. They, like the sepals, are, morphologically, appendages of leaf rank. In many families, they represent sterile stamens, but, in some primitive families, they are probably modified leaves (like the sepals), as evidenced by transitional forms and anatomical structure in Magnoliaceae. Petals have great range in size and form, from the large, elaborate organs of some orchids to minute structures; under reduction, they may become scales, bristles, or glands of many forms. Accompanying compacting of the flowers in inflorescence specialization, they are reduced to small size and lose their petaloid appearance. Theories of the nature and development of the perianth are closely bound up with theories of the origin of the flower. One theory is that the perianth, at least in part, was preexistent to the flower. In ancestral forms, sterile appendages were associated with groups of sporophylls, as seen in the bracts below the cones of some conifers and the "flower" of the Bennettitales. These appendages became, with the establishment THE FLOWER 89 of angiospermy (enclosure of the ovules), a part of the flower. Elabora- tion of the perianth, with the development of two series of organs, took place within the angiosperms. Under a second theory, all perianth parts are considered modified sporophylls. This is an old theory that has received little attention since 1900. The petals in most families have been shown to be sterile, petaloid stamens. But all sepals and the petals of some families, for example, Magnoliaceae and Calycanthaceae, have been shown by transitional forms and anatomical structure to be modi- fied leaves. A third theory is based on the view that the primitive flower was naked, and the perianth developed within the angiosperms as a new structure, accompanying the specialization of the flower. Under this theory, the perianth appeared in early forms as bractlike or scale- like organs, protective in function; these organs were increased in num- ber and size, with elaboration in color and complexity of form. The de- velopment of a distinct upper series, the corolla, was one of the later steps. The flowers of ranalian families may give evidence of the origin of the perianth from bractlike, protective organs. They show variety in number and form of bud-scale-like outer appendages. Etipomatia and Himantandra have one calyptralike organ; other genera, a few small, spirally arranged organs. Trochodendron has a few appendages below a naked flower. The distinction between bracts and sepals is hardly possible in some of these genera. There are probably no strictly peri- anthless flowers — with the exception of those that are greatly reduced in structure, as in the inflorescences of CercidiphijUum and many of the Amentiferae. All stages in the evolutionary development of the calyx from bracts are present in the Ranales and Dilleniales. The corolla has probably arisen in two different ways. A perianth of many spirally arranged organs, with gradual transition from bractlike to larger petaloid members — as in the Magnoliaceae, Calycanthaceae, Himantandraceae, Nymphaeaceae — apparently became separated into proximal and distal parts, the distal more petaloid and otherwise dis- tinct from the proximal. Stages in this specialization are present in the Magnoliaceae. The genus Magnolia itself shows reduction in number in the members of the perianth, with gradual change from spiral to whorled arrangement (Fig. 41A to C). Some species show definitely distinct corolla, as does the allied genus, Liriodendron (Fig. 41D to F). In the corolla, stages in transition from spiral to whorled arrangement are frequent, and there may be two whorls of petals. A second origin lies in the sterilization and petaloid elaboration of stamens. Anatomy demonstrates that the petals of many families are, in vascular structure, unlike the sepals but like the stamens, regardless of extent of superficial resemblance in form to sepals; the number of traces departing to a petal is like that going to the stamen and unlike that to the sepal. The corolla 90 MORPHOLOGY OF THE ANGIOSPERMS has undoubtedly arisen in two ways. It represents modified stamens in most families; in some families, it represents the distal part of a primi- tive unspecialized perianth. The effect of reduction of the perianth is primarily on the corolla; the petals are reduced in prominence, to small nonpetaloid organs, often easily overlooked, or to complete external loss. The "apetalous" condition has commonly been considered an early stage in flower development, one in which the perianth is still in primi- tive form. But, in many taxa, the apetalous condition has been shown to represent a high stage in perianth specialization rather than a low one — Salicaceae, Betulaceae, Juglandaceae, Urticaceae, Aceraceae, Plata- naceae, Proteaceae, Fraxiniis. The Fertile Appendages. The fertile appendages, also of leaf rank, are of two types: microsporophylls (stamens) , which bear microsporangia; and megasporophylls (carpels), which bear megasporangia. The stamens constitute the anclroecium; the carpels, the gijnoccium. (The term pistil is applied to a unit of the gynoecium: to a single carpel when the carpels are free from one another; to a group of carpels when they are fused to one another.) Where the flower has only one kind of sporo- phyll, it is unisexual; where both are present, it is bisexual. If flowers of both unisexual kinds are borne on the same plant, the taxon is monoecious; if staminate and pistillate flowers are borne on separate plants, the taxon is dioecious. The flower varies greatly in number of kinds of parts, from four to one; in number of organs of a kind, from many to one; in extent of fusion of organs to one another, from complete absence of fusion to connation and adnation in extreme form; in elaboration of organs in size and form. The range in flower form and structure is very great — from the (superficially) simplest type, in which the flower consists of a single sporophyll on a receptacle, as in the staminate flowers of Euphorbia (Fig. 36A), to those with all kinds of appendages present, extensively fused together and elaborate in form. Variety in form is recognized as largely the result of reduction and of fusion. (Closely similar conditions are seen in the cones of the conifers.) Broad comparative studies of flower structure show lines of spe- cialization and, in general, the sequence in which changes have oc- curred. From these, the more primitive structure is determined and the course of phylogenetic modification can be outlined. The structure of the flower has been the chief basis for the classifications, artificial and natural, of angiosperms. And, although it is now recognized diat evidence of natural relationships must come from all parts of the plant, vegetative and reproductive, and from all fields of plant study, flower structure provides a foundation upon which theories of evolutioriary relationships can be based. THE FLOWER 91 The Primitive Flower The type of flower now generally recognized as morphologically simple is one that shows the least change, under adaptive evolutionary modification, from the original primitive flower. Evidence now seems to support strongly the theory that the ancestral flower was bisexual, with numerous stamens and carpels, without a perianth, or with a uniseriate perianth of simple, bractlike organs. (This theory differs greatly from that still held by some botanists that the primitive flower was uni- sexual.) All the appendages were spirally arranged, and the flower was symmetrical and without fusion among its parts. From this theoretically basic flower— essentially the "pattern flower" of preevolutionaiy tax- onomy — have developed many lines of modification, with reduction producing types more simple in kinds and number of organs, and with elaboration producing complexity in form. The major principles of evolutionary modification, upon wliich is based the acceptance of this type of flower as morphologically simple and primitive among living forms, are the following advances: 1. From many parts, indefinite in number, to few, definite in number 2. From three or four sets of appendages — perianth, androecium, and gynoecium — to one 3. From spiral to whorled arrangement of appendages 4. From freedom of floral parts to fusion — connation and adnation 5. From radial symmetry ( actinomorphy ) to bilateral symmetry ( zygomorphy ) In these principles, there is important departure from those of the Engler system of classification, which has long formed the foundation of the natural system of plant classification and is still largely in use in herbaria. The major difference lies in the interpretation of simplicity. Under the Engler system, the primitive flower is unisexual; advance is seen to bisexual flowers and to increase in number of sporophylls. In more recent classifications, fewer kinds and smaller number of ap- pendages represent specialization — reduction by loss of parts. This viewpoint has brought major changes in phylogenetic relationships of both large and small taxa; taxa formerly believed primitive are now considered advanced, and new lines of apparent relationship have been drawn. Prominent among these changes is the interpretation of the unisexual flower with few sporophylls as advanced, rather than primi- tive. Reduction in the Flower Reduction in the flower may occur in many or all parts, simul- taneously in several parts, or progressively from part to part. Loss may 92 MORPHOLOGY OF THE ANCIOSPERMS be of individual organs or of entire whorls of organs; loss may be minor, as loss of one petal or one stamen, or may occur in all parts of the flower, so that there remain only the receptacle and one or few sporophylls of one kind. All stages in loss of function and reduction in size can be seen in closely related taxa. Organs in process of evolution- ary reduction, "vestigial organs," can be recognized by abortive form and structure and by position in the flower; vascular anatomy mav aid in the interpretation of vestigial structures where identity is uncertain. The vascular traces of lost organs are usually present when external form is reduced, and may persist in the receptacle after the organ itself has disappeared. Under reduction, organs may be reduced in form and structure and changed in function. Transformations of petals and stamens into glands and of stamens into staminodia are probably the commonest changes. Evidence of the change of petals and stamens into glands is usually apparent in position of the glands and type and origin of vascular sup- ply. (Some glands and glandular surfaces represent merely secretory areas, not modified organs.) Plants with nonfunctioning petals are sometimes called apetalous when the petals are still present in vestigial or greatly reduced form, as in the Proteaceae. In this family, some genera have the petals present as laminar scales — Placospcrmum, Austromuellej-a — or filamentous projections; others have glands in the positions and with the anatomy of petals. In the Salicaceae, glands that undoubtedly represent a lost perianth stand around, behind, or in front of the sporophyll. The glands have the position and vascular supply of perianth parts, and some of them are lobed and some- what petaloid. The presence of a single gland seems to represent the greatest reduction, because the willows, primitive in other floral features, have several glands, and remnants of the vascular supply to other lost perianth parts are found in species with a single gland. Reduction in the Stamen. Reduction in the stamen occurs in all stages, from abortion of sporangia only, to complete disappearance of the or- gan. The abortion of two sporangia — one of each pair — is frequent; abortion of three of the four is rare. Loss of the entire anther is fre- quent; the stamen survives as a sterile, laminar, or filamentous organ, which may be petaloid. The petals of the majority of families seem to represent completely petaloid stamens. Stages in the loss of the stamen by gradual reduction in size of a staminodium are well shown in the genera of the Scrophulariaceae. In genera where no external remnant of the lost stamen survives, the vascular trace of the stamen is still present in the receptacle. Reduction in the Carpel. Reduction in the carpel is primarily in size and in number of ovules. Primitively, the carpel contained many ovules. A B D ■^ «i- P, Fig. 42. Diagrams of carpel structure showing reduction of ovules, and fusion and reduction of vascular supply. A to £, follicles: A, Hellchoms viridis, typical follicle, with many ovules and three traces; B, Trollius laxus, ovules reduced in number, traces to lost ovules persisting; C, Aquilegia canadensis, upper ovules and their traces lost; D, Ht/drastis canadensis, all ovules but lower two lost, one of these abortive; E, Waldsteinia fragarioidcs, all ovules but one lost, trace to one other persisting. F to O, achenes: F to M, with basal ovule surviving; N, O, with an upper ovule surviving. All with dorsal and ventral traces united at the base, in some as far as the ovule, which then appears to be attached to the dorsal (F, G, K, L, M). F, Geum rivals, with distal parts of all bundles present; G, Duchesnea indica, dorsal and ventral bundles greatly shortened; H, Fragaria vesca, with dorsal bundle greatly shortened; I, Agrimonia striata, with dorsal bundle lost beyond its union with the ventral bundles; /, Ranunculus Ficaria, the ventral bundles re- curving, not entering the style; K, R. Flammula, the ventral bundles greatly short- ened; L, R. Cymhalaria, only vestiges of the ventral bundles persisting, the dorsal bundle greatly shortened; M, R. aquatilis, the ventral bundles lost beyond the ovule, the dorsal bundle continuing hardly beyond it; N, Potentilla recta, one ovule surviving, others vestigial, the ventral bundles unreduced; O, P. canadensis, one ovule surviving, the ventrals lost beyond the ovule. (From Eames and MacDan- iels, adapted from Chute.) 93 94 MORPHOLOGY OF THE ANGIOSPERMS Reduction in ovules has been to few, and to one in the achene type. Loss of ovules may take place progressively from either end; the sur- viving ovules may be either the proximal or distal ones. The per- sistence of a median ovule appears to be rare. The position of the single ovule — distal or proximal — may form a good generic character, as in the Proteaceae. The sequence in ovule loss is demonstrated by the position of abortive ovules and presence of ovule traces where no ovule vestiges remain. This sequence in ovule loss is well shown in carpels of the Ranunculaceae and Rosaceae (Fig. 42). Where there is but one ovule surviving, the position of that ovule in the ancestral follicle can often be shown only by vestigial ovule traces. Reduction in the stamen and carpel is considered more fully in Chaps. 4, 6, and 7. BIBLIOGRAPHY Arber, A.: The interpretation of the flower: a study of some aspects of morphological thought, Biol. Rev., 12: 137-184, 1937. Arber, E. A. N., and ]. Parkin: On the origin of the angiosperms, Jour. Linn. Sac. Bot., 38: 29-80, 1907. Bancroft, H.: A review of researches concerning floral morphology, Bot. Rev., 1: 77-99, 1935. Celakovsky, L.: tJber den phylogenetischen Entwicklungsgang der Bliite, Sitzber. Kais. Bohm Ges. Wiss. Math.-Nat. Kl, 1896: 1-91; 1900: 15-221. Emberger, L.: La valeur morphologique et I'origine de la fleur (a propos d'une theorie nouvelle), Annee Biol, (III) 54: 279-296, 1950. Fagerlind, P.: Strobilus und Bliite von Gnetiim und die Moglichkeit, aus ihrer Struktur den Bliitenbau der Angiospermen zu deuten, Arkiv Bot., 33A: 1-57, 1946. Fisher, M. J.: The morphology and anatomy of the flowers of the Salicaceae, Am. Jour. Bot., 15: 307-326, 372-394, 1928. Gliick, H.: "Blatt- und Bliitenmorphologische Studien," Jena, 1919. Grant, V.: The protection of the ovules in flowering plants, Evol., 4: 179-201, 1950. Gregoire, V.: Sporophylles et organes floraux, tige et axe florale, Rec. Trav. Bot. Neerl, 32: 453-466, 1935. Hallier, H.: Zur morphologischen Deutung der Diskusbildung in der Dikotylen- blute, Medel. Rijks Herb. Leiden, 41: 1-14, 1921. Henslow, G.: On the vascular systems of floral organs and their importance in the interpretation of the morphology of flowers, Proc. Jour. Linn. Soc. London Bot., 28: 152-196, 1889. Heslop-Harrison, |.: The experimental modification of sex expression in flowering plants, Biol. Rev., 32: 38-90, 1957. Home, A. S.: A contribution to the study of the evolution of the flower, with special reference to the Hamamelidaceae and Cornaceae, Trans. Linn. Soc, (II) 8: 239-309, 1914. Janchen, E.: Die Herkunft der Angiospermen-Bliite und die systematischen Stellung der Apetalen, Oesterr. Bot. Zeitschr., 97: 129-167, 1950. Just, T.: Origine et evolution de la fleur, Annee Biol, 28: 135-148, 1952. BIBLIOGRAPHY 95 Kasapligil, B.: Morphological and ontogenetic studies of Urnbellularia calif ornica Nutt. and Laurus nohilis L., Univ. Calif. Piibl. Bot., 25: 115-240, 1951. Kaussmann, B.: Vergleichende Untersuchungen iiber die Blattnatur der Kelch-, Blumen-, und Staubbliitter, Bot. Archiv, 42: 503-572, 1941. Kozo-Poljanski, B.: On some "third" conceptions in floral morphology, New Phijt., 35: 479-492, 1936. Leppik, E. E.: The form and function of numerical patterns in flowers. Am. Jour. Bot, 43: 445^55, 1956. Martens, P.: La grain et le tube poUinique: Reflexions sur les caracteres propres des phanerogames, Acad. Roy. Belg. Bull. CI. Sci., 5 ser., 33: 919-943, 1948. Mason, H. L. : The concept of the flower and the theory of homology. Madrono, 14: 81-95, 1957. Mathews, J. R.: Floral morphology and its bearing on the classification of angio- sperms. Trans. Bot. Soc. Edinburgh, 23: 69-82, 1941. Nast, C. G.: The comparative morphology of the Winteraceae. VI. Vascular anatomy of the flowering shoot, ]otir. Arnold Arb., 25: 454-466, 1944. Neumayer, H.: Die Geschichte der Bliite, Abliand. Zool.-Bot. Ges. Wien, 14: 1-112, 1921. Ozenda, P.: Recherches sur les dicotyledones apocarpiques, Publ. Lab. £cole Norm. Sup., ser. biol. II, Paris, pp. 1-183, 1949. Parkin, J.: The unisexual flower: a criticism, Phtjtomorph., 2: 75-79, 1952. : The unisexual flower again — a criticism, Phytomnrph., 7: 7-9, 1957. : The strobilus theory, Proc. Linn. Soc. London, 135: 51-64, 1922. Payer, J. B.: "Traite d'Organogenie Comparee de la Fleur," Paris, 1857. Plantefol, L.: Fondements d'une theorie florale nouvelle: L'ontogenie de la fleur, Ann. Sci. Nat. Bot., 11 ser., 9: 35-186, 1948. Rauh, W., and H. Reznik: Histogenetische Untersuchungen an Bliiten- und Inflores- zenzenachsen. I. Die histogenese becherformiger Bliiten und Infloreszenzenach- sen sowie der Bliitenachsen der Rosoideen, Sitzber. Heidelb. Akad. Wiss. Math.- Nat. Kl, 1951: .3-71. Sargant, E.: The reconstruction of a race of primitive angiosperms, Ann. Bot., 22: 121-186, 1908. Sporne, K. R.: Correlation and classification in dicotyledons, Proc. Linn. Soc. Lon- don, 160: 40-47, 1948. — : A new approach to the problem of the primitive flower. New Phyt., 48: 259-276, 1949. Statistics and evolution of dicotyledons, Evol, 8: 55-64, 1954. Stebbins, G. L., Jr.: Natural selection and the differentiation of angiosperm famihes, Evol, 5: 299-324, 1951. Thompson, J. M.: Studies in advancing sterility. VII. The state of flowering known as angiospermy. Hartley Bot. Lab. Liverpool Publ. 12, 1934. : Towards a modem physiological interpretation of flowering, Proc. Linn. Soc. London, 156: 46-69, 1943^4. Troll, W.: "Organization und Gestalt im Bereich der Blute," Berlin, 1928. Van Tieghem, P.: Recherches sur la structure du pistil et sur I'anatomie comparee de la fleur, Mem. Sav. Acad. Sci. Imp. France, 21: 1-261, 1875. Vautier, S.: La vascularisation florale chez les Polygonacees, Candollca, 12: 219- 341, 1949. Wilkinson, A. M.: The floral anatomy and morphology of some species of Cornus and of the Caprifoliaceae, Thesis, Cornell University, 1945. Wilson, C. L., and T. Just: The morphology of the flower, Bot. Rev., 5: 97-131, 1939. Chapter 4 THE ANDROECIUM Classification by Arrangement on the Receptacle The aggregate of the stamens of a flower constitutes the androecium. In number, the stamens of a flower range from many to one, from in- definite to definite. In arrangement on the receptacle, they are spiral, whoiied, or fasciculate (clustered) — the fascicles usually in whorls. The Spiral Androecium. Many stamens, spirally arranged, constitute the primitive androecium, from which have been derived the whorled and the fasciculate types. This androecium is present in many of the lower dicotyledons — Degeneriaceae, Dilleniaceae, Magnoliaceae, Eupo- matiaceae, Annonaceae, Calycanthaceae, Nymphaeaceae, and some of the Ranunculaceae, Monimiaceae, and Winteraceae. The Fasciculate Androecium. The nature of stamen fascicles was much discussed — with little agreement — in the earlier days of taxonomic morphology. It has recently received attention again, chiefly in connec- tion with the theory of the telomic structure of the stamen. The fascicle has been considered both a branching and a compound organ: its branching, the result of radial and tangential splitting (chorisis) of a simple stamen; its compound nature, the result of aggregation of simple stamens, connate in various degrees (Fig. 43). Fascicles consist of various numbers of stamens — from very many, to few, and to one. Within the fascicle, development is often, perhaps always centrifugal. Connation is of various degrees — by the stamen traces within the re- ceptacle and, externally, by the filaments. Stages in the connation are present in several families — Dilleniaceae, Paeoniaceae, Cactaceae. In these families, a few stamens in the cluster may be wholly free, both externally and in their traces. Ontogeny, in showing fascicles arising as a mound, with individual stamens developing on the surface, has been considered to show evidence of branching. But evidence from ontogeny is not in itself conclusive; organs phylogenetically fused arise con- genitally united, wholly or in part. A comparable condition exists in the sympetalous corolla where the lobes — morphologically, the petal tips — and the tubular base arise from a simple primordium. The fascicle has been interpreted as a primitive form of the stamen, a branching system of fertile telomes, reduced and compacted, with the 96 THE ANDROECIUM 97 simple stamen a surviving solitary branch of this compound system. Similarity of the fascicle to the stamen of Ricinus (Fig. 44) has been cited in support of this theory, and resemblance to tlie bennettitalian stamen has been noted as a part of the evidence for the origin of angio- sperms from the Bennettitales. The fasciculate arrangement seems to c \i f(^'^^^^ Aquilegia Paeonia Dillenia Fig. 43. Diagrams showing vascular supply to the androecium. Aquilegia, stamens with independent traces; Paeonia and Dillenia, "trunk" vascular supply to the stamen fascicles, c, carpel; d, disc; p, petal; s, sepal; st, stamen. (From Eames, 1953. Paeonia after Brouland; Dillenia drawing by Wilson.) Fig. 44. Sketches of parts of branching stamens of Ricinus communis. 2, 3, varieties from Java showing morphology of the anther; anther sacs free, attached to the base of the elongate connective. (After van der Pijl; 1, from Sack's Lehrbuch.) have arisen at least twice — among the Dilleniales and among the woody Ranales. A series in elaboration is seen in the Myristicaceae, Monimiaceae, and Lauraceae. In these three families, part or all of the members of a fascicle are transformed into nectaries, which often re- tain stamen form. The theory that the fascicle is merely an aggregation of simple stamens is supported by comparison of related taxa and by anatomy. In the Dilleniaceae, Paeoniaceae, Myrtaceae, and Hypericaceae, series 98 MORPHOLOGY OF THE ANGIOSPERMS in taxa show the clustering of numerous free stamens to form multi- stamen fascicles, and reduction in number of fascicles and number of stamens per fascicle. In the Dilleniaceae, Wormia, Dillenia, and Hib- bertia form an obvious series in evolutionaiy specialization in flower structure and in habit, from trees to small shrubs and vines. In Dillenia, where there is no external grouping among the many spiral stamens, tnmk vascular supplies bind clusters of stamens together anatomically. In the first two genera, the stamens are free; in Hibbertia, the stamens are in fascicles. Hibbertia is a large genus of small shrubs and woody vines, with high specialization in flower structure — zygomorphy of corolla and androecium in its most advanced species. The zygomorphy is expressed most strongly in the androecium. In the Hypericaceae, Ascijron and other genera have numerous free stamens; within the genus Hypericum, there is a series from free stamens to the fasciculate condition, with reduction from five to three fascicles, and in stamen number per fascicle from many to three. Anatomical stiucture supports the interpretation of some supposedly simple stamens and nectaries as end products of the reduction of fasci- cles. The androecium in the Lauraceae and in Parnassia shows evidence — both from comparison of external form and of vascular structure — of strong reduction. Some of the stamens and nectaries (in the Lauraceae, modified stamens) have lateral or basal appendages (Fig. 45). Ana- tomically, these appendages represent individual stamens connate in a fascicle with a fertile stamen or with other sterile members of an original fascicle. In Sassafras and Umbellularia (Fig. 45A to C), the nectariferous basal lobes of stamens have independent vascular traces, two extending from the receptacular stele into tlie lateral structures, the other to the anther; the typical fertile stamen has only one trace. In Benzoin, the glandular stamen has three traces, one extending into each lobe. Each stamen in Laurus has three vascular bundles, two supplying the lateral lobes. Each of the several lobes of the staminodium of Varruissia has an independent trace; the unlobed stamen has several traces arranged in the filament in a loose cluster. The stamen and the staminodium (nectary) of this genus are homologous structures, repre- senting greatly reduced stamen fascicles with connate members. (The presence of the vascular bundles of several connate stamens within the "filament" of die typical stamen is an excellent example of the per- sistence of vascular structure after external form has been lost.) The existence of fasciculate stamens in these taxa is of phylogenetic im- portance. It supports the view — maintained by cytological evidence — that Parnassia does not belong in the Saxifragaceae; it aids in determin- ing the relatives of the Lauraceae (Chap. 11). Though the existence of "trunk" vascular supplies for fascicles would. THE ANDROECIUM 99 in itself, seem to support the view that the fascicle is a unit organ, other anatomical evidence shows the evolutionary development of tlie fascicle. In the Paeoniaceae (Fig. 46), Dilleniaceae, Monimiaceae, and other families, some stamens — especially those on the outside of the fascicles — have simple, independent traces, and other stamens have traces Fig. 45. Stamen fascicles of Umbellularla showing lateral members transformed into nectaries and stages in the loss of the nectaries, vascular supply to nectaries inde- pendent and like that of the fertile stamen in type and origin, msp., micro- sporangium; fi., filament; st.n., staminal nectary. ( After Kasapligil. ) loosely coherent with the "trunk," or connate with it for only a short distance. (The anatomy of the vascular supply of fasciculate stamens, especially the histological structure of the trunk vascular supply, is little known.) The Whorled Androecium. Whorled arrangement in the androecium represents a modification of spiral arrangement, as does the similar condition in leaves. Where the stamens are in one whorl, the androecium is haplostemonous; in two whorls, the members of the outer whorl al- 100 MORPHOLOGY OF THE ANGIOSPERMS ternating with the petals, diplostemonous, and the members of the outer whorl opposite the petals, obdiplostemonous; where in more than two whorls, polijstemonous (Fig. 47, A, B). Polystemonous androecia are uncommon [four whorls in Lauraceae (Fig. 47B), several in Aquilegia, Delphinium, Nandinn, Trochodendron; three in some species of Illicium]; haplo- and diplostemonous are common. Obdiplostemonous Fig. 46. Diagrams of flower structure of Paeonia brownii, showing fascicled stamens. A, showing trunk vascular supply to stamens; B, showing various degrees of con- nation of stamens in fascicle, c, carpel; d, disc appendage; p, petal; sd, staminodia; St, stamen; s, sepal. (A, from Eames, 1953.) androecia are characteristic of the Caryophyllaceae, Geraniaceae, Oxalidaceae, Rutaceae, some Saxifragaceae, and some other taxa. Obdiplostemony represents an interruption in the usual sequence of alternation in floral whorls; its morphological nature was much dis- cussed in the 1870s and 1880s. Three interpretations of obdiplostemony were proposed: that it arose by the addition of a new or extra whorl of stamens (the intercalation theory); by the loss of a whorl between the corolla and the androecium (the reduction theory); and by ontogenetic THE ANDROECIUM 101 displacement (the displacement tlieory). Under the intercalation theory, a new whorl of stamens was considered interposed between the corolla and the lower stamen whorl. Under the dedouhlement theory, the alternate new members were formed by "dedouhlement," a Fig. 47. Floral diagrams of polystemonous androecia showing reduction in the androecium; transformation of lateral members of the stamen fascicles into nec- taries and reduction in number of the whorls. A, Umbelluhria calijornicu: the inner whorl vestigial, the next outer whorl of nectary-bearing fascicles, the two outer whorls of simple stamens. B, C, Laurus nobilis: staminate Hower B with two whorls of nectary-bearing fascicles and one whorl of stamens; pistillate flower C with one whorl of staminodes. ( After Kasapligil. ) Fig. 48. Floral diagrams showing paired stamens ( dedouhlement) in the outer whorl of androecium. A, Sagittaria sagittifolia; B, Butomus umbellatus. {After Salisbury.) doubling by radial splitting of the original outer whorl (Fig. 48). Under the Carpel Polymorphism theory, it has been claimed that obdiplostemony transgresses the law of alternation of successive whorls, but, in androecia with more than two whorls of stamens, there is com- 102 MORPHOLOGY OF THE ANGIOSPERMS monly no alternation (Fig. 48A). Under still another theory, that of origin by division, which is supported by the frequent presence of epi- petally, the new stamens were considered morphologically petals, formed by tangential splitting of the petals, with the new inner organs trans- formed into fertile appendages. (This theory arose from reading in the wrong direction the common and readily interpreted series in the forma- tion of staminodia and petals from stamens.) Evidence supporting the corolline nature of the supposedly inserted whorl was seen in the stipule- like appendages of the petals and stamens in some taxa. Strong objec- tions were at first raised to the acceptance of the dedoublement theory, even before abundant evidence showed that it has no support in ana- tomical structure. Under the displacement theory, the new members in the stamen whorl were considered to have reached their positions by differential growth, displacement from an inner position. The displacement theory found later support under the theory of carpel polymorphism, where "balloon- ing," caused by the development of carpellary locules, was considered to have pushed outward the antepetalous stamens, so that, in obdi- plostemonous flowers, they stand farther out than the antesepalous stamens. (Displacement of this sort would not be morphologically sig- nificant, because it does not affect the level of origin of the stamen traces. ) Involved in a discussion of possible intercalation of a whorl in the androecium is a related aspect of major flower modification — the con- densation of two stamen whorls by suppression of the internode between them, with the insertion of the members of one whorl between those of the other. Evidence of this step is seen in many eight- and ten-stamen taxa in a not-quite-perfect whorl (the genus Acer is a good example) and in the greater length or earlier development of alternate stamens. The stamens of this double whorl may be perfectly aligned on one level, but the members of one whorl usually stand a little outside those of the other and their traces arise below those of the other whorl. This condition gave support to the views of intercalation and to the long- maintained displacement theory — that differential growth in the re- ceptacle had brought the new whorl into union with the other. Growth of this sort was believed even to move the inner whorl to a position out- side the outer whorl and thus to bring about obdiplostemony. Evidence of this was supposedly found in the course of the traces through the cortex, but the position of trace origin was not noted. Union of two whorls is well demonstrated in several families. In the papilionate Leguminosae, the ten stamens may arise as two whorls but, in maturity, form one; differential growth brings the whorls together, with resulting THE ANDROECIUM 103 connation. In the Caryophyllaceae, obdiplostemony is incomplete; tran- sitional stages in union of whorls are numerous. In the Saxifragaceae, both diplo- and obdiplostemonous taxa are present, with whorls separate or fused. Evidence supporting the reduction theory — loss of the outer whorl of stamens — is strong. Polystemony is generally accepted as primitive and diplo- and haplostemony as advanced. (The story of simplification in the flower is one of progressive shortening of the receptacle and dropping out of whorls.) Stages in reduction of whorls in the androe- cium are readily seen in some families — Lauraceae, Caryophyllaceae, Saxifragaceae, Primulaceae, Sapotaceae, Myrsinaceae, Theophrastaceae. Most genera of the Primulaceae are diplostemonous, but, in Samolus and Naumburgia, the outer whorl of stamens is represented by rudi- ments; in some species of Primula, the outer whorl, usually absent, may be present in normal form. Objection has been raised to the interpretation of obdiplostemony as brought about by loss of a whorl of stamens, on the ground that it is merely a matter of spatial and mechanical possibilities and, therefore, has no morphological significance. But the presence of remnants of tlie lost organs and their traces, as in Primula, is evidence of the loss. Reduction in the Androecium In reduction in the androecium, individual organs or entire whorls may be lost. In related taxa, the inner whorl may be lost in some families, as in the Iridaceae; the outer whorl in others, as in the Burmanniaceae and Haemodoraceae. The Orchidaceae are an example of members lost from both ancestral whorls. The origin and course of the vascular traces to the stamens in obdi- plostemonous taxa support the theories of the loss of the outer androecial whorl and of the union of the two whorls in those taxa that have only one whorl but have double the number of members present in other floral whorls. Obdiplostemony and the union of two stamen whorls are steps in the shortening of the receptacle in specialization of the flower by reduction. Ontogeny provides no evidence of value in the interpretation of obdi- plostemony. Sequence in development among the various whorls of organs in the flower is variable; it is probably less commonly acropetal than otherwise. In Oxalis and Geranium, sequence in development of the two whorls may be in either direction; in Commelina, the inner whorl of stamens develops before the outer. Transformation of Fertile Stamens into Slaminodia. Reduction of the androecium, whether spiral or whorled, by transformation of fertile 104 MORPHOLOGY OF THE ANGIOSPERMS stamens into staminodia may occur in either outer or inner members (Fig. 47A) or both; typically the outer members. Transformation of the inner stamens seems to be the more primitive condition, as seen where the petaloid staminodia form a prominent pseudocorolla above and inside the androecium in Eupomatia (frontispiece). In other primitive flowers — Degeneria, Hi7nantandra, Calycanthus, and some of the Moni- miaceae, Nymphaeaceae, and Helobiales — staminodia are present both above and below the stamens. If the upper position is the earlier one in the evolution of the flower, as suggested by these primitive taxa, the first "corolla" was above the stamens, a position which accompanies pollination by beetles. Reduction in Stamen Number within Whorls. Aside from the reduction of whorls in the androecium, loss of stamens within the whorl is a prominent feature of specialization in many families. This loss accom- panies zygomorphy and other adaptations to pollination methods. There is often variety in reduction in stamen number within a family, as from five to four and to two — Scrophulariaceae; even within a genus, as in Polygonum, from nine or six to three and two. Reduction in stamen number is in one whorl only or in both whorls and is of all degrees; it ranges from suppression of a single stamen in the androecium to suppression of all but one. Extreme reduction — to a single stamen — is occasional, as in Euphorbia, Callitriche, Sarcandra, Najas, Casiiarina, Hippuris, Lilaea, Zostera, Triglochin, Wolffia, Mangifera, most orchids. Solitary stamens occur in flowers otherwise greatly reduced, chiefly those of aquatic genera, of elaborate zygomorphic form, or of greatly compacted inflorescences. In unisexual flowers, the solitary stamen has often been claimed to be terminal on the receptacle and therefore cauline in nature, but ontogeny and vascular structure of the entire flower show such a stamen to be appendicular, a true microsporophyll in a pseudoterminal position (Fig. 36). (In this respect, it is similar to the pseudoterminal carpel.) Some so-called terminal stamens are fused stamens; two or more connate sporophylls occupy a central posi- tion in the flower — Salix (section Diandrae), Zannichellia, Typha. Evi- dence that the solitary, apparently terminal, stamen is not morphologi- cally terminal and cauline is seen in its ontogeny, for it arises off center; in its form, it is not radially symmetrical, because two sporangia are larger than the other two; and the anther is definitely dorsiventral. The interpretation of solitary stamens and carpels as cauline involves the acceptance of sporangium position in the angiosperms as both cauline and appendicular. Indiscriminate sporangial position — cauline or foliar — is found only in the lower plants, and the presence of both types in the highest plants — even within a genus, as is seen under the Phyllo- sporae-Stachyosporae theory — is morphologically inconsistent. THE ANDROECIUM 105 There is great variety in the form and function of sterile modified stamens: they may be transformed into petaloid organs — Eupomatia, TroUhis; or into secretory organs — Coptis. Frequently they are repre- sented by abortive structures, mere stubs, or vascular traces ending in the receptacle — Scrophulariaceae, Anacardiaceae. Reduction to one fer- tile stamen and the sterilization of half the anther of that stamen, with the sterile part petaloid, are characteristic of some highly zygomorphic families — Cannaceae, Zingiberaceae, Marantaceae, some Aizoaceae. Reduction of Stamen Fascicles. The formation of fascicles is probably an early step in androecium reduction. Further steps consist of reduc- tion in number of fascicles and in number of stamens per fascicle. Both these steps are well shown in Hibbertw, where reduction of fascicles is from several to one, and of stamens, similarly, from many to very few; and in Hypericum, where five fascicles are reduced to three, and sta- mens per fascicle from many to about three. Extreme reduction in the fascicle is seen in the Lauraceae and in Tarnass'm; some apparently simple stamens are surviving members of reduced fascicles — those of Sassafras, Benzoin. Fusion in the Androecium Connation of members of a whorl — ontogenetic or phylogenetic — is frequent in both whorled and fasciculate androecia. An androecium with connation by filaments is monadelphous where all stamens form a single cluster; diodeJphous where two clusters are formed; polijadeJphous where there are more than two clusters. It is adelphous where there is no connation. An androecium with anthers united is termed syngenes- ious. Union of anthers is commonly ontogenetic but is congenital in some taxa. Ontogenetic fusion commonly covers cohesion as well as connation. (Cohesion implies a loose, not intimate, union of similar structures, brought about by glandular secretions, by close appression with interlocking of epidermal cells, or by cuticular projections; conna- tion implies histological union, with lines of union either evident or obscure. The anthers of the Compositae, Lobeliaceae, Solanaceae, some Gentianaceae are coherent.) Connate anthers are infrequent or rare^ the Typhaceae, some Cucurbitaceae. The anthers of the Lobelioideae seem to show stages from cohesion to connation. Fusion by filaments — monadelphous or diadelphous — may involve one, two, or perhaps more whorls. Union of two whorls to form an apparent one is frequent, as in some legumes and the Thymelaeaceae. Adnation of the Androecium Fusion of stamens to other organs of the flower, especially to the corolla (this is termed epipetaly), is common. Fusion to the calyx is 106 MORPHOLOGY OF THE ANGIOSPERMS less common than to the corolla — Proteaceae. Fusion to the carpels, where all the outer organs are together fused to the gynoecium in perigyny and epigyny, is common. Fusion to carpels alone is rare — Sarcandra, Monimiaceae. Fusion to styles and stigma, with the forma- tion of a gijnostemium, is characteristic of the Orchidaceae and Stylid- iaceae. Adnation may be by filaments for part or for all their length, where the anthers are sessile on other organs. (The term sessile is unfortunately applied both to anthers where free stamens consist of anthers only, and to those where filaments merge with the uniting organ.) Adnation of the anther to other organs varies in extent, from attachment near the base only to union by the entire dorsal or ventral surface — the dorsal surface to a sepal in many Proteaceae, the entire ventral surface to the gynoecium in the orchids. In Yiscum, fusion of the entire stamen to the sepal is intimate, and all external evidence of the fusion is lost; the sporangia are apparently borne on the ventral surface of the sepal, which is described as "polliniferous." Stages in fusion — ontogenetic or phylogenetic — of stamen to adjacent organs can be seen in many taxa. External evidence of the fusion may remain, the filament forming a ridge on the surface of the petal or sepal, or the only evidence may be internal, in the presence of the vascular bundles of the two organs, which lie, radially, side by side. Fusion, when phylogenetically established, may include the vascular tissues, and the bundles of the two organs may be merged and histologically in- distinguishable. The Androecium under Zygomorphy Number, form, and structure of stamens are greatly modified in the development of zygomorphy. Some of the stamens become sterile, are reduced in size, and lost, leaving little or no evidence in external form, though sometimes internal vascular traces may persist. Often, where there is more than one whorl of stamens, one entire whorl, or some members of either, or of both, whorls are suppressed. Of a whorl of six in Carina, one remains fertile, four become petaloid staminodia, and one is lost; in the Bignoniaceae, two stamens are fertile and three sterile. Correlated with sterilization and loss in the androecium under develop- ment of zygomorphy, are modifications in size and form, in time of maturation, and in adnation of the stamens to other organs. A number of terms are used in descriptive taxonomy for these variations but they have li\.tle morphological significance — for example, didtjnomous, where the androecium consists of two pairs of stamens of unequal length; tetradynamous, of four long and two short stamens. the androecium 107 Ontogeny of the Androecium The stamens commonly develop in acropetal, that is, centripetal, sequence. The existence of basipetal, or centrifugal, sequence in the androecium in several taxa was noted as early as 1871 but received little attention. Recently, centrifugal sequence in the androecium has been emphasized as probably an important character in the determina- tion of phylogenetic relationships between families and orders. In many families, order of origin and maturation is not on record; it is con- spicuous in taxa with massive, spiral, multiorgan androecia; in families with few stamens, it is difficult to determine. A sequence is present within fascicles in the Hypericaceae and Dilleniaceae. Undoubtedly, centrifugal development was derived from centripetal, the normal acro- petal sequence. Centrifugal development is known in the following multistaminate families and will doubtless be found in others — Actini- diaceae, Aizoaceae, Bixaceae, Cactaceae, Capparidaceae, Dilleniaceae, Hypericaceae, Loasaceae, Lecythidaceae, Malvaceae, Paeoniaceae, Thea- ceae; it is unknown in the monocotyledons. The major many-stamen families with centripetal stamens are Annonaceae, Lauraceae, Legum- inosae, Lythraceae, Magnoliaceae, Myrtaceae, Nymphaeaceae, Papa- veraceae, Punicaceae, Ranunculaceae, Rosaceae. The Geraniales and Centrospermae have been considered probably "referable to the cen- trifugal series," and, certainly, if the Capparidaceae belong in this group, so must their close relative, the Cruciferae. Examples of the value of sequence of development in determination of relationships can be seen in its use as an important character in the removal of Paeonia from the Ranunculaceae and the establishment of the Paeoniaceae as a member of the Dilleniales; and in the transfer of the Crossosomataceae from the Rosales to the Dilleniales. Centrifugal development seems to characterize certain groups of families commonly recognized as related and is apparently an important character in confirming supposed relationships and in suggesting others. But this does not mean that all taxa with centrifugal stamens belong in the same phylogenetic line; reversal of sequence of development has doubtless appeared more than once, as have all other advanced floral characters. Similar departure from normal sequence in develop- ment is seen in the inflorescences of some palms (Carijota), where the inflorescences appear successively down the trunk, and in some cauli- florous tropical taxa. Centrifugal sequence in flowering is occasionally seen in inflorescences — Vallisneria (staminate), Dipsaciis (proximal part). This sequence in inflorescences may be apparent only — the result of the condensation of large complex inflorescences. 108 MORPHOLOGY OF THE ANGIOSPERMS THE STAMEN The stamen, or microsporophyll, in contrast with the other floral appendages, is typically a slender organ, unleaflike in form. But it is, like the other organs, basically of leaf rank and is closely similar to all these organs in morphological and anatomical structure, in ontogeny, and in its relation to the stem (floral receptacle). Less important re- semblances are in the epidermis: stomata are frequent on the connective — less common on the filament and absent on terete forms; and epi- dermal appendages like those of other floral organs are occasionally present, even over the anther sacs, as in Cahjcanthus jeHilis. With the exception of the primitive laminar types, the stamen shows little super- ficial resemblance to the megasporophyll, the carpel, but the homology of the two organs is usually unquestioned. In most families, the stamen is a highly specialized organ, with great variety of elaborate form related to methods of pollination. Typically, it consists of two more or less distinct parts: a proximal, sterile part, the filament; and a distal, fertile part, the anther (Fig. 49P). The anther consists of the microsporangia and the sterile tissues between and within which the sporangia are borne. The term connective is usually applied to the strip of tissues that lies between the pairs of sporangia, but this median strip is not separable, histologically or morphologically, from the tissues of the anther-sac wall, which enclose the sporangia. The stamen is usually considered a rather simple organ, but its simplicity is false; it is, in general, more complex and more reduced than the carpel. Filament and anther are usually distinct, but in the more primi- tive families there may be no clear limitation of fertile and sterile parts. The filament merges into the anther (Fig. 49E to H). The connective may extend beyond the sporangia, sometimes forming a more or less distinct appendage (Fig. 49 A, B, F, I, K). The filament, the connective, and the appendage of the connective constitute continuous parts of the microsporophyll and are not morphologically distinct units. The con- nective is the sterile median part of the anther, connecting and, in various degrees, embedding the sporangia. Morphologically, it repre- sents a median part of the laminar sporophyll. The Primitive Stamen Like the primitive carpel, which has no distinction of ovary, style, and stigma, the primitive stamen has no distinction between filament and anther. The most primitive stamens in living angiosperms are probably those of some of the woody Ranales (Fig. 50). These are broad, more or less leaflike organs, without, or with weak, distinction between fertile and sterile parts. The sporangia are borne near the Fig. 49. Semidiagrams showing variety in stamen form from laminar to anther- filament with intermediate types. A, Belliolum haplopus; B, Eupomatia laurina; C, Illicium; D, Hillebrandia; E, Ccratophyllum; F, Descainea; G, Lactoris; H, Chloranthus glaber; I, Magnolia nitida; J, M. hypoleuca; K, Calycanthus floridus; L, Michelia; M, Euryale; N, Sagittaria; O, Hydrastis; P, Caltha. {Adapted from — A, Bailey and Nasi; C, I, J, L, Canright; D, photograph by R. Gauthier; G, Challenger Exped.; H, Swamy ''\' ^' J' 0"ter and central stamen of N. odorata, re- spectively; K stamen of Cabomha aquatica; L, stamen of Brasenia schreberi M stamen of Nelumbo nucifcra. (After Moscley.) 120 Fig. 55. Specialized stamens of the Cabombaceae and Nymjohaeaceae showing vas- cular anatomv', position, and dehiscence of the sporangia. The sporangia are united in pairs by endothecial tissue. 1, 2, abaxial views and cross sections of Cabomha aquatica and C. caroliniana, respectively showing anatomy simple, dehiscence latrorse-extrorse in 1, A- A, and varying from c.\trorse, B—B, to latrorse-extrorse, A-A in 2; 3, Nuphar variegatum, anther sacs protuberant in pairs, connective mas- sive; 4, Brascnia schreheri, anther sacs semiprotuberant, dehiscence latrorse; 5, Nelumbo iiiicifera, vascular supply complex, filament strands five, twisting and re- orienting in their course {B-B, C-C, D-D), evidence of differential growth in de- velopment of terete from laminar filament. (After Moseley.) 121 122 MORPHOLOGY OF THE ANGIOSPERMS pendage of the anther, an appendage that seems to be generally functionless. (In one species of Magnolia, it is large and apparently plays a part in pollination.) In the more advanced genera of the Nymphaeaceae and in die Cabombaceae — Barclaija, Eunjale, Cahomha, Brasenia — it is absent. The most prominent part of the reorganization in the specialization of the anther is the change in position of the sporangia (Fig. 53). The sporangia, remaining in pairs, are moved ("migrate," "slide") laterally, so that they come to lie, first, obliquely lateral ("trapezoidal") (Fig. 54D), with dehiscence latero-introrse; then lateral, with dehiscence latrorse (54L); and as a final position, halfway around the anther, ahax- ial, with dehiscence extrorse (Fig. 55, 2, h-h). This positional change, as exemplified by the Nymphaeaceae, has been likened to the "phyletic slide" of sori in the leptosporangiate ferns. The positional changes in the sporophyll seem to involve all the tissues except the midvein. The lateral veins are moved laterally and around toward the opposite side, as are two of the sporangia. In the Nymphaeaceae, the lateral veins have been described as "twisting and rotating in their upward course" in the stamen (Fig. 55, 5, c-c, d-d). In some taxa, some of these bundles in the anther are inverted ( Fig. 53D ) or have united with others to form amphicribral or bicollateral forms. Their course and orientation have been distorted during the more or less massive differential growth. Similar inverted bundles have been found in the stamens of Eiipomatia. This inversion of bundles in anthers resembles that of the ventral bundles of carpels, but, in the carpel, there is definite folding or rohing of the lamina and a chamber is enclosed; that in the stamen is by massive differential growth, and no chamber is formed. (Changes in the orientation of vascular bundles, including inversion, are frequent in specialized petioles.) The course and orientation of the lateral veins of the stamens in the Nymphaeaceae and Cabombaceae give strong support to the theory of sporangium migration. In the laminar stamens, the vascular bundles are distributed and oriented as in leaves, but, in stamens with narrow filament, thick connective, and lateral sporangia, the outer bundles are often obliquely oriented or inverted (Fig. 53). In Nelwnbo, the highly specialized filament shows stages in the twisting and inversion of the lateral veins. Accompanying this distortion in orientation is loss of the weaker lateral veins (Fig. 55, 5). Progressive change in sporangium position from adaxial to abaxial has been recognized as a feature of advancing specialization in the anther. This change has apparently occurred independently in several, perhaps many, families; the MagnoBaceae and the Nymphaeaceae are excellent examples, for they show many stages in the change. The THE STAMEN 123 small family, Cabombaceae, seems to complete the series in the Nym- phaeaceae, with the sporangia abaxial in Cabomba. It is apparent that, at least in many families, the primitive position for microsporangia is adaxial. The abaxial position in primitive families with especially primi- tive stamen form — Degeneriaceae, Himantandraceae, Lactoridaceae, An- nonaceae, for example — has already been discussed. The Ranales, in their most primitive stamens, present a major morphological problem in the variety of microsporangial position; there are both adaxial and abaxial types. Among ranalian families, some — Magnohaceae and Nymphaeaceae — show phyletic migration of microsporangia from adaxial to abaxial positions. ( Megasporangia in angiosperms are adaxial, without excep- tion.) There seems no question but that the adaxial position is one primitive position and that from this have been derived lateral and abaxial positions. But it seems unlikely that the abaxial position in the highly primitive stamens of Degeneria, Himantandra, and Lactoris is secondary — that these stamens are highly specialized in sporangium position. It is possible that both positions are primitive, a retention in the stamen of morphological structure of an ancestral taxon. (In the Pteridospermae, ovules were apparently borne on both surfaces of the leaf. ) The old search (1820-1850) for evidence of basic sporangium posi- tion by locating the stamen margins was well planned; it should be con- tinued, with further critical studies of marginal meristems in stamens. Ontogeny should add strong support to the theory of the migration of sporangia. Leaves, petals, carpels, and broad stamens (Fig. 50) show marginal meristems. In slender stamens, these meristems have been lost wholly or in part; terete filaments apparently have no marginal growth, but weak marginal growth is found in "flattened" filaments and in the broader connectives. The somewhat flattened filaments of Lirio- dendron have weak marginal meristems; the broader filaments of Mag- nolm have well-marked marginal meristems. ( Liriodendron seems to have been derived from ancient magnolian stock — its extrorse dehiscence derived from introrse. ) Much more information about the development of the sterile tissues of the stamen is needed, especially of semilaminar forms. If the margin of the filament can be distinguished by linear meri- stems and these meristems followed upward into the connective, the position of the sporangia can hardly be questioned. Evidence of another type that all four sporangia belong on one side of the sporophyll comes from the close histological association of each lateral pair. The sporangia on each side often fuse, in late stages, and the pollen grains mingle in a common chamber at dehiscence. Still further support for the theory that sporangia migrate is present in stamens where sporangium number is reduced to two. In these stamens, the 124 MORPHOLOGY OF THE ANGIOSPERMS missing sporangia are the outer members of the pairs of the ancestral lamina, often smaller and nearer the edges of the lamina wings than the others. Latrorse anthers also provide in their form — intermediate be- tween laminar and extrorse or introrse — evidence to support the ex- treme change in position of the two outer, smaller sporangia. Latrorse dehiscence has commonly been recognized as more primitive than -'^// //////////z^m^^ -< ^/y'y^//////////////M//y/zF?2 >- w///////////M/^/////////77:r r> g B Fig. 56. Diagrams of stamen structure. A, longitudinal sections of stamens showing stages in freeing of the anther sacs from the lamina; B, cross sections of anthers showing stages in the development of anther sacs by the reduction of the lateral and connective tissues. extrorse and introrse. The morphologists of the nineteenth century called it "normal"; the other types, specialized. In position on the sporophyll, latrorse anthers provide the step between laminar and ex- trorse or introrse; they are characteristics of many primitive families — Ranunculaceae, Rosaceae, Butomaceae, Crassulaceae, Monimiaceae, Tro- chodendraceae. In position on the sporophyll, the microsporangia of the typical anther are submarginal; in the primitive stamens, they are laminar. Dis- tribution of sporangia is much the same in micro- and megasporophylls, THE STAMEN 125 though the similarity is obscured. Primitively, both are laminar; in advanced forms, submarginal; in specialization, the laminar form is lost in microsporophylls, greatly reduced in megasporophylls; the mega- sporangia are primitively many; the microsporangia always four (higher numbers represent divided sporangia or united sporophylls ) . Reduction to two or one sporangia is uncommon in stamens, common in carpels. Parallelism in the two types of sporophylls is to be expected. Number and distribution of sporangia are correlated with form of sporophylls. Primitively, the sporangia were deeply embedded in the tissues of the sporophyll — Degenerla, Michelia, Himantandra, Ceratophijlhim (Figs. 67, 51, and 49). In the specialization of the anther, progressive reduc- tion and reorganization of the sterile tissues have freed the sporangia from their burial in the lamina, by the formation of anther lobes (Fig. 56). But some of the sterile tissues remained as protective layers about the sporangia and formed the heavy anther-lobe walls. The spo- rangia within the anther lobes are wall-less, as in the laminar sporo- phyll. Each lobe of the anther contains the pair of sporangia of one side of the ancestral stamen. As the anther sacs became strongly pro- tuberant, they were gradually freed from the connective (Fig. 56f/, e, f), until ultimately nearly free [versatile anthers (Fig. 56/)]. The separa- tion began at either the distal or the proximal end, or at both ends simultaneously, with resulting variations in point of attachment. Where the anther lobes are nearly "free," they are attached to a small median remnant of the connective, the wall is thin, and the anther consists largely of sporangia, two in each half ( Fig. 56/ ) . An anther of this type, freed at both ends, is typically X-shaped, with the ends of the halves often spreading — many grasses. Even where the anther lobes are nearly free and the form and position of the sporangia are prominent, the sporangia themselves are not free — in the sense of superficial — for they remain sheathed by a few layers of sporophyll tissue. Anther-sac Wall and "Sporangium Wall" The wall of the anther sac is usually described as made up of an epi- dermis, a hypodermal or fibrous layer, and one to several parenchym- atous layers. The term exotheciiim has been used for the two outer wall layers — the epidermis plus the fibrous layer — and endotheciuni for layers between the fibrous layer and the sporogenous tissues. Also, the epi- dermis alone has been called the exothecium, and all layers within it, the endothecium. There has been no consistency in the use of these terms; they cause constant confusion in interpretation and have no morphological value. In unspecialized anthers, the epidermis resembles that of the filament; in highly specialized anthers, with strongly pro- tuberant sporangial areas, the intermediate parenchymatous layers may 126 MORPHOLOGY OF THE ANGIOSPERMS be absent and the wall two-layered, consisting only of epidermis and fibrous layer, or of the fibrous layer alone, when the epidermis has been lost ontogenetically. The fibrous layer is associated, mechanically, with dehiscence and varies in position, extent, and structural details with type of dehiscence. In the primitive types, the fibrous layer covers the sporogenous tissue and extends laterally over adjacent sterile tissue somewhat beyond the fertile tissue, as in Degencria and Himantandra (Figs. 67 and 51); in advanced anthers, the fibrous layer may be more extensive, covering the protuberant areas and, in some high types, the entire anther. In anthers with apical dehiscence, the fibrous layer is usually absent, ex- cept in the region of the pores. In the elaboration of mechanical de- hiscence, a series apparently exists in increase in extent of the fibrous layer and in simplicity of wall structure, accompanying increase in pro- trusion and freeing of the anther sacs. The anther wall becomes simpler by reduction of cell layers; the layers between the tapetum and the fibrous layer are reduced to one layer, and this also may be lost in the highest types. Even the epidermis may be lost — in part or wholly — and the fibrous layer take its place as the outermost layer, its cells remain- ing fibrous or becoming epidermislike. The loss of layers may be ontogenetic — occurring early or late in the development of the anther — or phylogenetic. Ontogenetic breakdown of the inner layers occurs at about the time of sporocyte formation. In delicate anthers, like those of the grasses, the wall of the anther sacs may have lost all but the outer two layers, the epidermis and a one-layered endothecium. Rarely, the epidermis may degenerate early or be lost just before dehiscence and the anther wall described as one-layered. This one-layered wall consists, at least in part, even in highly specialized anthers, of connec- tive or laminar cells. There is, morphologically, no microsporangium wall in angiosperms — an important character of these plants. (Anther- wall structure is further discussed under Dehiscence of the Anther.) The anther lobe or sac is sometimes called a sijnangium, but, mor- phologically, this is inaccurate, especially when a "significant" resem- blance is seen to the synangium of the Bennettitales. The cycadophyte synangium consists of two connate sporangia; the angiosperm anther lobe consists of two sporangia individually encased in sporophyll tis- sues, sometimes widely separated by them. In the massive reorganization of the tissues of the fertile part of the laminar stamen to form the anther, the areas in which the pairs of sporangia lie become protuberant — progressively more and more markedly so, with increased specialization in sporophyll form. The sporangia of the more primitive stamens, sunken in the mesophyll, are wall-less (Fig. 67). With the reduction of the lamina, the sterile tissues THE STAMEN 127 around the sporogenous cells persist and, where the protuberances are large and unusually thick-walled, are perhaps added to from nearby regions. The anther-sac wall, thus formed, is laminar in nature, not a true sporangium wall. The entire anther-sac wall or one or more layers of it have frequently been described as the sporangium wall, but the stamens of the Ranales show clearly the naked character of the sporo- genous tissue. The protruding sporangia, though histologically and functionally walled, are "morphologically naked"; even in the most thin-walled and free anther lobes, the sporangium is "sunken." In anther sacs, the sporangia are in a position more favorable for pollen distribu- tion than in the center of a laminar stamen. (The megasporangium is probably also naked, though in interpretation of the morphology of the ovule, the nucellus is commonly considered the sporangium wall.) The importance of the wall-less character of the angiosperm micro- sporangium seems to have been neglected. The apparent presence of a sporangium wall — the anther-sac wall and the endothecium — has ob- scured the wall-less condition. Absence of a sporangium wall is prob- ably an important angiosperm character, one that will play a major part in the determination of the ancestry of the angiosperms. The Pollen-containing Chamber. Several terms have been applied to the pollen-containing chamber: anther chamber, pollen sac, locule, theca. Anther lobe, anther sac, and anther half, though obviously cover- ing more than the chamber and its contents, have sometimes been used as synonyms. All these are morphologically loose terms, because they may represent one, two, or four sporangia. The term theca is perhaps the most used and has unfortunately come to be well established, sup- planting others because it is less technical. Pollen chamber is an un- desirable term, because it may suggest the entirely different pollen- receiving chamber of the gymnosperms. All these terms must remain loose, general terms, but theca should be avoided in technical descrip- tions because it may represent one, two, or, rarely, more sporangia. It was called obsolete in morphological literature many years ago. Anthers with two lobes are termed dithecoiis; with one lobe, mono- thecotis. (Both these terms are morphologically unsatisfactory, because they involve the loose term theca.) The dithecous anther is a typical complete anther with four sporangia; the monothecous anther is usu- ally a half-anther with two sporangia, the sporangia of a lateral half of a typical anther. But the monothecous anther may have the sporangia of one lateral half abortive or absent, as in many Labiatae; it is rarely a transverse half of an anther, with one sporangium of each lobe abor- tive. An anther described as monothecous may have one, two, or four sporangia, with their contents united in a common chamber. The anther sac usually contains the pollen of two sporangia, but, in latrorse 128 MORPHOLOGY OF THE ANGIOSPERMS anthers, where the sporangia are isolated in the four corners, there is usually no fusion of sporangia at maturity and the "theca" represents one sporangium (the anther is "tetrathecous"). In some highly spe- cialized anthers, all four sporangia unite to form one chamber, and the "theca" represents four sporangia — CaUitriche, some Cucurbitaceae. Anther-dehiscence Types. On the basis of position or direction of dehiscence, stamen types — better called dehiscence or anther types — are distinguished in taxonomic use as introrse, when pollen is freed from the anther directly or obliquely inward in the Hower (adaxially); eoArorse, when dii-ectly or obliquely outward (abaxially). (In some genera, anthers that appear introrse in the bud appear extrorse in the flower — and those extrorse in the bud, introse in the Hower — because of changes in form or structure of the filament or of movement of the anther on the filament apex.) When pollen is shed laterally, dehiscence is termed Introrse; in older usage, "normal." Introrse anthers character- ize the majority of angiosperms; extrorse anthers characterize a rather small group of families — Calycanthaceae, Myricaceae, Cucurbitaceae, Fumariaceae, Lardizabalaceae, Aristolochiaceae, Iridaceae, Juncagina- ceae, Potamogetonaceae, and a few others. Some families — Liliaceae, Alismataceae — have both types; and, in the Magnoliaceae, Liriodendron has extrorse anthers, the other genera, introrse; in Fagopyrum and Persea, one whorl of stamens is extrorse, the other introrse; in Cinna- momum, two whorls are introrse, one extrorse; in CommeJina, two of the three stamens are extrorse, the third, introrse. Anthers introrse in the bud may become extrorse at flowering — Geraniaceae, Caryophyllaceae, some palms. The latrorse position is intermediate between the primi- tive laminar and the advanced extrorse and introrse — Ranunculaceae, Butomaceae, Menispermaceae. Figure 52 shows a series of theoretical steps in the modification of the ancestral lamina of the microsporophyll to form the latrorse and, ultimately, extrorse or introrse anther. Even in anthers that, in form, are extrorse or introrse, discharge of pollen may be in a lateral direction — Solarium, Lonicera, Valeriana, Primida. Such dehiscence as this characterizes genera but not larger taxa. All anther types could have been derived from sporophylls with, primitively, either adaxial or abaxial sporangia. Orientation of sporangia as indicat- ing extrorse or introrse is commonly apparent early in the ontogeny of the stamen, but often, in early stages, the sporangia are symmetrically placed near the four angles of the primordium and the final position is determined by differential growth. This growth may be sti'ong and even suggest cambiumlike activity (Fig. 57C, D). The terms applied to dehiscence have much descriptive and taxono- mic value and are morphologically significant, in that an evolutionary sequence is obvious from the simple extrorse and introrse dehiscence THE STAMEN 129 of the primitive laminar stamen to latrorse and to the elaborated ex- trorse and introrse of advanced anthers. The presence in the same taxon and even in the same flower of extrorse and introrse dehiscence is evi- dence that these types represent ecological adaptations. Dehiscence in laminar stamens is necessarily simple and determined directly by sporangium position. In advanced anthers, sporangium position is not simple; phylogenetic shifting has brought the outer members of each pair to the side opposite that of their original position, and the resulting obliquely inward or out\vard dehiscence is derived, secondarily ex- trorse or introrse. Both extrorse and introrse dehiscence have apparently Fig. 57. Diagrams and a sketch of cross sections of anthers showing modification in form of connective at time of flowering. A, B, Iris pumila, by differential growth: A, young, B, mature; C, D, Ntjinphaea colorata, by cambium-like growth, develop- ment of ridge on connective: C, young, D, mature. (A, B, after Engler; C, D, after Kaussniann. ) been developed in some taxa from ancestrally simple extrorse — Laura- ceae — and from ancestrally simple introrse — Magnoliaceae except for Liriodendron. Latrorse dehiscence similarly has two origins. It is clear that extrorse and introrse dehiscence do not necessarily indicate deriva- tion from ancestral forms with abaxial and adaxial dehiscence, respec- tively. The terms must be considered as chiefly of descriptive value. Union of Sporangia. In the primitive stamen, the sporangia lie close together in lateral pairs; in advanced stamens, they remain so or have been broguht even closer together by reduction of intermediate laminar tissues. In many families, they unite laterally by breakdown of the separating wall, usually late in ontogeny, as the pollen matures. The 130 MORPHOLOGY OF THE ANGIOSPERMS pollen grains then lie in a common chamber, from which they are shed through a single opening. Early breakdown of the wall occurs occa- sionally — Berberis, Nuphar, and some orchids. The two pollen masses may mingle or lie side by side without separation by sterile cells — Cassia, and some poricidal anthers. In the union of the two sporangia, the tissues of the entire separating wall or of some part of it disinte- grate and are largely absorbed. The adjacent outer anther wall, above and beside the separating wall, is commonly also involved in the disin- tegration; the outer anther-sac walls are retracted, and the intermingling pollen is shed through a large opening. In some taxa, the separating wall persists almost to the outside of the anther, and the chambers dis- charge through a common break which opens separately below into the two chambers. Often the pollen forms a mass which fills the furrow between the two sporangia. The details of opening vary greatly, espe- cially the extent of retraction of the anther walls, which may even com- pletely invert the chamber. The variations depend, at least in part, upon the extent of the fibrous layer and the thickness of the anther wall. The double character of the chamber is usually evident externally as well as internally, but the chamber may appear simple in late stages. Sporangium fusion is uncommon in anthers with latrorse dehiscence. Union of sporangia is probably always between an "outer" and an "inner" sporangium — the two that belong morphologically together on the laminar sporophyll. The association of the lateral pairs is obviously close, morphologically as well as histologically, and refutes the view that two sporangia are abaxial and two adaxial and the theory that, morphologically, the anther represents two conifer sporophylls "back to back." Unusual Forms and Arrangements of Stamens Unusual forms and arrangements of stamens that suggest splitting, forking, and branching received much attention from 1850 to 1900 and are still occasionally discussed. Chorisis, dedouhlement, and multiplica- tion of anthers were considered methods of origin for supposedly branched stamens. Today these terms are little used, even in taxonomic descriptions, and have long had morphological implications that are, at least in part, false. The stamen is fundamentally a simple, unbranched organ, as shown by the primitive genera Degeneria, Eupomatia, Austrohaileija. Many of the so-called branching forms can now, with the aid of anatomy and comparison with the stamens of related taxa, be interpreted more ac- curately than formerly. Longitudinal division ("splitting") of the stamen occurs in some taxa, as in the Mai vales, but, in other taxa, the ap- parent division represents partial connation of two or more stamens. THE STAMEN 131 "Dichotomous" stamens have been seen as occurring in Salix in the in- terpretations of the "New Morphology," as examples of retention of primitive structure. The "dichotomous" stamens of some species of this genus are pairs of stamens partly united, not forked organs. Two stamens borne on opposite sides of the flower and connate by the bases of their filaments may suggest dichotomy, but anatomy of the flower shows two independent vascular supplies, derived on opposite sides of the receptacle. Taxonomists have correctly interpreted this "doubled" stamen as an end product of reduction in the androecium. Dedoublement, a doubling — if repeated, a multiplication — of stamens has, since the middle of the nineteenth century, been used in both taxonomy and morphology to explain the presence of two — sometimes more than two — stamens, where one would be expected in a floral pattern (Fig. 48). Among the best known examples of supposed de- doublement are the pairs of longer stamens in the Cruciferae and the pairs of stamens opposite each petal in Alisma. In the crucifer flower, two pairs of stamens would fit into the dimerous floral diagram of the family, with two pairs (whorls) of sepals, two of petals, and two of carpels. But six stamens cannot be placed in alternating whorls of two with the other organs. Dedoublement of two of an original four (two whorls) has been used to explain this lack of conformity in the androecium. Neither anatomy nor ontogeny supports the doubling theory; the vascular supply and the primordia of all the stamens arise independently. The members of each pair of stamens in Alisma repre- sent independent organs with their vascular supplies derived well apart on the floral stele. Many objections were raised at the end of the nineteenth century to the acceptance of dedoublement as an explanation of the presence of pairs and clusters of stamens where a single organ is to be expected. Fifty years later, dedoublement, in the sense of a division, seems to be a sound explanation of the origin of those pairs or clusters of "half- anthers" where it is obvious, on anatomical and ontogenetic evidence, that the anthers have been divided longitudinally. In these stamens, the division may extend downward in the filament even to the base — some of the Malvales. But its use as an explanation of pairs of complete stamens, with independent vascular supplies is morphologically in- correct. Forked stamens, apparently the result of longitudinal division or "splitting," are occasional, as in Salix, Corijlus, and Ostnja, but these stamens represent pairs of stamens connate throughout part of their length, rather than forked individual organs. Dedoublement, as a true splitting, a dividing of the entire stamen, has been claimed for Adoxa, where pairs of stamens alternate with the petals, with separation oc- 132 MORPHOLOGY OF THE ANGIOSPERMS curring in the primordia. (It is doubtful whether this is true dedouble- ment, and it should be studied further.) The extent to which forking or doubling represents true division, rather than fusion, can be determined only by comparison with related taxa and by anatomical study. (Double flowers have been said to be formed by the ontogenetic splitting and transformation of stamens and carpels.) The stamen of Ricinus, massive, with many branches, each bearing a terminal "anther," has been used as a branched type (Fig. 44). But this interpretation is doubtless incorrect. The natine of this stamen can probably best be determined by comparative studies of inflorescences and flowers throughout the Euphorbiaceae, a family with many "flowers" of difficult interpretation. The staminate flower of Euphorbia has been reduced to a single stamen (Fig. 36A). The stamen of Ricinus is per- haps an inflorescence. Support for the theory of telomic organization of the sporophylls of angiosperms has been seen in what seems to be dichotomy of the branches of the stamen of Ricinus. The tips of the branchlets show an apparent dichotomous forking, with each branch terminated by a sporangium. But the several varieties of the castor bean show that each ultimate dichotomy represents an anther with the two anther sacs nearly free from the connective, which is scalelike and often deciduous (Fig. 44). The nature of fasciculate or clustered stamens was early discussed by morphologists and taxonomists, who considered the clusters branched or compound stamens, or "stamen systems." (Fasciculate stamens are discussed under Androecium. ) Anatomy of the Stamen Stamens with well-differentiated anther and filament usually have a single trace which continues as a simple vascular bundle through the filament to the anther and sometimes into the distal prolongation. Branching in the connective is occasional, but the branches bear no relation to sporangium position. In delicate stamens, the vascular supply may be vestigial and discontinuous or even absent; where absent within the stamen itself, it may be present as a vestigial "stub" in the re- ceptacle — some Scrophulariaceae. A two-trace supply to the stamen is rare: Austrohaileija, Sarcandra (a vesselless genus of the primitive Chloranthaceae), some of the Pro- teaceae, Victoria and some other genera in the Nymphaeaceae, Casuarina, and some of the Betulaceae. The filaments of Nuphar, Cyrtandra, Eranthemum, and Peristrophe were long reported to have a pair of bundles, and that of Donjanthes, four bundles. In Casuarina, THE STAMEN 133 two traces enter the filament and soon fuse to form a single concentric bundle. In the Betulaceae and Fagaceae, there is evidence of the double nature of the stamen bundle; in several genera, there are, at the base of the filament, either two separate bundles or one bundle with two xylem strands or two xylem poles; in Conjlus, the single bundle arises from two traces. The phylogenetic history of the trace supply of the stamen has been clarified by the better interpretation of nodal structure in the shoot. Changes in number and relation of traces have been the same as those of the leaf. From the basic two-trace supply — characteristic of all ap- pendages — have been derived the three- and the several-trace types by addition of lateral traces to a median double trace (representing the basic two, fused); from the basic two (by fusion) and from more than two (by loss of laterals) has been derived the single trace. Most laminar and semilaminar stamens have three traces and three major veins. Three traces characterize some of the primitive families: Degeneriaceae, Himantandraceae, Annonaceae, Lardizabalaceae, many of the Magnoliaceae, and Nymphaeaceae. Eupomatia has five or seven. Three traces, often with additional laterals, are present in most mono- cotyledons — Musaceae, Zingiberaceae, Marantaceae. The common one- trace condition has obviously arisen by reduction from a higher num- ber in association with the narrowing of the filament. Evidence of reduction in trace number from three or more to one by loss of the lateral traces — as also occurs commonly in leaves and carpels — is well shown in the Magnoliaceae. In Magnolia, the lateral bundles may be weak and discontinuous; in Michelia, some species have three traces, others one, with vestiges of the lost bundles. In these genera, distortion and rupture of lateral bundles are associated with reduction in width of the stamen, accompanied by the development of an anther and the migration of the sporangia. As early as 1824, there was a surprising understanding of stamen anatomy; it was stated that, in slender stamens, the lateral veins vanish and the median persists. The present-day view that the single-bundle supply was derived by reduction from a three-trace supply is very old, not recent, as generally believed. Branching of the vascular supply is uncommon or rare in stamens with one trace, but is frequent in stamens with more than one trace. Minor lateral branches may connect with major veins or anastomose to some extent with one another, but only rarely does the branching and fusion suggest the reticulate structure of most carpels. Appendages of the stamen usually have no vascular tissue, but, in the Melastomaceae, the anthers have prominent adaxial and abaxial projections or lobes that are completely vascularized. These appendages 134 MORPHOLOGY OF THE ANGIOSPERMS have been considered possibly remnants of an ancestral telome system, but they can, at least in large part, be interpreted as projections of the connective, often reflexed and adnate to the body of the anther. The Melastomaceae are an advanced family, and the projections suggest modification of anther form related to uncommon methods of pollina- tion, perhaps ornithophily. The "trunk" vascular supply to stamen fascicles is a compound structure. That it consists of the connate traces of several to many stamens is apparent from the presence in several primitive families — Paeoniaceae, Dilieniaceae, Monimiaceae — of stages in the union of a group of stamens (Fig. 46). "Branched stamens" are formed in this way. Fascicled stamens have been considered branched stamens that are perhaps homologous with the bennettitalian stamen, but the branched microsporophyll of the Bennettitales has a single, simple trace. Ontogeny of the Stameh^ Ontogeny of the stamen is basically like that of sepal, petal, carpel, and leaf; there is growth in length, width, and thickness. Early de- velopment is by apical and marginal meristems (Fig. 58). Marginal meristems are active for periods varying with the form of the filament and anther. In winged filaments, they are well defined and long per- sistent; in narrower filaments, they cease activity after periods varying with the width of the filament; in terete filaments, they are absent. Weak marginal meristems may be present in thick, subterete filaments. In the connective, marginal growth is difficult to determine, because it is weak and obscured by growth in thickness. Sufficient attention has not, however, been given to details of the ontogeny of the connective and filament. Critical ontogenetic studies should be made of anthers with broad connectives and subterete filaments to obtain evidence in support of the view that both margins are on one broad side of the anther. The stamen primordium arises on the floral receptacle as a crescent- shaped projection, where the filament is broad, or as a rounded pro- jection, where the filament is terete or nearly so. The primordium elongates rapidly and soon assumes roughly the form of the anther, with some indication of its orientation and relation to the filament. At this stage, the filament is represented merely by the base of the primordium. Witli increase in size of the primordium, bulging areas indicate the position of the sporangia. These regions develop rapidly and the rest of the anther grows more slowly. Differential growth in the connective and anther wall complete the development of mature anther form. At this stage, differential growth is an important feature THE STAMEN 135 in enlargement, because it is largely responsible for the orientation that determines direction of dehiscence. Increase in cell number on one side of the connective — sometimes by almost cambiumlike growth (Fig. 57) — displaces the sporogenous areas toward the other side — toward the adaxial side where dehiscence is introrse, and toward the abaxial side where dehiscence is extrorse. Latrorse dehiscence develops where a subcylindrical primordium becomes more or less four-sided and tliere is little or no differential growth in die connective. The Fig. 58. Cross sections of filament of stamen of Cleome gigantea showing marginal development. I to IV, progressive stages in ontogeny, d, v, dorsal and ventral sides; mi, initials of the meristems; pc, procambium. ( After Kaussmann. ) sporogenous areas remain in the corners. The strongly marked differ- ences of orientation in extrorse and introrse anthers are the result of differential growth in the connective — plus, in some taxa, a twisting at the filament tip, as in some palms — and do not indicate an important morphological difference. With evolutionary reduction of the sterile tissues of the sporophyll, the protuberance of the sporangial regions increases, but the sporangia remain enclosed by the several-layered anther-sac wall; they do not become "free." The wall layers are progressively reduced, with in- creasing elaboration of the anther lobes, until only two layers remain, 136 MORPHOLOGY OF THE ANGIOSPERMS an epidermis and one hypodermal layer (Fig. 59). "Uniseriate" anther walls — so-called "sporangium walls" — are probably misinterpretations of biseriate walls where the epidermis has been lost ontogenetically (Fig. 59C, G, H). Even in aquatic plants with extreme tissue reduction Fig. 59. Cross sections of anther-lobe walls showing reduction at maturity to epi- dermis and one hypodermal layer in A, D, I, or to the hypodermal layer with fragments of epidermis persisting in the other diagrams. E, swollen cells over septum which ruptures the wall at dehiscence. A, Humulus; B, Broussonetia; C, Dorstenia; D, E, Zannichellia; F, Trema; G, Casuarina; H, Ricinus; I, Zostera. {After Staedtler. ) throughout the flower, the anther wall is two-layered. In the angio- sperms, there is no sporangium wall distinct from enclosing tissues of the sporophyll. The description of the microsporangium wall as "many layered" is based on the interpretation of tlie anther-sac wall as the sporangium wall. The filament commonly remains very short — sometimes a mere base for the developing anther — until late stages in flower development. Its THE STAMEN 137 development may be rapid just before the flower opens — largely as the result of cell elongation— and may continue after the flower is open. Ontogeny of the Sporogenous Regions. Study of the development of the microsporangium in angiosperms began about the middle of the nineteenth century; it was a popular field of morphological study from about 1870 to 1900, when many taxa were investigated. In the large amount of detailed information available, much confusion in interpreta- tion and terminology exists. This applies particularly to the description of the anther wall and its ontogeny, where topographical terms have been applied loosely. This is not surprising. Most of the early studies were purely descriptive and were rarely compared with one another. Variation in structure of the anther wall (which has been generally assumed to be the sporangium wall) is great. Many of the descriptions were based on the theory that the stamen was cauline in nature, and special attention was given to origin of specific layers from "derma- togen" or "periblem." Deeply sunken sporangia were unknown; the sporangium was considered superficial. Anther wall and sporangium wall were often considered synonymous terms; commonly, the wall of the pollen sac was called the sporangium wall. The possibility that sterile tissues of the sporophyll may form part of the sporangium wall did not enter the picture. The loss of layers in the wall during ontogeny — even of the epidermis — was usually overlooked. Only with recogni- tion of sporangium position as primitively sunken in the sterfle tissues of the sporophyll has there been a basis for interpretation of the mor- phology of the anther wall. Part of the difficulty of interpreting the nature of the several wall layers outside of the sporogenous tissues lies in the variation in number of these layers. The stamen has, unfortunately, been commonly looked upon as a simple organ, readily interpreted morphologically; actually, it is the most highly specialized of floral organs, resembling an ancestral laminar sporophyll far less than the carpel, and showing much greater variety in form. Adaptation to various types of pollen distribution has brought about not only great range in external form but in the his- tological structure of the anther. The sporangium, originally buried in sterile tissue, seems to have become "superficial" and, in the highest types, is largely free from the surrounding sterile tissues. Variety of wall structure, seen in the many intermediate stages in this series, with phylogenetic and sometimes ontogenetic loss of cell layers, obscures the story of modification. Early in the ontogeny of a typical anther, when the beginning of anther form is evident in the primordium, groups of cells in the center of each of the four corners can be distinguished from the cells around 138 MORPHOLOGY OF THE ANGIOSPERMS them by their large size, denser cytoplasm, and larger nuclei. These cells, in elongate, often crescent-shaped or platelike rows, form the archesporium (Fig. 67). The term archesporium, like so many other terms used in Hower ontogeny, has had several meanings; it has been applied not only to these rather readily recognized cell clusters but to the central tissues at an earlier stage, from which arise not only the sporogenous tissue and the tapetum but inner wall layers. The archespo- rial cells — also called the primary sporogenous cells — enlarge radially and divide periclinally, cutting off outer cells which form the tapetum. The inner cells, after this division, form the primary sporogenous cells. From the tissues outside tlie archesporium arise the epidermis and wall layers. The terms applied to these cell layers as they develop — primary parietal layers, parietal layers, primary tapetal layers — have been so variously and inconsistently used that they are valueless. Doubt- less, there is much variation in the behavior of the meristematic layers, as in other meristems, and it is impossible here to ascribe more than general origin and function to specific layers. The subepidermal (fibrous) layer and one or more layers immediately below this are clearly derived from the hypodermis of the young anther; deeper-lying layers, even the layer that becomes the tapetum, apparently may have different origins. The terms exothecium and endothecitim have been commonly applied to outer and inner layers, respectively, of the mature anther-sac wall, but are used as loosely as they were in reference to layers in the de- veloping wall. Exothecium has been applied to the epidermis alone — the common usage, to the epidermis plus the fibrous layer, and to the outermost layer, regardless of its morphological nature. Endothecium has been applied to the fibrous layer alone, to the wall layer or layers below the fibrous layer, and to all of these together. The confusion in use lies, in part, in the history of the terms, which are old and were at first considered descriptive only. Early students of anther structure overlooked the ephemeral layer below the fibrous layer, as well as the occasional loss of the epidermis. The identity of layers in highly specialized anthers can probably be determined only by ontogeny and comparative studies of related taxa. Exothecium and endothecium have only ecologic and topographic value; they are not of morphological value. The anther wall consists, typically, of three layers: the epidermis; the fibrous layer, immediately below the epidermis; and one or more parietal or wall layers betweerj the fibrous layer and the tapetal cells. But the anther wall may be reduced to only one or two layers (Fig. 59). Fibrous layer is a poor term, since it implies that the layer consists of fiberlike cells. Its cells, though differing greatly from the surrounding THE STAMEN 139 cells, are not fiberlike, but the term is an old one and will probably continue in use. Histologically, the fibrous layer may be added to be- low the sporogenous and tapetal cells, by transformation of parenchyma cells of the connective so that the inner tissues are completely encircled by the fibrous layer. The parietal layers are continuous with sterile cells of the connective. In further development, the epidermis usually remains simple but its cells may be velvety papillose, as in Gloxinia and Gladiolus, and the hairs may cover the entire anther, as in species of Calijcanthus. Stomata — doubtfully functional — are frequent on the connective, rare over the sporangia. Where the fibrous layer is strongly developed, the epidermal cells may be tabular and thin, appearing collapsed and scalelike. They may be early deciduous or lost, as dehiscence approaches, over the entire anther or over restricted areas. As cells of the fibrous layer mature, lignified or suberized thickenings develop in the walls, contrasting strongly with the thin, cellulosic wall areas between. The thickenings take the form of bars, spirals, or an- nular bands and often resemble the thickenings of the cell walls of protoxylem, to which they have even been considered morphologically equivalent and, therefore, of phylogenetic significance. But they are doubtless of only ecological importance, representing a structural mech- anism involved in dehiscence (Fig. 68). Under drying, the thick and thin wall areas shrink unequally, and the layer, with attached layers, is ruptured. Continued drying of the wall tissues brings about retraction of the borders of the opening; changes in atmospheric humidity cause opening and closing of the rupture. In some genera, as in Liliinn, the rupture is through a stomium, a group of specialized epidermal cells similar to those of the stomium of some ferns. It was long believed, with the existence of the stomium as evidence, that dehiscence was brought about by the epidermis alone. But the epidermis may be lacking at time of dehiscence, especially over the lines of rupture. Where the sporangia are deeply sunken, the sporogenous cells are covered externally by only two layers, the epidermis and the fibrous layer. The fibrous layer extends a short distance beyond the sporangium, as in Degeneria (Fig. 67) and Himantandra. With increasing specializa- tion of the anther, the fibrous layer extends farther and farther around the anther lobes, developing internally from parenchyma of the con- nective where it is often irregular in thickness. The parietal layers are often flattened by enlargement of the tapetal and sporogenous cells and may be crushed and absorbed. This is especially true where there is but one layer. Where the connective is greatly reduced, the fibrous layer may extend all over the anther. In anthers with poricidal or val- vular dehiscence, the fibrous layer is usually absent or very weak. 140 MORPHOLOGY OF THE ANGIOSPERMS The Thin-walled Anther. It was long ago stated that the microsporan- gium of the pteridophytes and gymnosperms, with the exception of Ginkgo, had only an exothecium (epidermis); that the angiosperms, in contrast, had both an exothecium and an endothecium (cell layers be- tween the epidermis and the sporogenous and tapetal tissues). The validity of this distinction was questioned when it was found that the anther sacs of many angiosperms had walls of only one or two cell layers. It had been pointed out that thin-walled anthers belonged chiefly to taxa then generally considered primitive — Casuarinaceae, Piperaceae, Proteaceae, Urticaceae — and the similarity in sporangium structure to that of the gymnosperms gave support to the view that these taxa were among the most primitive angiosperms. But it was, even at that time, argued that these families were highly specialized in many other char- acters, that the simple wall structure of the anthers was the result of reduction in specialization. Only with the general recognition that these families are not liighly primitive, and that the woody Ranales are the primitive angiosperms, has the phylogenetic significance of the thin- walled anther become apparent. The deeply sunken sporangia of laminar stamens, without a sporan- gium wall and with their sporogenous tissues surrounded by sterile cells of the microsporophyll, have important phylogenetic significance. They set the angiosperms well apart in method of microspore-bearing from other vascular plants, except the primitive eusporangiate ferns. The multiseriate anther wall suggests possible relationship of the angio- sperms to ancient eusporangiate ferns. In the 1920s, the uniseriate anther wall was correctly recognized as a reduced structure, but the interpretation received little attention. The evidence presented then can well be outlined here, because it supports the growing opinion that the families in which it occurs are advanced, not primitive. Reduction of the wall layers comes about by ontogenetic or phylo- genetic loss of the innermost or the outermost layers, or of both of these layers. Loss of the epidermis may occur early in ontogeny of the anther or just before dehiscence, or vestiges may remain, especially over the line of fusion of the two sporangia in a sac (Fig. 59C, G, H). Scattered remnant epidermal cells may persist over the entire anther sac. Where the epidermis is degenerate or reduced, the fibrous layer takes its place as the outermost layer, retaining its characteristic cell-wall thickenings, as in Casuarina. Variety of structure in thin-walled anthers and the evidence of re- duction from thicker-walled types are shown in the following examples. All stages in the reduction can be seen in the Urticiflorae. Fictis — per- haps alone in the order — shows a typical anther wall, with well-de- THE STAMEN 141 veloped fibrous layer. Genera of the Urticaceae, with a uniseriate wall and no fibrous layer, show the extreme reduction structure. Transition forms occur in some Moraceae, the Ulmaceae, and Cannabinaceae. In Dorstenia and Broussonetia (Fig. 59B, C), a few epidermal cells are present over the region of wall fusion between the two sporangia of the anther sac. Hiimuhis (Fig. 59A), Ulmus, and Cannabis show epi- dermal cells scattered over the fibrous layer. The isolated epidermal cells (Fig. 59G, H) are survivors of the typical epidermis of earlier stages of anther development. The simple wall of the Urticiflorae is the result of reduction. Evidence is the presence of isolated epidermal cells, survivors of a typical epidermis of earlier ontogenetic stages, and the absence of the layer of wall cells below the fibrous layer, also often suppressed in ontogeny. Explosive dehiscence characterizes many of the genera of the Urtici- florae, especially those of the Urticaceae. The pollen grains are hurled into the air by sudden bursting of the anther-sac wall. A remarkable feature of this type of dehiscence is the simultaneous explosion in some tropical genera of great numbers of anther sacs, forming smokelike puffs of pollen grains. The mechanism of explosive dehiscence is not under- stood. It has been reported to be the result of uniform shrinkage of a uniseriate wall, of internal turgor, and of pressure of maturing pollen grains upon a delicate enclosing wall. A structural feature of the anther sac, unknown in other types, may be a part of the mechanism: there is a short, longitudinal slit in the wall of the young anther. Explosive dehiscence is a highly specialized type developed in association with anemophily. In Casuarina, the simple anther wall suggests superficially the spo- rangium wall of gymnosperms (Fig. 59G), but its outer wall is a typical fibrous layer and the epidermis persists, as a few isolated cells, over the partitions between the sporangia. The anther is not explosive but dehisces normally. In the Proteaceae, the anthers of some taxa — Leucadendron — have lost the epidermis completely, and others — species of GreviUea — have only scattered surviving cells. In the Piperaceae, Pepewmia also has persisting remnants of the epidermis. In submersed aquatic plants, the entire plant is structurally reduced in adaptation to its habitat. The anther sacs of these plants — with the exception of Zostera and Zannichellia — have uniseriate walls. Zostera, a transitional form, has a biseriate anther wall. The outer wall layer is of large parenchyma cells — a true epidermis; the inner layer is of thin cells with weak wall thickenings, which do not function in dehiscence. The pollen is freed by local swellings that break open the sac. Zannichellia has a simple, biseriate wall (Fig. 59D). The swelling of a small cluster of cells over the septum ruptures the wall (Fig. 59E). In 142 MORPHOLOGY OF THE ANGIOSPERMS Najos, the simple-walled anther is enclosed by two sheathing structures. Freeing of the pollen is by a forcing apart of the enclosing sheaths, the outer one by growth of the receptacle and peduncle below the flower. Whether the single wall layer of Nojas represents the epidermis, the fibrous layer, or some sheathing organ is not known. The Tapetum. The primary sporogenous cells enlarge radially and divide periclinally. The outer cells so formed constitute the tapetum, a layer that encases the sporogenous cells and serves as a nutritive tissue for the pollen mother cells and the microspores (Fig. 67). The layer is typically uniseriate, rarely biseriate, but often irregular in outline and thickness, especially on the inner side of the sporogenous tissue. Tapetal cells may be formed also by transformation of outer sporogenous cells and, occasionally, from isolated sporogenous cells. The tapetum is described as derived, in some taxa, from the innermost layer of primary wall cells; in other taxa, partly from this layer and partly from the sporogenous cells. On the inner side of the sporogenous cell mass, it may arise from parenchyma of the connective and here complete the enclosure of the fertile cells. The tapetal cells enlarge rapidly, and their cytoplasm becomes denser. The tapetum is commonly uninucleate but may become bi- or multinucleate. The presence of many nuclei has suggested amitotic division, but an appearance of amitosis is perhaps given by incomplete mitosis. The mature tapetal cells may be uniformly uni-, bi-, or multinucleate or may vary in number of nuclei. At the time of spore formation and immediately afterward, the tapetal cells break down, and their cytoplasm unites to form a tapetal plasmodiiim. The cytoplasm is freed in different ways — by disintegration of the walls or by extrusion through collapsing walls. Cells of the connective abutting on the tapetum may also break down, but it is uncertain whether their contents are added to the plasmodium. The tapetal plasmodium extends among the spore tetrads or the free spores, isolating them singly or in clusters. As the spores mature, the plasmodium nourishes them and builds their walls. Two types of behavior of the tapetal cells in the development of the pollen grains are distinguished: the amoeboid and the secretory. In the amoeboid type, the protoplasts of the tapetal cells enlarge and intrude among the spore mother cells, spores, and young pollen grains, before merging as a periplasmodium; in the secretory type, the tapetal cells, without collapsing, become rich in content and "secrete" a periplas- modium. In both types, the periplasmodium ultimately surrounds the maturing spores or pollen grains. Four types of amoeboid formation of the periplasmodium have been distinguished, but these differ perhaps too little to be important. In the Sagittaria type, the plasmodium develops rather late — about the time THE STAMEN 143 the microspores are freed from the mother-cell wall — by the intrusion of "tongues" of cytoplasm from the individual tapetal cells among the spores. The nuclei of the tapetal cells migrate into the tongues later, as the processes develop among the pollen grains. Ultimately, the tongues merge to form the periplasmodium, in which the embedded pollen grains enlarge greatly. Vacuoles appear in the periplasmodium, and the nuclei and remnants of protoplasm disappear as the exine de- velops on the grains. The Butomus type of plasmodium is similar to the Sagittaria type but appears earlier, while the spores are still in tetrads. The enlarging tapetal-cell protoplasts press among the spores and unite, embedding them. In the Sparganium type, the tapetal cells — 4-nucleate and vacuolate — enlarge and free their protoplasts at the time of the first reduction division of the spore mother cell. Tetrad formation occurs in the nucleus-rich plasmodium. In the Triglochin type, the tapetal cells disintegrate where they lie, and the cytoplasm and later, during the reduction divisions, the nuclei push between the pollen mother cells. The secretory type of tapetum, like the amoeboid type, varies in be- havior; the cell walls disappear or collapse as their cell contents merge in a plasmodium on the periphery of the fertile tissue. Forms of taxonomic significance can perhaps be distinguished, but too little de- tailed information is yet available. This type occurs in the higher mono- cotyledons and in many of the dicotyledons. It has been called the primitive type, but the amoeboid type, which characterizes the lower monocotyledons generally and is found in some of the lower dicotyle- dons, is probably primitive. The resemblance of this amoeboid peri- plasmodium to that of some of the lower vascular plants supports this view. The formation of a peripheral layer in situ — the Triglochin type — is the simpler and apparently the most advanced method. Placentoids. Projections of sterile tissue into the sporogenous tissue have been called placentoids. These projections may be large lobes of connective parenchyma on the inner side of the sporangium — much re- sembling the placentae of carpels, with which they were once con- sidered homologous — or plates of laminar tissue extending into or through the sporangia, dividing it more or less completely. In the Gentianaceae and Menyanthaceae, sterile cells break up the sporogenous tissue; in Limnanthemum and Menyanthes, they surround clusters of fertile cells and, in Gentiana, individual fertile cells. In the Onagraceae, two to six spore clusters are set apart in a sporangium; in some Orchidaceae (Phajus), Rhizophoraceae, Loranthaceae, and Mimoseae, there are several to many well-separated clusters. Transverse partitions give the appearance of multisporangiate anthers — Butomus. The parti- tions may become tapetal wholly or in part, and the groups of 144 MORPHOLOGY OF THE ANGIOSPERMS sporogenous cells formed in tliis way may persist together, even to the mature pollen stage. Large, so-called placentoids are merely lobes of connective tissue, not division walls. The sporangial position where all four sporangia lie in one plane, as in Sassafi'as and other lauraceous genera, suggests placentoid division of two sporangia but is the result of major displacement of sporangia by differential growth in the de- veloping anther. Placentoids occur in rather few taxa, chiefly in Labiatae, Acanthaceae, Bignoniaceae, Solanaceae, Scrophulariaceae. Ontogeny of Spore Mother Cells. The primary sporogenous cells may differentiate directly as spore mother cells — many Compositae, Labiatae, Malvaceae, Portulacaceae — or may divide, usually sparingly, and their daughter cells become spore mother cells. Where division does occur, the adjacent tapetal cells may also divide, maintaining the tapetal sheath about the enlarging cell mass. The spore mother cells (here "pollen mother cells") rapidly increase in size. The nuclei enlarge, the cytoplasm becomes dense, and the walls thicken. Wall thickening is uneven in cells which form pollen grains that remain in tetrads or larger clusters; the outer walls of the cluster are thicker than the internal ones. Microspore Formation. Microspores are formed from the spore mother cells in tetrads by two closely successive divisions, the meio'tic or re- duction divisions, during which the chromosome number is reduced from 2n to n. Meiosis is nearly simultaneous throughout a sporangium or anther sac, except in very long anthers, as in Liriodendron, where there is an acropetal succession in stages of pollen development. Meiosis follows two somewhat different methods: the simultaneous method, where no wall is formed after the first division, the two divisions being almost simultaneous, with four free nuclei in the mother-cell protoplast; and the successive method, where a wall is formed after the first di- vision, and each of the cells so separated is then divided. The simul- taneous method is characteristic of dicotyledons and the successive, of monocotyledons, but there are many exceptions. There is probably little or no phylogenetic significance in these differences. Cell division in spore formation also takes place by cell-plate formation or by furrow- ing, sometimes called constriction, which occurs by means of the divi- sion of the protoplast by the extension inward from the mother-cell wall of wedges of new wall, as the protoplast contracts after nuclear division. There seems to be much variety in relationship of mother-cell wall and the walls of the four spores. The mother-cell wall may remain in- tact, even until the pollen grains are mature, enclosing them when they are shed as clusters, or may disappear early or late in pollen matura- tion. Where the spores remain together as compound pollen grains and THE STAMEN 145 as poUinia, the abutting walls are said to remain uncutinized. The ex- tent to which the mother-cell wall takes part in the formation of the pollen-grain wall by secretion or by modification and adnation to the spore wall probably varies greatly. The wedge-shaped projections that furrow the cytoplasm have been claimed to be formed within a gelatinous inner layer of the mother-cell wall. The spores are variously placed in the tetrad (Fig. 64); those formed by the simultaneous method are usually in tetrahedral positions; those formed by the successive method, in isobilateral positions. The decus- sate arrangement is occasional; the linear, rare. The T position is also rare and is probably abnormal. More than one type of arrangement may occur in a species. The spore mother cell usually forms four spores but may form more or less than four — two, when the second division fails to take place; three, when one of the first pair of cells fails to divide; more than four (pohjspory), occurring chiefly in hybrids, where divisions are irregular or incomplete. Of four formed, one only may mature; the other three degenerate, often remaining as three vestigial nuclei beside the surviv- ing spore inside the mother-cell wall — Cyperaceae (Fig. 60). Polyploid nuclei and other complex nuclear conditions may result from failure of wall formation in meiosis and from fusion of spore mother cells. The Microspore Wall. The microspores, at first enclosed by a delicate wall, which may be deposited upon the inside of the mother-cell wall wherever the spore abuts upon it, lie within the mother-cell wall, usu- ally completely filling it. The mother-cell wall may hold the tetrad to- gether for a time, even permanently when the pollen grains remain in tetrads; or it may gelatinize and merge with the spore wall to form a part of the pollen-grain wall. Sometimes, it is apparently absorbed. The walls of the microspores enlarge, expanding and thickening by intussusception. The exine — outer layer — is thick and its function is protective, both mechanically and physiologically. It is adapted to changes in volume of the grain — expansion and contraction with changes in humidity. The intine — inner layer — is thin and delicate and also readily adapts itself to changes in size of the pollen grain. The wall may become very thick and complex in structure, and the outer layer become sculptured with projecting ridges, spines, and granules ("orna- mentations"). Inner and outer layers, usually increased in wall thick- ness as the spore enlarges, are distinct; the inner is cellulosic, the outer cutinized. In the Epacridaceae, however, the wall of the young spore is thick and becomes thinner as it enlarges. The tapetal plasmodium aids, in large measure, in the increase of size of the spores and, especially, in the thickening of the wall and the building of the projec- tions. When this thickening and ornamentation is completed, the spore 146 MORPHOLOGY OF THE ANGIOSPERMS is mature. The spore, after germination, when the first cells of the male gametophyte appear within the wall, becomes the pollen grain. A so-called "third layer" of the anther wall — the layer adjacent to the tapetum — has been called the sporangium wall and persists in only some taxa. But this layer is prominent and persists only in thin-walled, highly specialized anthers, not in primitive stamens, where it would be expected to persist if it were a vestige of a true sporangium wall. B Fig. 60. Scirpus, germination of microspore mother cell, three of the daughter cells degenerate. A, S. lacustris: 1, the four daughter cells within the mother-cell wall; 2, one nucleus enlarging, three beginning to degenerate; 3, the surviving cell has divided, the other three reduced to remnants. B, S. uniglumis: 1, surviving daughter cell has divided (surrounded by phragmoplast ) , the other three spores aborted; 2, later stage of B\, vegetative and generative nuclei maturing. {After Piech.) THE STAMEN 147 Morphological Nature of the Stamen The basic nature of the sporophylls is discussed in Chap. 3 but is further treated here in the hght of the comparisons of form and struc- ture discussed in this chapter. The primitive form of the stamen is ob- viously laminar, like that of the carpel. (Under the New Morphology, the terete-filament type has been called primitive; the broad filament is interpreted as specialized.) In ontogeny, the broader types have mar- ginal meristems like those of carpels and leaves. In specialization, the stamen has been modified more extensively than the carpel and the laminar form largely lost. Sporangium position has also been greatly modified, in such a way that two of the four sporangia appear to be morphologically on the side of the sporophyll opposite its original posi- tion, and all four, from an originally sunken position, appear to have become superficial. But, morphologically, all the sporangia are on one side of the sporophyll and are embedded in sporophyll tissues, even though only lightly so in highly specialized anthers. The change in the microsporangium position is largely phylogenetic, but, in some taxa, is still in part ontogenetic (Fig. 57). The arrangement of sporangia in two pairs has been looked upon as persisting evidence of ancient telomic structure, but the laminar form, clearly primitive in living forms, does not support the view that the pairs of sporangia represent fused terminal telomes. The vascular struc- ture of the laminar stamen shows no evidence of a basic telomic struc- ture. The sporangia lie in pairs between the midrib and the lateral veins and have no connection with the vascular meshwork; if the sporophyll consists of telomes, the sporangia should terminate vascular bundles. Relationship of the simple and the fascicled stamen is discussed under androecium. The Peltate Stamen. In the older, descriptive literature, stamens in which more or less free, dorsifixed, or versatile anthers stand somewhat oblique to the filament — a common condition in highly specialized anthers — were called shield-shaped or peltate, but this term did not continue to be used in so simple a sense. Its revival in the early decades of the twentieth century came with its use in extended and modified senses. In one treatment, stamens fall in two major classes: peltate (shield-shaped) and impeltate, or epeltate (not shield-shaped). In impeltate stamens, the connective is the direct continuation of the filament; in peltate stamens, the filament is attached laterally to the anther, which usually stands at an angle to the filament. The impeltate stamen is arrow-shaped if the anther sacs extend beyond the attach- ment of the filament, and not arrow-shaped if the sacs do not so project; 148 MORPHOLOGY OF THE ANGIOSPERMS it is unseparated if there is no clear limit of anther and filament. (Ap- parently, no term is proposed for the stamen where the sacs project at both ends of the connective.) Peltate stamens are epipeltate if filament attachment is on the ventral side; hypopeltate, if on the dorsal side. In the second, more elaborate and morphologically more complex interpretation of "peltate" form in the stamen, the stamen is considered not merely shield-shaped but, like the carpel, a hollowed organ, ascidi- form or utriculate. (For a general discussion of the peltate theory as applied to sporophylls, see Chap. 5.) In the stamen, typical hollowed forms do not exist, but examples have been incorrectly described as evidence that they do. Evidence of peltate form in the stamen is seen, under this interpreta- tion, (1) in ontogeny, where the early form is called peltate; (2) in the so-called three-dimensional (as contrasted with so-called two-di- mensional, dorsiventral ) form; (3) in organs transitional from stamen to petal; and (4) in petals. Ontogenetic evidence is seen, under this interpretation, in the early ontogeny where the anther is prominent and may stand somewhat obliquely on a basal plate, which, much later, becomes the filament. The oblique position suggests the form of a simple, peltate leaf. (This obliqueness is, actually, merely the early ex- pression of later extrorse or introrse dehiscence.) A slight median fur- row, often present at the apex, is considered, under the peltate theory, a vestige of the hollow of an ancestral utriculate stamen (Fig. 61a). If the bulging anther base overhangs the filament primordium, as in arrow- shaped anthers (Fig. 61e), the enclosed space is said to represent a vestigial cavity. (These minor indentations are merely beginnings in the elaboration of anther form, not vestiges of ancestral cavities; the distal hollow is the beginning of the transverse furrow separating the anther sacs; the lower hollow is formed by differences in diameter of anther and filament, which are evident at an early stage.) The stamen is looked upon, under the peltate theory, as a "three-dimensional" organ, because of the thickness of the anther and the arrangement of the four sporangia in four corners. But this structure is the result of the specialization of the anther, the bringing of two of the four sporangia, originally in a single plane, into a second plane. In some ranalian genera, there are distal hollows in organs transitional between stamens and petals, as in the petals of Coptis, Hellcborus, Aquilegia, Delphinium, Aconitum. These hollows are secretory areas and definitely glands in petals or staminodes (sterile stamens); they are not vestiges of the cavity of a tubular, an- cestral stamen, as suggested under the peltate theory (see Chap. 6). The best evidence that the stamen is not fundamentally a peltate, three- dimensional organ is the type of primitive stamen clearly shown by the Ranales — a simple, dorsiventral, laminar organ. THE STAMEN 149 The theory that peltate form is fundamental in angiosperm sporo- phylls closely parallels the obsolete theory of carpel polymorphism in its lack of sound morphological basis, lack of convincing evidence, and continuing extension to include all appendages. The "Diplophyllous" Stamen. Closely related to the theory that the stamen is basically so-called three-dimensional (tubular or four-angled) is the tlieory that it is fundamentally double in nature, consisting of two Fig. 61. Diagrams to show, under the peltate theory, the origin of stamen from young peltate leaf, a, peltate primordium from which develop the types h, c, d, e, each with a median furrow; f to h, i to /, median longitudinal sections of a carpel or stamen primordium to show two series of stages in the development of the utriculate form by growth of the cross-zone meristem x. Dotted lines outline spo- rangia. (After Baum, 1949.) united laminae. This theory is suggestive of the old idea that the angio- sperm stamen represents two conifer microsporophylls, connate back to back. Evidence in support of this later theory is seen largely in organs transitional from stamen to petal and in so-called diplophyllous ("four- winged") petals, such as occur in teratological forms in the Ranuncu- laceae, Rosaceae, Saxifragaceae, and Myrtaceae. The so-called double organ is four-winged, suggesting two laminae united by their mid- ribs, and having a common median vascular vein. The abaxial half of the double organ is seen as the true leaf blade; the adaxial half, a pro- 150 MORPHOLOGY OF THE ANGIOSPERMS longation of the "cross zone." The organ is considered fundamentally tubular (peltate), even though it has a vascular core rather than a central cavity. The distribution, orientation, and type of the lateral veins in the sta- mens of some taxa have been considered proof that stamens are basically diplophyllous in nature; the filaments in the Nymphaeaceae show some vascular bundles that are inverted and amphicribral or bicollateral ( Fig. 55). But study of the complete vascular system of the stamens in the Nymphaeaceae shows that the position, orientation, and type of these bundles are the result of crowding and twisting as the broad ancestral sporophyll is narrowed and, consequently, displaces the bundles. The in- version of some of them is like the inversion of the sporangia which have been shifted to the opposite side of the sporophyll. The inverted bundles give no support to the theories of the peltate or diplophyllous nature of the stamen. The presence, in some families, of organs transitional from stamen to petal and the supposed, basically diplophyllous nature of both are considered, under the diplophyllous theory, evidence that stamens have been derived from petals. Support for this interpretation — the reverse of the generally accepted theory — is found in the view that simplicity of form is primitive, complexity advanced. That simplicity may represent reduction from complexity is disregarded. The Telome Theory as Applied to the Stamen. Under this hypothesis, the typical stamen is considered to represent, morphologically, a dichot- omous system of telomes, with terminal sporangia greatly reduced and compressed, and with only two distal dichotomies and their sporangia surviving. Fascicles of stamens are seen to represent the survival of several or many branchlets; the basal axis is greatly shortened, and the distal tips are brought together in pairs. Support for this view is seen in the side-by-side pairs of sporangia of the typical anther and in the presence, in some fairly primitive orders — Dilleniales, Parietales, Mal- vales — of a strong "trunk" vascular bundle which constitutes the central axis of the telome system and supplies several or many stamens. The pairs of sporangia, characteristic of stamens, naturally suggest dichotomies, and the pairing is perhaps even more prominent in the laminar stamens where all the sporangia lie in one plane. (The two- plane position of the sporangia in the typical anther has clearly been derived from the one-plane position. ) The pairing of the microsporangia, prominent in all stamens and especially in the primitive laminar types, contrasts strongly with the basic, scattered position of the megaspo- rangia. If the dichotomy of telomic structure persists in the sporophyll, some similarity should be found in the megasporophyll. And it is per- haps significant also that the outer member of each pair of sporangia is THE STAMEN 151 smaller than the inner, a character that is so deep-seated that it is present in most highly specialized anthers. The vascular structure of stamens, as a whole, gives no support to the hypothesis of telome structure in the stamen; no vascular tips lead to or directly toward the individual sporangia. (Vascular branches are frequent in the connective of some taxa but rarely penetrate the wall between members of the pairs.) The stamen of Ricinus has often been cited as an example of solitary sporangia terminal on dichotomizing branches of the stamen, but the position of these sporangia is now better understood. The anther in this genus has remarkable form; the connective is scalelike, subtending the stalked unisporangiate anther lobes (Fig. 44). The telomic theory has been considered to aid in the interpretation of stamen fascicles with their complex structure; under this theory, the fascicle can be interpreted as a compacted system of telomes. But, in the vascular system of a fascicle, there is no evidence of dichotomy. And comparative anatomical study of fascicles in several families, espe- cially in the Dilleniaceae and Paeoniaceae, shows that the fascicle is an aggregation of simple stamens (Fig. 46). Staminodia Staminodia are sterile floral organs that apparently represent stamens morphologically, resembling them more or less in form and closely in relation to the receptacle and to other floral organs. They have great variety of form. In most angiosperms, the petals are, morphologically, elaborated, sterile stamens and could well be called staminodia. Some taxa have staminodia which are transitional in form and loss of fer- tility from stamen to petal, and no line can be drawn between stamen and staminodium and between staminodium and petal — in genera of the Nymphaeaceae, Calycanthaceae, Magnoliaceae. Some staminodia are secretory organs, stamens or petals that have become primarily nec- taries — Coptis. Vestigial stamens, stamens in process of reduction and disappearance, are commonly called staminodia. The remnants of sta- mens, often mere stubs, in pistillate flowers are good examples of this type. Staminodia, typically, have the vascular structure of stamens in more or less reduced form. Where the stamen has lateral traces, these may be weak or absent. Staminodia that are mere remnants of reduced sta- mens may have no vascular supply or a mere basal stub. Vascular stubs of the traces of stamens entirely lost externally may be present in the floral receptacle, as in the flowers of the Scrophulariaceae and in the pistillate flowers of some unisexual genera. 152 MORPHOLOGY OF THE ANGIOSPERMS Nectaries Nectaries are surface areas, emergences, or organs where nectar is secreted. Floral nectaries are, morphologically, of two types: localized areas where nectar is secreted, or organs transformed from their original form and function. Nectaries that are modified organs may have re- stricted secretory areas or be secretory over most of their surface. Secretory areas, such as those on the receptacle, may be superficial only or may be proliferated, forming emergences that suggest organs. Re- duced organs, most commonly stamens or petals, are often represented by nectaries. These nectaries may have extensive secretory surfaces — some of the Proteaceae — or only minor nectariferous areas — Coptis and Salix. Superficial nectaries may be minor or major areas on organs other- wise largely normal — Aquilegia, Ranunculus, Viola. Septal glands, nec- taries on the abutting and partly connate walls of carpels in syncarpous gynoecia, are perhaps the most elaborate of superficial nectaries. They are characteristic structures of many Liliaceae, Amaryllidaceae, Palmae, Bromeliaceae. Where the carpels are not fused by their external edges, the nectar is excreted along a slitlike lateral or distal opening. Where the marginal areas are fused, the nectar flows out through a tubelike pas- sage to a small aperture on the top of the ovary. Direction of dehiscence of anthers is biologically correlated in insect pollination with position of nectaries: extrorse where the glands are below the stamens, introrse where they are above the stamens. Histologically, the secretory tissue of a nectary consists of subepi- dermal, small, closely packed cells, with rich cytoplasmic content. The epidermis overlying this tissue varies in structure; rarely, its cells also are secretory. A cuticle, usually thin, is commonly present, al- though nectaries without a cuticle have been reported. Stomata, usually with enlarged openings, through which the nectar exudes, are present in many nectaries. In other nectaries, the nectar diffuses through the epidermis and cuticle. Rarely, a nectary may consist of a secretory area in the epidermis alone, where the cells are papillose or replaced by multicellular hairs. The morphological nature of nectaries can best be determined by comparisons with the flower structure of related taxa and by anatomical structure. The number, origin, and type of vascular bundles supplying them give important and usually definite evidence of their nature. The traces of nectaries that represent organs are few and well defined; the vascular supply of superficial nectaries is diffuse, with many minor bundles from several or many points of origin, which may be on more than one floral organ; for instance, on receptacle and ovary, on stamen and petal. The nature of nectariferous discs, especially ring-shaped discs, THE STAMEN 153 can often be determined only by the number and position of the major vascular bundles. Nectaries are present in many places in the flower, on all the floral appendages and, frequently, on the receptacle. No structural nectary occurs in the primitive flowers of many woody Ranales. Their absence in "simple" flowers, such as those of the Amentiferae, has been con- sidered support for the view that anemophily is the primitive method of pollination. But these flowers are obviously greatly reduced, and lack of nectaries may represent loss of these structures. The general story of development of nectaries can be seen in living forms. Where pollination is by beetles exclusively, or nearly so — Eupomatia, CaJijcanthus — there are no nectaries; the insects feed on nonsecretory "food bodies" (Fig. 69). (Superficially and in cell content, these food bodies may resemble nectaries.) In the Magnoliaceae {Magnolia, Talauma), nectar is secreted, but not in localized areas; it is described as diffused through petal cuticles and excreted through enlarged stomata on petal bases and carpel surfaces. The entire flower of Magnolia has been called a nectary. In some of the Nymphaeaceae, a primitive family, nectar is reported secreted by the perianth. Since a small amount of nectar has been reported in one species of Magnolia and one species of Calijcanthus, genera which have food bodies and beetle pollination, nectar secretion apparently originated while the food-body method existed. It has been suggested that, in the evolution of the flower, a pro- gressive change has occurred in nectary position, from the base of the flower and the receptacle inward and upward on the appendages. In primitive dicotyledons, nectaries occur on the perianth and on the outer parts of the receptacle; in the higher families, they occur chiefly on the sporophylls or on the "receptacle" (fused perianth and sporo- phylls) around and above the ovary. In the monocotyledons, the nec- taries are on the perianth, stamens, and especially the carpels (septal nectaries). Differences in the morphological nature of the nectaries seem not to have been considered in this view of a phylogenetic change in nectary position, but a general change is evident. Change in nectary position necessarily accompanies fusion of appendages, and, in advanced floral types, nectaries are present higher up in the flower. The location of nectaries that represent reduced floral appendages may be of much importance in the determination of phylogenetic rela- tionships. For example, the nectariferous disc of some genera of the Proteaceae (four separate organs in other genera) is, from the evidence of comparative form and anatomy, a vestigial corolla, and this family can no longer be placed in the supposedly primitive Apetalae. The nec- taries of the Salicaceae, which are reduced perianth parts, demonstrate that the flowers of this family are not primitively simple. 154 MORPHOLOGY OF THE ANGIOSPERMS The variety in form, position, and morphological nature of nectaries indicates that these structures have arisen independently in many angio- sperm lines. Position and type of nectary are surely important in phylo- genetic studies. Flowers now generally accepted as highly primitive — those of the Eupomatiaceae, Himantandraceae, Magnoliaceae, Calycan- thaceae, Nymphaeaceae — are chiefly or entirely pollinated by beetles and have no nectaries. Nectar glands are characteristic of angiosperms, ex- cept those that are pollinated by beetles and wind, BIBLIOGRAPHY The Stamen and Androecium Aboy, H. E.: A study of the anatomy and morphology of Ceratophxjllum demersum, Unpubhshed thesis, Cornell University, 1916. Arber, A. On the structure of the androecium in Farnassia and its bearing on the affinities of the genus, Ann. Bot., 27: 491-510, 1913. Bailey, I. W., and C. G. Nast: The comparative morphology of the Winteraceae. I. Pollen and Stamens, Jour. Arnold Arh., 24: 340-346, 1943. Baillon, H.: Recherches organogeniques sur les Eupomatia, Adan.sonia, 9: 22-28, 1868-1870. Baum, H.: Beitriige zur Kenntnis der Schildform bei den Staubblattern, Oesterr. Bot. Zeitschr., 96: 453-466, 1949. : Die Unabhangigkeit der diplophyllous Gestalt der Staubblattspreite von ihrer Funktion als Trager der Pollensacke, Oesterr. Bot. Zeitschr., 100: 265- 269, 1953. Brouland, M.: Recherches sur I'anatomie florale des Renonculacees, Le Botaniste, 27: 1-278, 1935. Canright, J. E.r The comparative morphology and relationships of the Mag- noliaceae. I. Trends of speciahzation in the stamen. Am. Jour. Bot., 39: 484-497, 1952. Celakovsky, L.: Uber den "eingeschalteten" epipetalen Staubbhittkreis, Flora, 58: 481^89, 497-504, 513-524, 1875. : Teratologische Beitrage zur morphologischen Deutung des Stamengef asses, Jahr. Wiss. Bot., 11: 124-174, 1878. Challenger Expedition: Rpt. Sci. Results Bot., III. Rpt. Bot. Juan Fernandez and Masafuera, 1884. Chatin, A.: "De I'Anthere." Paris, 1870. Ghifflot, J. B. J.: Contributions a I'etude de la classe des Nympheinees, Ann. Lyons Univ., n.s., I (10): 1-294, 1902. Clausen, P.: Uber das Verhalten des Antheren-Tapetum bei einiger Monocotylen und Ranales, Bot. Archiv, IS: 1-27, 1927. Cooper, D. C: Nuclear divisions in the tapetal cells of certain angiosperms. Am. Jour. Bot., 20: 358-364, 1933. Corner, E. J. H.: Centrifugal stamens, Jotir. Arnold Arh., 27: 423-437, 1946. Daumann, E.: Zur morphologischen Wertigkeit der Bliitennektarien von Laurus, Beih. Bot. Centralhl, 48 (I): 209-213, 1931. Diels, L.: Uber die Gattung Himantandra, ihre Verbreitung und ihre systematische Stellung, Bot. Jahrb., 55: 126-134, 1917. Eames, A. J.: Floral anatomy as an aid in generic limitation, Chron. Bot., 14: 126-132, 1953. BIBLIOGRAPHY 155 Eckardt, T.: Untersuchungen iiber Entwicklungsgeschichte und systematische Bedeu- tung des pseudomonomerous Gynoecium, Nova Acta Leopold., N.F., 5: 1-112, 1937. Engler, A.: Beitrage zur Kenntnis der Antherbildung der Metaspermen, Pring- sheim's Jahr. Wiss. Bot., 10: 275-316, 1876. Frye, T. C: A morphological study of certain Asclepiadaceae, Bot. Gaz., 34: 389- 412, 1902. Guerin, P.: Le developpement de I'anthere chez les Gentianacees, Bull. Soc. Bot. France, 2 ser., 72: 5-18, 1926. Hannig, E.: Kritische Untersuchungen iiber das Vorkommen und die Bedeutung von Tapeten und Periplasmodien, Flora, N.F., 2: 335-^382, 1911. : Uber Bedeutung der Periplasmodien, Flora, N.F., 2: 209-278, 1911. Hirmer, M.: Beitrage zur Morphologic der polyandrischen Bliiten, Flora, 110: 140- 192, 1918. Hjelmquist, H.: Studies in the floral morphology and phylogeny of the Amenti- ferae, Bot. Not. Suppl. 2: 1-171, 1948. Howard, R. A.: The morphology and systematics of the West Indian Magnoliaceae, Bull. Torreij Bot. Club, 75: 335-,357, 1948. Ivancich, A.: Der Bau der Filamente der Amentaceae, Oesterr. Bot. Zeitschr., 56: 305-^09, 385-394, 1906. Janchen, E.: Die sogenannte Schildform der jungen Staubgefasse, Phijton. {Ann. Rei Bot.), 2: 267-270, 1950. Kasapligil, B.: Morphological and ontogenetic studies of Umbellularia calif ornica Nutt. and Laurus nobilis L., Univ. Calif. Ptibl. Bot., 25: 115-240, 1951. Kaussmann, B.: Vergleichende Untersuchungen iiber die Blattnatur der Kelch — , Blumen— , und Staubblatter, Bot. Archiv, 42: 503-572, 1941. Leclerc du Sablon: Recherches sur la structure et la dehiscence des antheres, Ann. Sci. Nat. Bot., 7 ser., 1: 97-134, 1885. Leinfellner, W.: Die petaloiden Staubblatter und ihre Beziehungen zu den Kron- blattem, Oesterr. Bot. Zeitschr., 101: 373-406, 1954. : Die blattartig flachen Staubblatter und ihre gestaltlichen Beziehung zum Bautypus des Angiospermen-Staubblattes, Oesterr. Bot. Zeitschr., 103: 247- 290, 1956. Inwieweit kommt der peltat-diplophylle Bau des Angiospermem-Staub- blattes in dessen Leitbiindelanordnung zum Ausdruck? Oesterr. Bot. Zeitschr., 103: 381-399, 1956. Matthews, J. R., and E. M. Knox: The comparative morphology of the stamen in the Ericaceae, Trans. Bot. Soc. Edinburgh, 29: 24.3-281, 1929. Money, L. L., I. W. Bailey, and B. G. L. Swamy: The morphology and relationships of the Monimiaceae, Jour. Arnold Arb., 31: 372-404, 1950. Moseley, M. F., Jr.: Morphological studies in the Nympheaceae. I. The nature of the stamens, Phtjtomorph., 8: 1-29, 1958. Neumann, R. : tjber Antherae anticae und posticae vmd deren tjbergiinge in einander, Bot. Zeit., 11: 353-363, 371-383, 399-403, 1854. Ozenda, P.: Remarques sur quelques interpretations de letamine, Phtjtomorph., 2: 225-231, 1952. Parkin, J.: The protrusion of the connective beyond the anther and its bearing on the evolution of the stamen, Phtjtomorph., 1: 1-8, 1951. Piech, K.: Uber die Entstehung der generativen Zelle bei Scirpus uniglumis Link durch "freie Zellbildung," Planta, 6: 96-117, 1928. Pijl, L. van der: The stamens of Ricinus, Phtjtomorph., 2: 130-132, 1952. Salisbury, E. J.: Floral construction in the Helobiales, Ann. Bot., 40: 419-445, 1926. 156 MORPHOLOGY OF THE ANGIOSPERMS Schaeppi, H.: Vergleichend-morphologische Untersuchungen an den Staubblattem der Monocotyledonen, Nova Acta Leopold., N.F., 6: 389^47, 1939. Schumann, K. : Bluthenmorphologische Studien: Die obdiplostemonen Bliithen, Jahr. Wiss. Bot., 20: 349-426*^ 1889. Schwarze, C: Vergleichende entwicklungsgeschichtliche imd histologische Unter- suchungen reduzierter Staubbliitter, Jahr. Wiss. Bot., 54: 183-243, 1914. Shoemaker, D. N.: On the development of Hatnamelis virginiana, Bot. Gaz., 39: 248-266, 1905. Smith, A. C: Studies of Papuasian plants. V. Jour. Arnold Arb., 23: 417-443, 1942. Stadtler, G.: tJber Reduktionserscheinungen im Bau der Antherenwand von Angio- spermen-Bluten, Flora, N.F., 116: 85-108, 1925. Swamy, B. G. L.: A contribution to the life history of Casuarina, Proc. Am. Acad. Arts and ScL, 77: 1-32, 1948. and I. W. Bailey: Sarcandra, a vesselless genus of the Chloranthaceae, Jour. Arnold Arb., 31: 117-129, 1950. Thoday, D., and E. T. Johnson: On Arceuthobium pusillum Peck. II. Flowers and fruit, Ann. Bot., 44: 813-824, 1930. Thomson, B. F.: The floral morphology of the Caryophyllaceae, Am. Jour. Bot., 29: 333-349, 1942. Troll, W.: tJber Diplophyllie und verwandte Erscheiningen in der Blattbildung, Planta, 15: 355-406, 1931. Van Tieghem, P.: Structure de I'etamine chez les Scrophulariacees, Ann. Sci. Nat. Bot., 8 ser., 17: 363-371, 1903. : Sur les antheres symmetriquement heterogenes, Ann. Sci. Nat. Bot., 9 ser., 5: 364-370, 1907. Von Mohl, H.: "Beobachtungen iiber die Umwandlung von Antheren in Carpelle." In "Vermischte Schriften," pp. 28-44, Tiibingen, 1845. Warming, E.: Untersuchungen iiber pollenbildende Phyllome und Caulome, Bot. Abhandl, 2: 1-90, 1873. Vl^ilson, C. L. The phylogeny of the stamen. Am. Jour. Bot., 24: 686-699, 1937. : The telome theory and the origin of the stamen. Am. Jour. Bot., 29: 759- 764, 1942. : Vascularization of the stamen in the Melastomaceae, with some phyletic implications, Am. Jour. Bot., 37: 431-444, 1950. Nectaries Anderson, C. E.: Some studies on the floral anatomy of the Liliales, Thesis, Cornell University, 1940. Behrens, W. J.: Die Nectarien der Bliiten, Flora, 62: 2-11, 17-27, 49-64, 81-90, 113-123, 145-153, 23.3-247, 304-^14, 369^75, 443-457, 1879. Bonnier, G.: Les nectaires, Ann. Sci. Nat. Bot., 6 ser., 8: 1-212, 1879. Brongniart, A.: Memoire sur les glands nectariferes de I'ovaire, Ann. Sci. Nat. Bot., 4 ser., 2: 5-23, 1885. Brown, V^. H.: The bearing of nectaries on the phylogeny of flowering plants, Proc. Am. Phil. Soc, 79: 549-595, 1938. Daumann, E.: See reference under Pollination. : Zur Phylogenie der Diskusbildungen: Beitrage zur Kenntnis der Nectarien, Beth. Bot. Centralbl, 48: 183-208, 1931. Diels, L. : See reference under Pollination. BIBLIOGRAPHY 157 Fahn, A.: On the structure of floral nectaries, Bot. Gaz., 113: 464-470, 1952. : The topography of the nectary in the flower and its phylogenetic trend, Phtjtomorph., 3: 424-426, 1953. Grassmann, P.: Die Septaldriisen, Flora, 67: 113-144, 1884. Jaeger, P.: Nectaires floraux et phylogenese, Annee. Biol, (III) 54: 111-117, 1950. Jordan, K. F. : Die Stellung der Honigbehalter und dcr Befnichtungs-werkzeuge in den Bhimen, Flora, 69: 195-226, 243-252, 259-274, 1886. Porsch, O.: Die Abstammung der Monocotylen und die Bliitennektarien, Ber. Dcutsch. Bot. Ges., 31: 580-590, 1914. Schniewind-Thies, J.: "Beitrage zur Kenntnis der Septabiektarien," Jena, 1897. Werth, E.: See reference under Pollen and Male Gametophyte. Wolff, G.: Zur vergleichenden Entwicklungsgeschichte und biologischen Bedeutung der Bliitennektarien, Bot. Archiv, 8: 305-344, 1924. Chapter 5 POLLEN In most angiosperms, germination of the microspore begins when the flower bud is small, with the enlargement of the spore and the first cell division within the spore. After one or two divisions within the spore wall, a resting period usually occurs. The spore wall, together with its contents of two or three cells in a dormant state, (an early stage of the male gametophy te ) , is the pollen grain (Fig. 62F). The microspore is mature long before the dehiscence of the anther, often while the anther is still essentially sessile and the filament undeveloped. In plants that flower in early spring, the anthers usually pass the winter with the sporogenous tissue in the mother-cell stage or, less commonly, in the microspore stage; rarely, the microspore has germinated. In these plants, development may continue in the warmer periods during the winter. Collectively, masses of pollen grains — individual grains or clus- ters of various sizes — constitute the pollen. Where the grains formed from the spores of a tetrad remain permanently together, the grains have been called "compound pollen grains" but are better described as "pollen grains in tetrads." Pollen grains show great variety of form, size, and sculpturing of the wall (Fig. 63). In form, they are usually globose, ellipsoid, or fusiform, but they may be lobed or angular. Their shape depends, in part, on their moisture content. Drying may greatly change their form, but, when moist, the grains will return to their original shape. Pollen may be shed in either the moist or dry state: that of many Rosaceae is described as shed in the moist state; that of the Compositae, in the dry state. The most extreme shape is confervoid — as in many submersed aquatic genera, Zostera, Thalassia, Fosidonia, Rtippia (crescent-shaped). Unusually large grains characterize the Cucurbitaceae and very small ones some of the Boraginaceae. Arrangement of Pollen Grains in the Tetrad. The pollen grains formed by a mother cell are associated in several ways (Fig. 64). The most common arrangements are the tetrahedral — the grains lying at the four corners of a tetrahedron — and the tetragonal — the grains lying at the corners of a square or rhombus. The tetrahedral arrangement is com- mon in the higher dicotyledons. In the formation of the spores, the tetrahedral arrangement usually results from simultaneous divisions; 158 POLLEN 159 Drimtjs is a prominent exception. The tetragonal arrangement results from successive divisions; this arrangement is frequent in monocotyle- dons, lower dicotyledons, and gymnosperms. Pollen grains are commonly shed individually or in loose clusters but may remain in tetrads — Tijpha, Droseraceae, Winteraceae, Lactorida- Fig. 62. Microsporogenesis in Lobelia cardinalis. A, second meiotic division within the mother-cell wall; B, microspores nearly formed, partition walls developing; C, mature microspore; D, spore enlarged, first cell division completed; E, vegetative and reproductive nuclei mature; F, mature poUen grain with sperm nuclei. (After G. O. Cooper.) ceae, Juncaceae, several sympetalous families. They are held together sometimes by the persistent mother-cell wall, sometimes by close ap- pression or by adhesive surfaces. The grains may remain in tetrads even in pollinia. Rarely, they remain in pairs called "dyads" — Scheuchzeria. They may form pollinia or massulae, the compacted contents of indi- vidual sporangia, anther sacs, or fused anther sacs — Asclepiadaceae, Fig. 63. Pollen grains of angiosperms showing variety in form and sculpturing. A, Carex striata, side view; B, Liriodendron tulipifera, ventral view; C, Festuca elatior, side view; D, Populus sargentii; E, Agrostis palustris, ventral view, contracted; F, Fagus grandifolia, polar view, expanded; G, Castanca dentata, side view, fully ex- panded; H, Salix fragilis, polar view, expanded; /, Barnadcsia herheroides, polar view; /, Taraxacum officinale, polar view. {After Wodehouse, 1935.) 160 POLLEN 161 Orchidaceae. In the Mimoseae, pollen is often shed in masses of 8 to 64 grains. Structure of the Exine. The exine of the pollen grain is often sculptured in many and elaborate patterns, which characterize major or minor taxa (Fig. 63). Pollen grains that remain in tetrads have the surfaces of con- tact with other spores of the tetrad smooth or only lightly sculptured when exposed by enlargement at germination. The patterns have con- siderable taxonomic and phylogenetic value. Projections in the form of ridges, spines, and granules are the prominent features. The thin areas between the projections are of two types: germinal furrows and ger- minal apertures, or pores. The furrows, various in form and distinctness of limit, represent elongate areas where the wall is thin and elastic. They provide structural adaptation to changes in volume of the grain due to changes in atmospheric humidity. The apertures are small thin areas, r\ Fig. 64. Diagrams to show the types of arrangement of pollen grains in the tetrad. A, tetrahedral; B, cross; C, square; D, rhomboidal; E, linear. (After Erdtman, 1945.) which represent positions of possible emergence of pollen tubes. They are usually located within the furrows but may be present in the thicker wall areas. Apertures in some taxa are probably phylogenetically short- ened furrows. In walls of elaborate structure, the apertures may be covered with "caps" of thicker wall, which are readily lifted by pres- sure of the developing pollen tube — Cucurbitaceae. Where no apertures are present, the pollen tubes grow out from the furrows. Germinal fur- rows are absent in some families, and rounded, thin areas, pores, take their place functionally. These areas may be numerous — about thirty in the Polygonaceae. Specialization of these basic types within the angio- sperms has apparently followed several lines, with greatly modified types resulting — both more elaborate and simpler in form. Similarity of type has evolved by convergent changes in lines not closely related. Pollen-grain Types. There are apparently two major types of pollen grains: monocolpate (Fig. 63B, G) and tricolpate (Fig. 63F, H), with one and three germinal furrows, respectively. The monocolpate type is elongate or rounded, with the furrow on the side not in contact with the other grains in the tetrad. This seems to be the primitive form. It is characteristic of the monocotyledons and most of the Ranales, espe- cially the woody families — Winteraceae, Degeneriaceae, Magnoliaceae, 162 MORPHOLOGY OF THE ANGIOSPERMS Eupomatiaceae, Calycanthaceae, Lauraceae, and a few other families in Other orders (Piperaceae, Saururaceae, Chloranthaceae ) . Pollen grains of the monocolpate type occur also in the Cycadales, Bennettitales, and pteridosperms. The tricolpate grain, with three meridional furrows, is apparently the basic type in the dicotyledons. It is not known in other seed plants. Position, form, and number of furrows and other structural features of the exine are important diagnostic characters in the identification of pollen grains. Larger numbers of furrows and their position on the grain are related in part to the arrangement of the spores in the mother cell. The sohtary furrow in a distal position — distal as related to posi- tion of the spore in the tetrad — seems to represent a primitive form, from which have been derived, along one line, the proximal monocolpate, dicolpate, polycolpate, and acolpate (without a furrow) types; along another line, primitive tricolpate has led to polycolpate and acolpate. In the monocolpate line, specialization tends to eliminate the furrow^ both under reduction of the flower, with establishment of anemophily, as in the Gramineae, where the furrow is reduced to a small pore; and under extreme zygomorphy, with elaboration of corolla and androecium in entomophily, as in the Cannaceae and Musaceae. The end products of specialization in the pollen grain seem to be the acolpate and poly- colpate types. These have probably been derived from both monocolpate and tricolpate types. Some doubt arises as to the homology of the mono- colpate grain with distal furrow with the similar grain with proximal furrow — Annonaceae. Evidence from pollen morphology of apparent relationship among major taxa can be considered important only to- gether with that from other fields. The fundamental characters of the pollen grain are probably number and position of furrows, form and position of apertures, and pattern of sculpturing. Shape and size are probably of little or no basic importance, though very large and very small size characterize some taxa. Most large families show a considerable range in pollen characters, but pollen type in some is remarkably uniform, as in the Gramineae, where the grains are spherical or ellipsoid, with one furrow. The pollen of anemophilous plants is usually small, rounded, smooth, rather thin-walled, "dry" (nonadhesive), and with shallow furrows or none at all; it often becomes angular when dry. The pollen of plants distributed by insects and birds is large, sculptured, and often coated with an adhesive waxy or oily substance. Beetle-pollinated plants have simple, thick-walled Tpo\\en~Eupo7natia. Characteristic smooth, thin- walled pollen is present in the wind-pollinated Fopulus, Gramineae, Cyperaceae, Platanus, Plantago, Betulaceae, Fagaceae, Juglandaceae, Ambrosieae. The Compositae, as a whole, have perhaps the most elab- POLLEN 163 orately sculptured pollen, but, within the family, they show a series in simplification toward loss of sculpturing in the anemophilous genera (Fig. 65). Artemisia has smooth pollen; Xanthiiim has lightly sculptiu-ed walls, with weak spines. This series is considered evidence that ane- mophily has been recently established in the Compositae. Scattered, "vestigial" patches of an adhesive layer on wind-borae pollen have been considered evidence of the derivation of anemophily from en- tomophily. The sculpturing of the exine is clearly correlated with method of pollination. Fig. 65. Pollen grains of Ambrosieae. Diagrammatic equatorial sections showing pro- gressive thinning of exine and reduction in size of spines in a phylctic series from insect to wind pollination. A, Oxytcnia acerosa; B, Chorisiva nevadcnsis; C, Cijcla- chaena xanthifolia; D, Ambrosia elatior; E, Xanthiiim speciusum. (After Wodchoiise, 1935. ) Pollen-grain Development. Three types of pollen-grain development have been distinguished: the normal, Triglochin, and Jimcus types. In the normal type, characteristic of most angiosperms, the spore enlarges greatly in volume before dividing to form the generative and tube cells (Figs. 62C, D and 66A, B). In the Triglochin type (Fig. 66A to D), cell division occurs in the spore before there has been much increase in size; the major increase in size follows the first cell division. The Juncus type, restricted to the Juncaceae and Cyperaceae, seems in- sufficiently distinct from the other two. The Triglochin type is present in lower monocotyledons — Najas, Triglochin, Ruppia, Lilaea, Aponogeton. 164 MORPHOLOGY OF THE ANGIOSPERMS Sequence within the Sporangium. Development of the pollen grains within a sporangium and within an anther sac is commonly simultane- ous, or nearly so, but is successive from one part to the other in some taxa, especially the Ranales and Helobiales. The long anthers of Lirio- dendron show, from one end to the other, successive stages in pollen ontogeny. Palynology. The cutinized wall of spores and pollen grains is one of the most resistant of organic structures; it persists as a fossil in peat and Fig. 66. Ceratophtjllum submersum L. showing the Triglochin type of pollen-grain development. A, spore; B, C, generative cell formation and beginning of vacuolation before enlargement; D, E, increase in size and vacuolation. (After Wiilff, 1939.) in sedimentary rocks of many kinds and all ages — especially in coal and oil shales — when other plant material has been destroyed. Fossil spores and pollen grains are therefore of great importance in the study of the phylogeny of plants. Palynology, the study of spores and pollen grains, has recently been established as a morphological, taxonomic, and phylogenetic section of botanical science and has already contributed information of great value in the determination of relationships among angiosperms. For example, the pollen grains of the Casuarinaceae and of the Amentiferae in general are not of primitive type and support the present generally accepted view that these taxa are advanced, not primi- tive. Palynology is also of much value in studies of historical distribution and ecology. DEHISCENCE OF THE ANTHER 165 The pollen of the monocotyledons appears to be, in general, more primitive than that of the dicotyledons; the monocolpate type is dom- inant, and the tricolpate type not present. It has been suggested that the prominence of the monocolpate type is perhaps connected with phylo- genetic homogeneity, hygrophily, and geophily. The characteristic monocolpate pollen of the monocotyledons has been called "evidence of their antiquity," because a derivation from the tri- colpate Ranunculaceae — as often suggested for the monocotyledons — cannot be considered. The primitive monocolpate pollen of the mono- cotvledons is evidence at least of their very early origin, whatever their ancestral stock may have been, DEHISCENCE OF THE ANTHER Pollen is commonly shed from the anther tlirough longitudinal, slitlike openings in the anther or lamina wall — longitudinal dehiscence. Trans- verse slits occur occasionally — Alchemilla, Hibiscus, Euphorbia, Chrijso- splenium. Where the sporangia are united in pairs, the longitudinal opening follows the furrow between the sporangia of a pair along the line of sporangial fusion. Dehiscence by small, rounded openings, "pores" — poricidal or porosc dehiscence — is characteristic of a few families — Ericaceae, Epacridaceae, Tremandraceae, Melastomaceae (most genera), Myrsinaceae, some Leguminosae, Ochnaceae, Solanaceae. Pores are usually located at the distal ends of the sacs, rarely at the proximal ends. In the anthers of some Ericaceae, the distal ends are, morphologically, the proximal ends; the anther is inverted in ontogeny. Arrangement of the sporangia in pairs brings about variety in longi- tudinal dehiscence. In laminar stamens, the sunken sporangia often de- hisce independently; the sporangia have individual endothecial caps. In anthers where the sporangia lie close together in lateral pairs, dehiscence is usually by a common slit in the furrow between the two sporangia. As the pollen matures, the anther-sac wall separating the two sporangia breaks down, and the pollen grains mingle as they escape into the furrow (Fig. 67F). Where the four sporangia are isolated in the corners of an angular anther, dehiscence is from each sporangium separately and in a more or less lateral direction. Dehiscence results generally from hygroscopic shrinkage of the fibrous layer. It may be initiated in regions of existing structural weakness, such as absence of epidermis along the line of dehiscence. Where de- hiscence is longitudinal, continued drying brings about retraction of the borders of the openings, and the chambers are opened widely. Changes in atmospheric humidity may cause repeated opening and closing of the anther sacs. The pollen may all be freed at once or may gradually 166 MORPHOLOGY OF THE ANGIOSPERMS escape. In fusion chambers, the pollen grains probably do not inter- mingle, except as they escape. Dehiscence is not always due to changes in the fibrous layer, for this layer is not always present, especially in poricidal anthers. Details of dehiscence in anthers of this type vary considerably, but the pores usually form by disintegration of the anther wall in a small area, with shrinkage of the surrounding tissues. Rarely, a slit occurs in tlie epi- Fig. 67. Degeneria vitiensis showing development of microsporangia, spores, and pollen grains. A, B, groups of archesporial cells in cross section of the lamina; C, D, organization of the tapetum; E, sporangia with microspores and the beginning of endothecial caps, tapetum degenerated; F, pair of sporangia united by breakdown of intermediate tissue, two-celled pollen grains, and endothecium; G, microspore with germination furrow, dividing; H, pollen grain at shedding stage, generative cell dark, tube nucleus Hghter; /, young tetrad in mother-cell wall, showing initia- tion of furrow on distal face of grains; /, pollen grains of a tetrad, just separated, showing furrows; K, germinating pollen grain with tube projecting from the fur- row. (After Swamtj.) DEHISCENCE OF THE ANTHER 167 dermis where the pore will develop, and the opening is enlarged by shrinkage of the surrounding tissues — some of the Tremandraceae. Where the anther sacs have elongate tips, the sporogenous tissue in these tips may become sterile, forming delicate parenchyma cells, which dis- integrate when the pore opens and leave hollow tubes extending to the pore. Cellular breakdown may involve the connective tissue between the tips of the anther sacs, and the two sacs — four sporangia — open through one pore. Poricidal dehiscence has apparently been derived inde- pendentlv in different taxa from longitudinal dehiscence by shortening of the slit. It reaches high specialization where the pollen is freed from four sporangia through a single pore, as in Cassia. Forms transitional to longitudinal have some fibrous tissue, restricted usually to areas around the pore — Solanaceae. Association of poricidal dehiscence and a fibrous layer throughout the length of the sac is rare. Complex form in the anther is often associated with dehiscence. An- thers with poricidal dehiscence may have long, tubular projections ("awns") at the ends of the sacs, through which the pollen is gradually sifted out. In anthers with slitlike openings, the slit may be so shaped as to free a valvelike flap of tissue; this is valvular dehiscence — Lauraceae, Berberidaceae. In Hamamelis, the valve flap is folded out- ward and backward and carries the pollen mass out of the chamber as it lifts. In some aquatic genera with submerged flowers, the pollen is freed by disintegration of the anther sac. Although the general structure of the anther wall has already been discussed, the structure and distribution of the outer layers in their relation to dehiscence need further consideration. The epidermis is characteristically simple but, at anther maturit)', may show unusual structure. Its cells may be tabular and greatly flattened, appearing col- lapsed, as in Carina, Balsamea, and many Compositae. It may be lost during ontogeny, either wholly or in part — Vitis, Grevillea, Asarum, many Compositae. Along the line of dehiscence, it may be absent — Aristolochia, Nepenthes. A narrow band of cells may enlarge greatly when dehiscence approaches — Crocus, Iris, Bignonia — and form a spe- cialized opening stiucture, a stomium. The well-developed stomium of Lilium is frequently figured. A crest of thickened epidermal cells may develop along both sides of the line of dehiscence — Passiflora, Lycoper- sicum. Epidermal hairs in a fringe along the sides of the slit perhaps aid in the freeing of the pollen grains by hygroscopic movements. Thick- ening bars like those of the fibrous layer have been reported in the epi- dermis of Clandestina. The fibrous layer is obviously structurally adapted, by the uneven thickenings of its cell walls (Fig. 68), as a mechanism at least largely responsible for the opening of the anther sac. But there are families and B Fig 68. Fibrous layer of anther-sac wall associated with dehiscence (cells elongated fr^risve se to the Lher axis) in face view. A, Malva syvestris, inner tangential S bars stellate, no bars on outer wall; B, Borage ^f '-f\-"- ^^^1:^1" d reticulate, hgnilied; C, Lychnis dioica, inner tangential wall, bars U-shaped, curved on Inner wall and ar;is extending over radial walls; D, L. dwica, outer tangential wal w hout bars, showing bars o'n radial walls; E, Delphinium --"f ^' T^ ^J gential wall, bars fused scllariformly; F, Enjthraea Centaunum, ^/f^ll no barT layer, bars forming a mesh; G, H, Aquilcgia vulgaris, outer tangential wall no bars but showing sections of bars on radial wall; 1, Alopecurus agrestis, radial wall of epidermis. {After Leclerc du Sablon.) 168 POLLINATION 169 genera in which a typical fibrous layer is absent, and the sacs open by longitudinal breaks — Orchidaceae, Araceae, Asclepiadaceae, Oroban- chaceae, Diospyros. The anther wall is described as consisting of the fibrous layer only in Althaea, Vitis, PhijfeJephas. (The epidermis has been reported ephemeral in Vitis. ) The Orchidaceae and Orobanchaceae, considered to lack a typical fibrous layer, have some wall cells with sparse or incomplete bars; these cells probably represent vestiges of an ancestral fibrous layer. Nonfunctioning fibrous cells are also present in parts of the anther sac that do not open — Berberis. The shape of the cells of the fibrous layer and the position of the thickened bars are doubtless related to the functioning of the layer, though there seems to be no consistency in these features. The long axis of the cells is described as parallel to the slit in many genera — SiJene, Erodiiim, Plantago, Commelina; and as at right angles to the slit in many other genera — Pijnis, Tiarella, Lychnis, Geranium. Shape of the cells appears unimportant, since it differs in related genera — Silene and Lychnis; Erodiiim and Geranium. The thickening bars usually extend at right angles to the slit; only rarely are they parallel to the slit — Salvia. Any function in dehiscence of the "third layer," that below the fibrous layer, is probably indirect or accessory only. The small, rather loosely packed parenchyma cells of this layer commonly disintegrate just before deliiscence, freeing the outer layers from attachment to the con- nective below; this disintegration may aid indirectly in the contraction of the external tissues. The role of the fibrous layer has perhaps been exaggerated, though it is surely the important layer in this function. Apparently both this layer and the epidermis can function alone, but, in most families, the mechanism responsible for deliiscence seems to involve two, or all three, of the layers. Dehiscence of the anther is similar to that of the carpel in that it involves a sporophyll wall. POLLINATION It was long assumed that pollination in the angiosperms was primi- tively by wind, anemophily; that pollination by water, hydrophily, and by animals, entomophily and ornithophily, were derived methods. This conclusion was based largely on the presence of anemophily in the conifers and in simple, and therefore supposedly primitive, flowers, especially those of the Amentiferae. The opinion that the primitive method of pollination was by insects — though argued at the beginning of the twentieth century — received strong support only twenty to forty 170 MORPHOLOGY OF THE ANGIOSPERMS years later. The views now generally accepted are that the conifers are not on the direct ancestral line of the angiosperms; that the ane- mophily of the Amentiferae is derived, accompanying high specializa- tion in inflorescence and flower structure; that the primitive flower was bisexual, not unisexual. Without question, anemophily is advanced in some taxa: in Fraxinus, a member of a highly specialized entomophilous family with petaloid flowers; in the Platanaceae, a family of the petaloid Rosales, with greatly reduced flowers and inflorescences; in the grasses and sedges, likewise with greatly specialized and reduced flowers and inflorescences. The recognition of the woody Ranales as the most primitive living angiosperms further emphasizes entomophily as primitive. Pollination in Eupomatia — in flower structure, one of the most primitive angio- sperms — is by beetles, the ancient and primitive group of insects. The flower structure of this genus is such that pollination by wind is impos- sible; pollen can reach the carpels only by beetles eating through an enclosing sheath of staminodia (Fig. 142). Pollination is also only, or largely, by beetles in Cahjcanthiis, Magnolia, lUicium, Paeonia, and in the herbaceous Nymphaeaceae, especially Victoria, now generally recog- nized as primitive taxa. Little is known of methods of pollination in other genera of the Magnoliaceae and in the primitive families, Winter- aceae, Degeneriaceae, Himantandraceae, Trochodendraceae, and Tetra- centraceae. There are strong similarities in features of floral structure related to pollination in Eiipomatia and Cahjcanthiis. In both genera, the stigmas are enclosed in a chamber readily accessible only to chewing insects; food, also available only to chewing insects, is abundant. This food con- sists of "food bodies," richly protoplasmic, succulent tissue. These food bodies are pads on the surface of the inner staminodia in Eupomatia laurina (Fig. 69F); apical clusters of cells on inner petals, stamens, and staminodia in Cahjcanthus (Fig. 69A to D) (a small amount of nectar has been reported in one species) and E. hennettii; in lUicinm, they are "succulent knobs in the center of the flower." There are no nec- taries as such in these taxa. In Eupomatia laurina, the beetles, trapped inside the floral chamber, rarely escape until the outer floral organs are shed; evidence of their long feeding and imprisonment is the presence of debris and excrement. Pollination by beetles is crude but effective. In higher families, it is replaced by true nectaries, surface areas or organs that supply fluid food over a considerable period. Beetles are also associated in varying percentages with other pol- linizing insects, especially in families with small flowers in dense in- florescences, such as the Cornaceae, Caprifoliaceae, Fagaceae {Cas- tanea), Saxifragaceae ( Hydrangeae ) . Fig. 69. Food bodies of beetle-pollinated angiosperms. A, B, C, D, terminal food bodies on inner floral appendages of Calijcanthus occidentalis: A, on stigma; B, on staminode; C, on stamen; D, on inner petal. E, part of longitudinal section of food body on inner staminodium of Calijcanthus florida showing thin-walled tissue of food body and, below, cells of connective; F, longitudinal section of staminodes of Eupomatia laurina, enclosing pollination chamber x, the inner two with ventral food bodies; G, inner staminode of Eupomatia bcnncttii showing globose food bodies. {Ato D, after Grant; E, after Daunian; G, after Schnizlein.) 171 172 MORPHOLOGY OF THE ANGIOSPERMS The majority of angiosperms are pollinized by other groups of insects — ^bees, flies, moths, and butterflies. Nectaries of many forms and varied morphology secrete attractive fluids in positions correlated with the habits and structure of the insects and with the locations and dehiscence types of the stamens. With extrorse anthers, nectaries are usually at the base of the flower outside the stamens; with introrse dehiscence, inside the stamens. (Nectary position is further discussed at the end of this chapter. ) In many tropical and some temperate taxa, birds are the chief pol- linizers. It is believed by some students of ornithophily that the extent of this method of pollination is underestimated. Birds have been re- ported as visiting the flowers of more than four hundred genera in many families scattered throughout angiosperms. Visitation of the flowers of a species by birds seeking nectar does not, of course, necessarily mean that pollination occurs in this way, but the evidence to be found in many of these taxa seems unquestioned. Among families in which bird pollination is prominent are the Proteaceae, Gesneriaceae, Nyctagina- ceae, Passifloraceae, Epacridaceae (part), Marcgraviaceae, Violaceae (tropical genera), Leguminosae (woody, tropical genera), Myrtaceae, Bignoniaceae. Some well-known genera reported as pollinated in part by birds are Loronthus, FAicahjptus, Vanilla, Batihinia, Carina, Gladiolus, Fuchsia, Acacia, Xanthorrhoea, Cattleija, Pritcliardia, Cereiis, Anigozan- thus. Aloe, Lonicera, Agave, Miisa, Jacaranda, Callistemon. Some unusual features of flower structure have been considered adaptations to bird pollination: the stipitate, compound ovary (not the stipitate follicle, which is primitive) — the ovary borne above the basal nectaries and away from possible injury by the beaks of birds, as in the Proteaceae and Capparidaceae; the ovary protected by sheaths of fused stamens, as in primitive tropical Mai vales. Pollination by bats is believed to occur in a few tropical genera, such as Kigelia, Diirio, Freycinctia, Erythrina, Barringtonia. Snails are be- lieved to pollinate some aroids and aquatic plants. Pollination by water is obviously an acquired method. It may be by contact of flowers on the surface of the water where the staminate flowers break free and float to the pistillate, as in Elodea and Vallisneria; or by pollen floating submersed, as in Zostera, Halophila, Cymodocea, Najas, and some species of Pofamogefon. (Other aquatic plants have aerial inflorescences, with pollination by wind or insects — many species of Potamogeton.) Perhaps the most remarkable method of pollination is that of Cemtophi/lhim, where distribution of pollen is by both air and water. The stamens, freed from submersed floweis, rise to the surface of the water and there discharge the pollen, which sinks slowly through the water to the pistillate flowers. POLLINATION 173 Anemophily characterizes the Betulaceae, Fagaceae (part), Juglanda- ceae, Urticaceae, Casuarinaceae, Gramineae, Cyperaceae, Ulmaceae, Restionaceae, Plantaginaceae, Saururaceae, Juncaceae (part), Amaran- thaceae (part), isolated genera in other famihes — Artemisia, Ambrosia, and Xanthium in the Compositae, Fraxinus in the Oleaceae, Thalictntm and Cimicifuga in the Ranunculaceae, Sanguisorha and Acaena in the Rosaceae, Mercurialis in the Euphorbiaceae, Rtimex in the Polygonaceae. From this hst, it is clear that anemophily is distributed throughout the angiosperm system and in taxa that are highly specialized, not primitive. In these taxa, the flowers are mostly numerous, small, inconspicuous, and odorless, with perianth absent or greatly reduced, not brightly col- ored. Other features of flower and inflorescence structure that occur frequently in anemophilous taxa are flexuous staminate inflorescences — many Amentiferae; greatly elongate and flexuous pedicels of staminate flowers — species of Acer; elongate filaments with versatile anthers — grasses and sedges, many Restionaceae, TJialictriim, Plantago. The genus Acer shows, in its staminate flowers, a series in loss of perianth, nec- taries, and odor, and in elongation of pedicles accompanying the change from entomophily to anemophily. Species like A. riibrum and A. sac- charinum with subsessile flowers, corolla, and odor are entomophilous; A. saccharitm, and A. Ncgundo with long pedicels and naked, odorless flowers are anemophilous; other species show transitional stages. Explo- sive deliiscence, especially the simultaneous dehiscence in many flowers, as in some Urticaceae, is accessory to wind pollination. Some of the Amentiferae, a chiefly anemophilous group, are apparently in transi- tion from entomophily to anemophily. In the Salicaceae, Salix has showy stamens, well-marked vestigial perianth parts, serving as nectar-secreting glands, and pollination is partly by wind and partly by bees; Popnltis is strictly wind-pollinated, without nectar secretion, and has inconspicuous stamens. In the dominantly wind-pollinated Fagaceae, 60 per cent of the insect visitors of Castanea (C. sativa) are beetles. In entomophilous plants, transfer of pollen by beetles is apparently primitive; that by bees, butterflies, moths, and flies, is later acquired. Insect pollination may be secondarily acquired from wind pollination, as is apparently the condition in Ficiis, Euphorbia, Hevea, Ricinus, the Nyctaginaceae, and in some of the Araceae. It is phylogenetically signif- icant that beetles, primitive insects, are associated with the primitive Eupomatiaceae and Calycanthaceae to the exclusion of other insects, and with the also primitive Magnolia, lUiciiim, Poeonia, and the Nym- phaeaceae, where they seem to play the major part in pollination. The association strongly supports the now generally accepted view that the woody Ranales are the most primitive living angiosperms. It is evidence that the angiosperms were well established and probably diverse before 174 MORPHOLOGY OF THE ANGIOSPERMS the higher insects arose. Angiosperms, with well-defined nectaries and stigmas, and higher groups of insects, with hairy bodies, perhaps de- veloped together in Jurassic and Cretaceous times. The closure of the stigma canal and of the carpel itself have also been considered an accompaniment of pollination by the higher insects. Entomophilous flowers that are visited by bees, moths, and butterflies, have conspicuous perianth, fragrance, and nectar glands. Flies visit chiefly flowers with disagreeable odor and reduced or dully-colored perianth. Flowers pollinated by birds are usually large or closely grouped in large, conspicuous inflorescences, and secrete quantities of nectar. The large, bird-pollinated inflorescences of the Australian waratah, Telopea speciosissima, and some species of GrevUlea drip with glistening nectar, even under dry atmospheric conditions. The highly colored organs in Telopea are the bracts and calyces; in GrevUlea, the calyces. The nectar is secreted by glands that represent vestigial petals in both of these genera. The pollen grains of wind-pollinated plants are small, light, smooth, and usually nonadhesive, not cohering in clusters; those of entomophi- lous and ornithophilous plants are strongly adhesive, borne in pollinia, tetrads, or free but cohering in masses. The pollen grains of genera pollinated under water are elongate — filamentous, or confervoid and not adhesive. Vestiges of adhesive material have been found on the wind-borne pollen of taxa with close entomophilous relatives: Acer ( anemophilous species), Riimex, Ambrosia, Sanguisorba. Pollen type, method of pollen distribution, and structure of the gynoecium are, biologically, closely correlated. THE MALE GAMETOPHYTE The newly formed microspore has a large, centrally placed nucleus, dense cytoplasm, and a delicate wall. Before germination, it increases greatly in size and volume (Figs. 70B and 71F). A central vacuole de- velops, and the nucleus takes a peripheral position. UsuaUy germina- tion occurs only some days or weeks after the spore is mature, though, in tropical plants, it may be immediate. In temperate-climate plants that flower in the spring, the sporangenous tissue may overwinter at the spore-mother-cell stage, the mature microspore stage, or at the two- cell stage of the gametophyte; perhaps more taxa overwinter at the spore-mother-cell stage than at later stages. Germination of the spore begins with cell division and the formation of a small, walled cell, which lies against the spore wall (Figs. 70D and 71G). This first cell of the gametophyte is rarely formed centrally in the spore by "free cell forma- tion," development without typical wall formation; the phragmoplast forms a delicate cytoplasmic "wall," which delimits the protoplast — THE MALE GAMETOPHYTE 175 Scirpus (Fig. 60). The formation of this first cell as a wall-less, free protoplast probabl)' marks a forward step in the reduction of the gametophyte. Though definite positions in the spore have been claimed for the first walled cell — and there may be a constant position in a given species — several or many diflFerent positions have been reported for angiosperms as a whole. The first-formed cell, walled and peripheral, Fig. 70. Diagram showing development of male gametophyte in angiosperms. A, microspore; B, microspore, enlarged, vacuolate; C, spore nucleus dividing; D, ga- m.etophyte, two-celled, smaller generative cell, larger vegetative (tube) cell; E, freeing of generative cell from spore wall; F, generative cell embedded in vegetative cell; G, H, di\'ision of generative cell to form two male gametes within the spore wall; I, J, division of generative cell within the pollen tube. (After Maheshwari.) is the generative cell (Figs. TOD and 71G); the larger, naked cell, cen- tral and wall-less, filling the remainder of the spore-wall cavity, is the vegetative or tube cell. The nuclei of these two cells differ not only in size but in structure and staining qualities. The vegetative nucleus has a prominent nucleolus; the generative nucleus, a small nucleolus or none. The generative cell is soon freed from the spore wall and be- comes ellipsoid or fusiform. Prominent cytoplasmic differences set it 176 MORPHOLOGY OF THE ANGIOSPERMS sharply off from the cytoplasm of the vegetative cell in which it is embedded. The formation of the generative cell takes place during late stages in anther development. In a sporangium, the divisions in the development Fig. 71. Microsporogenesis and development of the male gametophyte in Porttilaca oleracea. A, row of spore mother cells in anther sac; B, C, meiotic divisions in spore mother cell; D, tetrad of spores nearly mature; E, detail of spore-wall formation in tetrad; F, mature spore, enlarged before germination; G, H, generative and tube cells within spore wall; /, ], mature pollen grain with generative cell and two sperms; K, germinating pollen grain; L, part of pollen tube showing tube nucleus and sperms. {After D. C. Cooper.) of the gametophyte are not as nearly simultaneous as are the meiotic divisions, but are not far apart, except in long anthers, such as those of Liriodendron, where there is acropetal succession in germination of the spores. In tetrads that remain permanently together, and in pollinia, the divisions may be simultaneous. THE MALE GAMETOPHYTE 177 The Pollen Tube. Germination of the pollen grain on the stigma occurs after a period of time that varies greatly: from almost immediately — Sacchariim, Sorghum; five to ten minutes — Taraxacum, Zea; two hours; three hours; two days. The grain swells, probably by absorption of the secretions of the stigmatic surfaces; the shape may be greatly modified; furrows are filled out and flat surfaces rounded. The inner wall pro- trudes at one or more of the thin areas in the outer wall — in the fur- rows or apertures. In triangular pollen grains, tlie apertures are at the corners of the grain, and the tube develops from one of these corners — Proteaceae. By rapid apical growth, one of the protrusions becomes the pollen tube. Rate of elongation of the tube varies greatly — with the taxon, and with conditions of temperature and humidity. In early stages of tube development, the contents of the pollen grain, or most of them, pass into the tube, where further development of the cells may occur. The tube and its contents — together with any part of the contents that may remain within the pollen-grain wall — constitute the male gametophyte. The "pollen tube" — the tube plus its contents — is commonly called the male gametophyte; the contents alone are some- times described as the gametophyte. Objection has been raised to the inclusion of the tube itself as a part of the gametophyte, but the tube is a development of the microspore wall, and the spore is the first cell of the gametophyte. Both the pollen tube and the embryo-sac wall are en- larged and modified spore walls (walls of the spore mother cell or dyad in tetrasporic and disporic sacs). The contents of the mature pollen grain and of the young pollen tube, commonly called the male gametophyte, represent the immature gametophyte. (The generative nucleus may not yet have formed the male nuclei; the cytoplasm of the tube continues to increase in volume as the tube develops.) The male gametophyte of the angiosperms remains enclosed within the greatly modified spore wall until the male cells are formed. In this en- closure of the male gametophyte within the spore wall, the angiosperms resemble other major heterosporous taxa. The pollen tube itself, as a male-cell carrier, is characteristic of the highest gymnosperms and angiosperms only; by its aid, the male gametes reach the egg. Accompanying this high specialization of the spore wall has been loss of motility in the gametes. Only in the highest seed plants is fertilization accomplished by nonswimming male cells. The pollen tube represents one of the most difficult adaptations to life on land, the step from fertilization in a fluid by free-swimming sperms — the ancient method— to fertilization by nonswimming sperm cells, carried to the egg by the gametophyte. No animals, and only the highest plants, have made this advance. The pollen tube is probably an example of parallel development. The 178 MORPHOLOGY OF THE ANGIOSPERMS coniferophyte gymnosperms — Coniferales and Ephedrales — are so un- like angiosperms as to be unquestionably a stock unrelated in origin to the angiosperms, and their pollen tubes are probably of independent origin. The cycads and Ginkgo have tubes functionally different from those of conifers and angiosperms. In these taxa, the tubes, short and haustorial in nature, develop on one side of the pollen grain. The pollen grains become enlarged with increased cytoplasm and break open, dis- charging the sperms and cytoplasm into a pollen chamber, where the sperms swim in the cytoplasmic fluid to the archegonial necks. In the angiosperms, nonswimming sperms are discharged into the embryo sac. Germination of pollen grains at a distance from the ovule has been considered an angiosperm character, but, in Tsiiga and Araucaria, the grains germinate far from the ovule and form long tubes extending to the micropyle; the pollen grains of Agathis form extensive pollen tubes in the cone axis and ovular tissues before reaching the archegonium. Stages in the phylogenetic development of germination at a distance from the ovules are present in living Ranales. In Dcgeneria and some species of Drhmjs, the pollen grains germinate on a long, ventral stig- niatic crest, and their short tubes penetrate between the unsealed "lips" of the carpel directly to ovules close by (Fig. 83). In other species of Drimijs and other genera of the Winteraceae, progressive acropetal shortening of the crest — the pollen tubes still short — restricts ovule position to the area directly below the crest. With the acquisition of a terminal stigma and a style, the pollen tubes become long, and ovule position is not restricted. The pollen tube, as it increases in length, usually pushes between the cells of the stigmatic surface and the inner tissues of the stigma and style. In some families, it destroys the adjacent cells, eroding a path to the ovules. Its course may be along the surface of the stylar canal or deep within the stylar tissues. It may pass through vascular tissues (protoxylem) in its course — Casiiarino. (It resembles in this the course of pollen tubes in the phloem of araucarian conifers. ) Where a stylar canal is present, the tube may pass wholly or partly along the wall of the canal. Entering the ovarian chamber, the tube may continue along the wall to the ovule or cross directly to the micropyle. It follows lines of transmitting tissue (often wrongly called conducting tissue), the extent and distribution of which within the stylar canal and the ovary have not commonly been recognized. The course of the tube, though obviously toward the ovules, has been considered not directed structurally. But, in many taxa, the tube follows direct lines through the ovary, predetermined by the distribution of transmitting tissue (Fig. 77), which is by no means restricted to the style, as often implied. In the more specialized taxa, the tube seems to take the shortest course THE MALE GAMETOPHYTE 179 along the surface of the ovary wall and placenta to the ovule. To reach the micropyle, the tube may pass along the surface of the ovule to the micropyle, but it may enter the ovule and follow the vascular bundle, running through protoxylem cells in funicle and raphe — Casuarina. The crossing of a space, as a "short cut" to the micropyle — described in some taxa — has been called an "extraordinary accomplishment," be- cause it involves the continuing growth of the tube tip, without contact of the tip with tissue of the ovary. Where no ovule, as such, is present — some parasites, such as Viscum — the tubes pass through carpellary tissues. Branching of the tube is rare except in chalazogamous taxa, where it may take place at various levels in the ovule. Time of growth of tube from stigma to embryo sac varies greatly and is not controlled by distance from stigma to embryo sac; the tip may grow rapidly or slowly, and many factors enter into its elongation. The tube may grow from pollen grain to the egg over some millimeters in a few hours. But many days, even weeks and months, may pass between pollination and fertilization — in the witch hazel, about six months; in the black oaks, about twelve months; but these periods include winter dormancy. Entrance of the pollen-tube tip directly through tissues of the ovule, rather than by the micropyle, was first found in Casuarina in 1891 and termed chalazogamij ( Fig. 103 ) , because the tube penetrates the chalaza on the way to the embryo sac. Chalazogamy was soon found in several other amentiferous genera, and, since the Amentiferae were then con- sidered the primitive angiosperms, chalazogamy was considered a primi- tive character. (A classification of angiosperms as Chalazogams and Porogams was supported by a few botanists. ) But chalazogamy is now known to be present in taxa in other parts of the angiosperm system — Anacardiaceae, Cucurbitaceae, Rosaceae, Chloranthaceae — and must be considered a specialized, rather than a primitive, character. Penetration of the tube through the integuments is sometimes termed mesogamtj, and the term hasigamij was suggested to replace chalazogamy. "True chalazogamy" is said to occur only when the tube enters the embryo sac at its base. Entrance of the tube by the micropyle has been called porogamij; all other methods, aporogamij. In a few taxa, there is varia- tion in place of entry of the tube — Epilobium, Brassica. Entrance of the tube into the embryo sac at any point other than the micropylar end is probably rare; tubes entering the ovule below the micropyle usually pass along the embryo-sac wall until they reach the position of the egg. Pollen tubes may persist in the nucellus long after fertilization, a con- dition frequent in the Proteaceae. The parasitic behavior of the pollen tube is reported to reach an extreme in Casuarina, where it is described as destroying germinating 180 MORPHOLOGY OF THE ANCIOSPERMS megaspores; in this, it resembles the tubes of the araucarian conifers, which destroy the contents of immature archegonia in their course. The Tube Nucleus. The nucleus of the vegetative cell, the tube nucleus, commonly shows signs of degeneration as the generative cell matures, and may change in shape. It has been generally stated that, as the de- velopment of the pollen tube begins by extrusion of the intine of the pollen-grain wall, the tube nucleus passes into the tube and, as the tube grows, maintains a position in tlie distal part, where it "governs apical growth." But the supposed function of this nucleus has been questioned. The nucleus may remain within the spore wall; it is often degenerate before the beginning of tube growth; it may not enter the tube until the tube is well developed; it often follows, rather than pre- cedes, the generative cell or the gametes (the relative position of the two or three nuclei is perhaps a matter of chance); it may not progress far, and the male cells may pass it. The tube may branch, even re- peatedly — Fagiis and other Amentiferae — and the branches which have no nuclei develop as well as that with the nucleus. The vegetative cell has also been called a vestigial cell, the last survivor of the vegetative tissue of an ancestral, free-living gametophvte. As such, it would be homologous with the one or two prothallial cells of the cycads, Ginkgo, and the conifers. Male Gametes. Division of the generative cell, forming two male gametes or sperms, may follow soon after it is mature or may be de- layed until the cell has passed into the pollen tube. Within the tube, division may occur close to the base or far along, even well down in the style — LiJitim. In some species, it apparently may occur in either place. Division of the generative cell in the ungerminated pollen grain as it lies on the stigma also occurs. The pollen grain is either 2- or 3-nucleate — 3-nucleate if the male gametes have been formed before shedding. It is uncertain whether the two-celled or the three-celled pollen grain is more common. The male gametes are definite cells, with limiting cyto- plasmic sheaths, not naked nuclei, as often formerly believed. In shape, they are spherical, ellipsoid, lenticular, or vermiform. Their shape may change as they pass through the pollen tube. The two gametes are usu- ally alike in size. DiflFerences in size and staining quality — the gamete that unites with the egg, often smaller and staining less deeply — have been reported in a few species. The question of possible motility of the male cells has received con- siderable attention. In the lower gymnosperms, the sperms are ciliate and actively motile; in the higher gymnosperms, the pollen tube is a "male-cell carrier," as in the angiosperms, and independent movement seem„s to have been lost. It has been claimed that the sperms of angio- sperms have some capacity for "active movement" independent of THE MALE GAMETOPHYTE 181 cytoplasmic streaming in the tube. Part of the evidence for motility of the sperms is the frequent fusiform or vermiform shape of the gametes — shapes that suggest the sperms of lower vascular taxa. The generative cell also, when in the tube, has been described as moving "actively and independently." Elongate and vermiform shape of the male-cell nuclei is not, in itself, evidence of motility; nuclei of this type are fre- quent in elongate cells of many tissues. The older view, generally maintained, was that the gametes are carried by cytoplasmic stream- ing in the pollen tube. Movement of the gametes along the pollen tube where two-way movement in the cytoplasm occurs has been considered further evidence of independent movement of the gametes and proof that these cells are not carried simply by streaming of the cytoplasm as the tube rapidly elongates. Changes in turgor of the tube have also been considered the major cause of passage of the gametes or genera- tive cell along the tube. Terminology of the Male Gametophyte. Since the later part of the nineteenth century, there have been many changes in the terms applied to the various structures and stages in the development of the male gametophyte. Attempts to homologize the pollen grain with the an- theridium of lower groups have been, in some measure, responsible for the choice of terms. Changes in terminology have also come about by the demonstration, with new techniques, that some stages described as represented by "nuclei" are definite cells, with delimited cytoplasmic sheaths. Tube nucleus and vegetative nucleus — long used for the nucleus of the vegetative cell — are satisfactory and useful terms. They will continue in use, even though, as generally believed, the tube nucleus probably does not control the development of the pollen tube. The vegetative cell is a naked cell, lying within the microspore wall. The pollen tube is an extension of part of the inner layer of the spore wall; it does not constitute "the male gametophyte," even though it may en- close most of the male gametophyte. The generative cell was early called the spermotogcnous cell, a term as good, or better, than genera- tive cell, which is now used in the broader treatments. Both of these terms have been supplanted by anfhcridial cell in many ti^eatments since 1912. The term antheridial cell brings confusion to the descrip- tion of the male gametophyte, unless it is used as a part of a compara- tive study of the antheridium throughout vascular plants. The term generative nuclei, as applied to the nuclei of the male gametes, is un- fortunate, because it suggests nuclei of the generative cell. The nuclei formed at division of the generative cell are the nuclei of the male gametes. They have been called male nuclei, generative nuclei, and spewiatogcnous nuclei, all good, in a descriptive sense, for the nuclei of the gametes themselves. But it has commonly been assumed these 182 MORPHOLOGY OF THE ANGIOSPERMS nuclei are the gametes, that definite male cells are not usually formed. Improved techniques have demonstrated that each nucleus is sur- rounded by a definitely limited sheath of cytoplasm. The male gametes are cells. BIBLIOGRAPHY Pollen and Male Gametophyte Bailey, I. W., and C. G. Nast: The comparative morphology of the Winteraceae. I. Pollen and stamens, Jmir. Arnold Arh., 24: 340-346, 1943. Beer, R.: Development of the pollen grain and anther of some Onagraceae, Beih. Bot. Centralhl, 19 (I): 286-^313, 1905. Canright, J. E.: The comparative morphology and relationships of the Magnoliaceae. II. Significance of the pollen, Phytomorph., 3: 355-365, 1953. Cooper, D. C.: Microsporogenesis and the development of the male gametes in Tartulaca oleracea. Am. Jour. Bot., 22: 453-459, 1935. Cooper, G. O.: Microsporogenesis and development of seed in Lobelia cardinalis, Bot. Gaz., 104: 72-81, 1942. Cranwell, L. M.: New Zealand pollen studies: The monocotyledons. Bull. Auckland Inst, and Mu.s., 3: 1953. Engler, A.: Beitriige zur Kenntnis der Antherenbildung der Metaspermen, Jahr. Wiss. Bot., 10: 275-316, 1876. Erdtman, G.: "An Introduction to Pollen Analysis," Waltham, Mass., 1943. : Pollen morphology and plant taxonomy. V. On the occurrence of tetrads and dyads, Svensk. Bot. Tidskr., 39: 286-297, 1945. Garcide, S.: The developmental morphology of the pollen of the Proteaceae, Jour. So. Afr. Bot., 12: 27-34, 1946. Goebel, K.: "Die Entfaltungsbewegungen der Pflanzen," Jena, 1920. Hanf, M.: Vergleichende und entwicklungsgeschichte Untersuchungen iiber Mor- phologic und Anatomic der Griffel und Grilfelaste, Beih. Bot. Centralhl., 54A: 99-141, 1935. Hannig, E.: See first reference under Stamen and Androecium. Hotchkiss, A. T.: Pollen and pollination in the Eupomatiaceae, Proc. Linn. Soc. N.S.W., 83: 86-91, 1958. Knuth, P., and E. Loev^^: "Handbuch der Bliitenbiologie," vols. I-III, Leipzig, 1904-1905. Maheshwari, P.: The male gametophyte of angiosperms, Bot. Rev., 15: 1-75, 1949. : "Introduction to Embryology of Angiospenns," New York, 1950. Martens, P.: La graine et le tube pollinique: Reflexions sur les caractcres propres des phanerogames. Bull. Acad. Roy. Belg. CI. Sci., 5 ser., 33: 919-943, 1949. Matthews, J. R., and E. M. Knox: See reference under Stamen and Androecium. Murbeck, S.: Uber das Verhalten des Pollenschlauches bei Alchemilla arvensis und das Wegen der Chalazogamie, Lunds Urnv. Arsskr., N.F., (Avd. II) 9: 1-46, 1901. Newman, I. V.: Studies in the Australian acacias. IV. The life history of Acatia haileyana F.V.M. Part 2. Gametophytes, fertilization, seed production and germination, and general conclusion, Proc. Linn. Soc. N.S.W., 59: 277-313, 1934. Parmentier, P.: Recherches sur le pollen des dialypetalees, Jour. Botanique, 15: 150- 166, 194-204, 218-223, 419-429, 1901. Piech, K.: Zur Entwicklung der Pollenkorner bei Scirpus lacu.stris L., Bull. Acad. Polon. Sci. et Lett. Cl. Sci. Math, et Nat., ser. B., 1924: 113-123. BIBLIOGRAPHY 183 : Uber die Entstehung der generativen Zelle bei Scirpus uniglumis Link durch "freie Zellbildung," Planta, 6: 96-117, 1928. Pohl, F.: Der einfaltige Pollen, seine Verbreitnng und phylogenetische Bedeutving, Beih. Bot. Centralbl, 45: 59-75, 1928. : Die Kittstoffreste auf der PoUenoberfliiche windbliitiger Pflanzen, Beih. Bot. Centralbl, 46: 286-305, 1929. Porsch, O.: Geschichtliche Lebenswertung der Kastanienbliite, Oesterr. Bot. Zeitschr., 97: 269-321, 1950. Schnarf, K.: Studien iiber den Bau der Pollenkomer der Angiospermen, Planta, 27: 450-465, 1938. Shoemaker, D. X.: See reference under Stamen and Androecium. Swamv, B. G. L.: See reference ( 1948) under Stamen and Androecium. '-: Further contributions to the morphology of the Degeneriaceae, Jmtr. Arnold Arb., 30: 10-38, 1949. Tammes, P. M. L.: On the origin, number and arrangement of the places of exit on the surface of pollen grains, Rec. Trav. Bot. Neerl, 27: 1-82, 1930. Werth, E.: Ueber die Bestiiubung von Viscum und Loranthus und die Frage der primitivitiit der Windbliitigkeit \1e der Pollenblumen bei den Angiospermen, Ber. Dcutsch. Bot. Ges., 41: 151-164, 1923. W'odehouse, R. P.: The phylogenetic value of pollen-grain characters, Ann. Bot., 42: 891-934, 1928. : "Pollen Grains," New York, 1935. : Evolution of pollen grains, Bot. Rev., 2: 67-84, 1936. W'ulff, H. D.: Beitriige zux Kenntnis des mannlichen Gametoph\-ten der Angio- spermen, Planta, 21: 12-.50, 1933. : Die Entwicklung der Pollenkomer \on Triglochin pahtstris L. und die verschiedenen T\-pen der PoUenkbmentwicklung der Angiospermen, Jahr. Wiss. Bot., 88: 141-168, 1939. and P. Maheshwari: The male gametoph\-te of angiosperms, a critical re- xiew, Joitr. Indian Bot. Soc, 17: 117-140, 1938. Dehiscence Artopoeus, A.: Uber den Bau und die Offnungsgeweise der Antheren und die Ent- wickkmg der Samen bei den Ericaceen, Flora, 92: 309-345, 1903. Chatin, A.: Causes de la dehiscence des antheres, Compt. Rend. Acad. Sci. Paris, 70: 201-203, 410-413, 644-648, 1870. Leclerc du Sablon: Recherches sur la structure et de la dehiscence de antheres, Ann. Sci. Nat. Bot., ser. 1: 97-134, 1885. Matthews, J. R., and C. M. Maclachan: The structure of certain poricidal anthers. Trans, and Proc. Bot. Soc. Edinburgh, 30: 104-122, 1930. Schips, M.: Zur Offnungsmechanik der Antheren, Beih. Bot. Centralbl, 31: 119- 208, 1913. \'an Tieghem, P.: Observations sur la structure et la dehiscence des antheres des Loranthees, sui\-ies de remarques sur la structure et la dehiscence des antheres en general. Bull Soc. Bot. France, 42: 36.-3-^368, 1895. Woysicki, Z.: Recherches sur la dehiscence des antheres et le role du stomium. Rev. Gen. Bot., 36: 196-212, 250-268, 1924. POLLIXATION Daumann, E.: Das Bliitennccktarium von Magnolia und die Fiitterkoerper in der Bliite von Calijcanthus, Planta, 11: 108-116, 1930. 184 MORPHOLOGY OF THE ANGIOSPERMS Delpino, F.: "Ulteriori Osservasioni sulla Dicogamia nel Regno Vegetali," Milan, 1868-1875. Diels, L.: Kiiferblumen bei den Ranales und ihre Bedeutung fiir die Phylogenie der Angiospermen, Ber. Dciitsch. Bot. Ges., 34: 758-774, 1916. Grant, V.: Pollination systems as isolating mechanisms in angiosperms, Evol., 3: 82-97, 1949. : The protection of the ovules in flowering plants, Evol., 4: 179-201, 1950. ( Bird pollination. ) -: The pollination of Calijcanthus occidentalis, Am. Jour. Bot., 37: 294-297, 1950. Hotchkiss, A. T.: See reference under Pollen and the Male Gametophyte. Pijl, L. van der: Fledermiiuse und Blumcn, Flora, 131: 1-40, 1936. Pohl, F.: Beziehungen zwischen Pollen l^escliaffcnheit, Bestiiubungsart und Fruch- knotenbau, Beih. Bot. Ccntralhl, 46: 247-285, 1929. : See second reference under Pollen and the Male Gametophyte. Porsh, O.: Kritische Qucllenstudien iibcr Blumcnbesuch durch Vogel, Biol. Gen., 2: 217-240, 1926; 3: 171-206, 475-548, 1927; 5: 157-210, 1929; 6: 133-146, 1930. : Vogelblumenstudien, ]ahr. Wiss. Bat., 63: 553-706, 1924; 70: 181-277, 1929. : Die Vogel als Blumenstauber, Biol. Gen., 9: 239-252, 1933. : Siiugertiere als Blumenausbeuter und die Frage der Siiugertierblume, Biol. Gen., (I) 10: 657-685, 1934; (II) 11: 171-188, 1935. -: Windpollen und Blumeninsckt, Oestcrr. Bot. Zeitschr., 103: 1-18, 1950. Sahni, B.: Pollen grains in the stylar canal and in the ovary of an angiosperm, Ctirr. Sci., 4: 587-589, 1936. Schnizlein, A.: "Iconographia," vol. Ill, Bonn, 1843-1870. Stebbins, G. L., Jr.: "Variation and Evolution in Plants," New York, 1950. Werth, E.: See reference under Pollen and the Male Gametophyte. Chapter 6 THE GYNOECIUM The ovule-bearing organs of the flower, the carpels or megasporo- phylls, make up the gynoecium. They range in number from many to one and, in arrangement, from spiral to whorled. The carpels may be free from fusion with one another — the gynoecium apocarpous — or connate in various degrees — the gynoecium sijncarpous. In the early decades of die twentieth century, the term coenocarpous has been ap- plied by some authors to gynoecia with connate carpels that ai-e several- chambered, with axillary placentae; and the term paracarpous to syn- carpous gynoecia with a single chamber and parietal, basal, or free central placentation. This distribution has morphological value, but paracarpy has perhaps arisen in more than one way: by phylogenetic union of open carpels, and by modification of coenocarpic gynoecia. A line between apocarpy and syncarpy is difficult to draw, because fusion may be slight and even developed late in ontogeny. Syncarpy is dis- cussed later in this chapter. The gynoecium perhaps shows more simply than the androecium and the perianth the major changes in the evolution of the flower. Especially prominent are the advances from spiral to whorled arrange- ment, from free to fused members, and from many to one which is pseudoterminal. Numerous spirally arranged carpels characterize rather few families — Magnoliaceae, Annonaceae, Eupomatiaceae, many of the Ranuncula- ceae and Rosaceae; usually apocarpous, they may become syncarpous in fruit — Annonaceae. Gynoecia with few, spirally arranged carpels are few, but, in syncarpous forms, the spiral arrangement may be de- terminable only anatomically — Berheris. The flower of Scheuchzeria has been cited as an example of the spiral placing of a few carpels, but this "flower" is perhaps an inflorescence. Spiral arrangement is difficult to determine in gynoecia with only two carpels, but, in these gynoecia, one carpel overtops the other — Jeffersonia, Epimeditim. Oblique adna- tion to the receptacle by individual carpels brings about "false coeno- carpy," a type of union difficult to determine morphologically, because distinction must be made between the tissues of the receptacle and those of the carpels. False coenocarpy is discussed further under Syncarpy. Reduction of fertile carpels to few and to one is common. 185 186 MORPHOLOGY OF THE ANGIOSPERMS Whorled carpels represent modification of spiral arrangement. This modification is often readily seen in internal structure where there is little superficial evidence. The Terms Pistil and Locule. The term pistil is unfortunately loosely used and confusing to students in its various morphological implica- tions. It has been defined as the "unit of the gynoecium," but this "unit" may be a single carpel — in apocarpous gynoecia — or a group of carpels — in syncarpous gynoecia. The use of modifying adjectives — simple where the gynoecium consists of a single carpel, and compound where it consists of more than one carpel — are helpful, but "compound" is sometimes used for both apocarpous and syncarpous gynoecia. "Pistil," as a term applied to gynoecia, should be avoided as far as possible. The term locule is also loosely used, morphologically. It is applied to the space enclosed within a single carpel, and to that within connate carpels, both the common chamber and that within a ring of closed carpels — Sempervimim, Phytolacca. Discussions of syncarpy and the inferior ovary conclude the discussion of the carpel. THE CARPEL The angiosperm megasporophyll, the carpel, is, like the stamen, an elongate appendage, primitively of laminar form. It is like the stamen also in that it is leaflike in its relation to the stem — position, arrange- ment, vascular connections (traces), anatomy, and ontogeny (apical and marginal meristems). It differs from the microsporophyll in that the megasporangia are always on the adaxial side, whereas the micro- sporangia may be on either the adaxial or the abaxial side. In the carpel, the sides of the lamina have been folded or rolled adaxially toward the midrib. The folding or rolling encloses the megasporangia (ovules) in a chamber, the locule. Form of the Carpel In its typical form, the carpel resembles a folded or rolled leaf lamina, with margins usually appressed or fused. Ancestrally, it was undoubtedly a stalked, dorsiventral organ. The folded organ is still in- completely closed at pollination time in some taxa. Genera with the stipe, a stalk, are frequent in both monocotyledons and dicotyledons — Helobiales (Zannichellia, Scheiichzeria, Ruppia, Althenia); primitive Liliaceae {Tofieldla, Veratrum); many woody Ranales {Dcgeneria, Drimijs, Calycanthus, Bnbbia, some Annonaceae, Austrobaileya); many Ranunculaceae {Coptis, Eranthis, Cimicifuga, Hellehorus, even some achene-bearing genera, such as Thalictrum); Cercidiphyllum; Euptelea; Faeonia; many Dilleniaceae; primitive Rosaceae (Physocarpiis) . The THE CARPEL 187 stipe is sometimes confused with the gynophore, an elongate part of the receptacle that bears the entire gynoecium. Anatomy shows that the stipe is a part of the carpel, not of the receptacle; the vascular traces of the carpel arise from the receptacular stele and are independent in the stipe, with the ventral traces inverted (Coptis, Tofieldia). Stigma, style, and ovary are parts of the lamina of the sporophyll, functionally dis- tinct but often merging. The style, in some taxa, represents a sterile part of the ovary; in other taxa, it seems to be formed by a secondary elongation of the distal part of the carpel. The carpel varies from simple, without differentiation into fertile and sterile parts, to complex. The complex form is divided into a proximal, ovule-bearing part, the ovary; a distal, pollen-receptive part, the stigma; and a median sterile section, the style. Usually terminal on the ovary, the st)'le appears to be lateral in some highly specialized taxa — the Labiatae and achene types in the Rosaceae ( Potentilleae, Chrysobalanoideae). In these plants, ontogeny and the course of the midrib bundle show that the lateral position is secondary. The st\'le is terminal in early stages, but growth of the ovule and seed on the dorsal side displaces the style. The style appears nearly basal, seeming, in later stages, almost independent of the carpel. But anatomy shows that the dorsal bundle of the carpel is continuous from the base of the carpel over the ovule — along the dorsal side and top of the enlarged ovule-bearing base, down to the base of the st^-le and upward through the style. Obviously, the two structures are parts of one organ. (Under the carpel polymorphism theory, the two parts of the carpel were in- terpreted as two independent organs — one solid, the style, and one valvular, the ovary.) The carpels of some of the woody Ranales illustrate the simple type (Fig. 72). They are elongate, usually without distinction of ovary, style, and stigma. The many ovules are borne throughout an elongate cham- ber. The pollen in many is received on a longitudinal stigmatic crest (Fig. 72C, F, H, I), consisting of the papillose surfaces of the margins and borders of the lamina, which have come together in the closing of the carpel. The double nature of the crest is evident in its more or less strongly two-lobed form, the margins flaring back somewhat from the line of contact (Fig. 72F, H). In some species of Drimys, in Hi7nan- tandra and Degeneria (Fig. 83A, B), the crest extends from the apex to the base of carpel (taxonomic descriptions read "stigma decurrent"); in other taxa, it is restricted to the distal part (Fig. 72D, E, H, I). In these primitive taxa, the marginal areas of the lamina are appressed but not fused histologically; closure of the carpel is by interlocking of projecting papillae. The pollen germinates on the stigmatic crest over its full length, and pollen tubes may penetrate directly to the nearest ovules. Elongate stigmas may represent primitive form, as in many 188 MORPHOLOGY OF THE ANGIOSPERMS Ranales, or advanced form, an adaptation to wind pollination. In Euponmtla (Fig. 72A), connation of the carpels restricts the exposed stigmatic area to the apex of the carpel, and the papillose surface ex- tends downward inside the carpel to and around the ovules, forming an Fig. 72. Some primitive carpel types. A, Eupomatia laurina, part of longitudinal section of gynoecium showing connate carpels (spirally arranged) and stigma re- duced to a few papillae on gynoecial "floor"; B, Cimicifuga racemosa, showing early stage in development of style and no definite stigma; C, Drimys membranea, show- ing decurrent stigma, stigmatic crest; D, Caltha palustris, showing slightly decurrent stigma, no style; E, Hydrastis canadensis: El, showing weak style and two-lobed stigma with flaring borders, E2, face view of stigma; F, Dcgcneria viticnsis, show- ing stigmatic crest; G, Chamaewps humilis, open carpel with style; //, Biitomus umbellatus: HI, carpel with style and stylar canal, H2, detail of stylar canal; 7, Coptis trifolia, showing stigma decurrent on style and a long stipe; /, Physocarpiis opulifolius, showing elongate style and terminal stigma. ( F, ajter Bailey and Smith; G, after JuJinke and Winkler.) "internal stigma." The restriction here of pollen-tube transmitting tissue to the inside of the carpel, except at the apex, resembles the phylo- genetic shortening of the stigmatic crest by the closure of the carpel as the typical stigma developed. The basic form of the carpel is that known, when mature, as the follicle. The follicle is the classical carpel, the form long accepted as THE CARPEL 189 the primitive type, the type from which have been derived various speciahzed types, such as the achene. (The terms folhcle and achene are usually applied to mature carpels, fruits.) The follicle is primitive in its elongate form, numerous ovules, simple dehiscence, and the fre- quent absence of a style; it is advanced in the delimitation of ovary and stigma. It is characteristic of many genera of the more primitive families of both dicotyledons and monocotyledons. Carpels of the fol- lowing genera have many folliclelike characters: Dcgeneria, Drimijs (sect. Tasmannia), Butomiis, Trolliiis, Calfha, Coptis, Phijsocarpus, Akebia, Telopea, Paeonia, Toficldia, Schcuchzeria. Closure of the Carpel One of the distinguishing characters of the angiosperms is the en- closure of the ovules by the megasporophyll; the pollen grains do not reach the ovule but germinate at a distance, in contrast with the condi- tion in the gymnosperms, where the ovule is "naked" and the pollen is received and germinates on the ovule. There are exceptions in tlie conifers to the place of germination of the ovule; in Araucaria, Agathis, and Tsuga, the pollen germinates at a distance from the ovule. In the angiosperms, there are taxa in which the carpel is not completely closed at pollination time — Degeneria, Drimijs, Butomaceae, Hydrocharitaceae, Sparganium, Sassafras, Coptis, Tiarelh, Platajius. The "opening" is very narrow, and pollen grains probably never enter the ovarian locule through it. (Pollen has been reported in the stylar canal in Butomopsis, Hillebrandia, Reseda, but probably does not germinate there.) The opening is usually quickly closed but may still be present in the mature fruit. "Union" of the carpel borders shows all stages from the presence of an open slit to loosely and tightly interlocking papillae and to histologi- cal fusion — a union so complete that no evidence remains in tissue structure (Fig. 74A, B). In the open carpels of Degeneria and Drimijs, the opening is partially closed by the papillose cells of the stigmatic crest (Fig. 73). In carpels where the uniting margins or borders are closely appressed but without histological union, the epidermal layers may be merely coherent, without elaboration of surface-cell form, but the epi- dermal layers are often rugose or toothed and interlock as they mature; evidence of the identity of the individual layers may be only in cell form and position. Ontogenetic adjustments among the uniting layers often obscure the line of fusion. All these conditions are found in carpels in which union is ontogenetic; where fusion is congenital, there is no histological evidence of the union. Morphological nature of the carpel may be evident in the maturing and mature fruit, as in the peach and cherry. faj^-i 2^ c ss ^ -",■"00 CO ~ ■^ -^ 3 ^•Q .a. Qj ■^'^ a c« ^ O §^2 ss: 5 1 § « g^|e>^ &-0 ^ s r ^ir--"^ i J3 5 ..^2 oi ^ ^ -c c/:i rom c umbel rophtjl ercidi) , after ^ 0, ^ObJ oT 3 S " :.:: bc = ~= a. 3. i3 5 „ .. 3 3 i~ v- Co ^"^§"2^ floweri A, B, , Sagi ScJiisa after S >— ( t^l^^OQ QJ ^J f-«i • f> • *\ C ts S 3 =3 S a;S S 2 ,2 P,<\ £ &-• ■-^H 'T^ -■-< a «i o -^^ S-. OJ 'rt r ^~ 3 ^ 3ty in degr G to I, ba vitiensis; N, Tofiela natum. (C •C ■ -^ -2 3 S 1^3 rt^ ?*»•—'+-» > C Si^ 3 > c r^ 3 '^ O o v> ^ .^ J". . " ;:d c "S 3 y C >« S ^^„|t^ tt, O 3 .V i .-5^^ o c o Hi, ■„ s secti ntatio E, Tr His tri ptelea VI 'u -i. a 73. Cro inar plac aloides; s; L, Co us; S, E\ .^ e s s ^ fa^ ^-2i§~ 190 THE CARPEL 191 Conduplicate and Involute Closure. The closing of the carpel comes about, structurally, by an upturning of the sides of the lamina and the bringing together adaxially of the surfaces or edges, with more or less complete fusion. (Recent description of abaxial folding in Cercidiphijl- lum was based on the error of interpretation of the inflorescence as a flower.) The upturning may be a simple folding on the midrib region as an axis, with the margins lying side by side (Fig. 74A), or it may be rr\ &^^ Fig. 74. Semidiagrammatic sketches of cross sections of carpels to show two methods of closure of carpels. A, conduplicate method, borders of the lamina meeting face to face by ventral surface, vascular bundles half-inverted; B, involute method, margin of lamina meeting margin (marginal initials meet), vascular bundles in- verted; C, involute method, borclers of the lamina meeting by their dorsal surfaces, margins free. a more or less extensive incurving or inrolling, with contact face to face by the edges (Fig. 74B) or, where the inrolling is strong, by parts of the abaxial surface, with the margins free within the locule (Fig. 74C). The carpel has apparently closed independently in many lines and in a variety of ways. A primitive method of closing is that of a simple folding so that considerable parts of the adaxial surface come together; a carpel, closed in this way, is called conduplicate (Fig. 74A). Conduplicate closing has been considered the basic type from which all others have been derived, but a survey of many families sug- gests that this method is only part of the history of carpel closure. Carpels in which the lamina sides appear rolled upward and inward 192 MORPHOLOGY OF THE ANGIOSPERMS have been termed involute. The rolling appears commonly to have brought the edges of the lamina together (Fig. 74B); where the inroU- ing is greater — an uncommon condition — contact is by strips of the abaxial surface. It seems improbable that all the types of involute carpel have been derived from the conduplicate; this derivation would in- volve a change from contact by the adaxial surface only to contact by the abaxial surface — a major change, far more complicated and cir- cuitous than usually found in evolutionary derivation. The vascular anatomy of the carpel supports the view that the involute closure is not a modification of the conduplicate. In primitive carpels, a narrow Fig. 75. Semidiagrammatic sketches of cross sections of syncarpous gynoecia show- ing examples of involute closure of carpels. In B, C, D, E, the carpel margins are obvious. A, Enjthraea Ccntaurium; B, Tetraclca coultcri; C, Isayithus hrachiatus; D, Clewdcndron fallax; E, Premna japonica; F, Limnophila heterophylla. (A, after Baum, 1949; B, C, after Stauffer; D, E, after Junell; F, after Hartl. ) marginal stiip of the lamina is sterile. In specialization of the carpel, this sterile band is progressively narrowed (Fig. 74A, 2) and, in some taxa, lost (Fig. 75B to £), and the ovules seem to be borne on the margin itself ( Fig. 75A, F ) . Where, in involute carpels, the sterile strip is lost and the edges meet directly, the ventral bundles and the pla- centae are brought close together and often fuse into a common bundle (Fig. 74B) and common placenta. Where the sterile margin is nar- rowed but persists, contact is by the abaxial surface, and the sterile band projects inward as two flaring edges (Fig. 74C, 1); contact is in the region of the vascular bundles. Where the inrolling is greater, con- tact is farther back on the abaxial surface, and the ventral bundles and placentae are well inside the locule (Fig. 74C, 2). Extreme inrolling is present in the Labiatae, where the inrolled borders may reach the mid- vein and sometimes become fused with tissues of that region. THE CARPEL 193 The difRculty of determining whether a carpel of advanced form, with apparently inrolled borders, is involute or conduplicate is in- creased by the trend throughout carpel specialization toward reduction of the sterile borders. In primitive conduplicate carpels, as in Degeneria, these borders are wide; in less primitive types, they are narrower and, in advanced carpels that can still be recognized as conduplicate, they are narrow. In carpels that are probably involute, the sterile borders range from fairly wide to absent, with ovules borne on the very edge. Intermediate stages show vestigial borders. The recognition of these reduced borders as such demonstrates an inrolling that brings together borders of the dorsal surface of the carpel. Evidence supplied by posi- tion of the marginal meristems in carpels of the rolled type is probably not on record and is important for the determination of basic types of carpel closing. Where the sterile margins are greatly reduced or lost, the two placentae are brought close together and unite, in greater or less de- gree, in a common placenta. The placenta may show evidence of its double nature in bilobed form. The ventral bundles, which underlie the placentae, may lie side by side or be fused as one simple bundle ( Figs. 74A, 4, 5 and B, 2, 3 ) , regardless of external form of the placenta. A deeply two-lobed placenta with the lobes divaricate and the epidermal layers continuous through the line of union is evidence of fusion by the dorsal surface well back from the edges (Fig. 74C, 2). Orientation of the ventral bundles may be of aid in determining type of folding. In typical conduplicate carpels, these bundles are half inverted; in carpels where edge meets edge or the edges are rolled in, they are typically completely, or nearly completely, inverted. In carpels apparently of involute type, whether in apo- or syncarpous gynoecia, inversion of the ventral bundles is typical. Variations in degree of inversion occur occa- sionally in individual carpels, but this accompanies variations in external form. The assumption that all carpel types are derived from the condupli- cate involves complicated changes in orientation of the borders and margins in the phylogenetic development of the more common types. In the conduplicate type, ventral surfaces of the carpel borders are in contact, and the margins lie side by side (Fig. 74A, 1, 2); in the more common types, the margins may come face to face, with no contact of the borders, or the margins, still side by side, are turned inward, and the dorsal surfaces of the borders are in contact (Fig. 74C, 1). Where the inrolling is continued beyond the line of contact, the borders flare and may recurve strongly (Fig. 74C, 2). It is difficult to see, in the involute-margin types, modifications of the conduplicate condition, especially in the light of the existence to- 194 MORPHOLOGY OF THE ANGIOSPERMS day of open carpels in unrelated taxa and the undoubted occurrence of independent closure in several or many phylogenetic lines. (Outstand- ing among the contributions of comparative morphology is the dis- covery of the frequent occurrence in phylogenetic progress of the same structural modification — dioecism, gamopetaly, zygomorphy, the in- ferior ovary, the solitary ovule, the vessel.) The carpel is closely like the leaf in many ways. In vernation, the leaf is folded or rolled in sev- eral ways: conduplicate, involute, revolute. In ecological modification, the leaf blade may, under reduction, be folded or rolled upward or downward, and no type apparently is basic to the others. It would be strange, in the light of the behaviour of the leaf in vernation and in ecological modification, if the carpel, so like the leaf in form, ontogeny, and anatomy, closed in only one way — that all forms of closed carpels have been derived from the conduplicate type. That the closed carpel is not an ancient angiosperm character is evi- dent not only in the presence of open carpels, even in some fairly ad- vanced families, and the frequent presence of those with borders merely appressed, but also in the presence of carpels closed by adnation to the receptacle. Fusion, by both connation and adnation, clearly pre- ceded closure of the carpel. Position, number, and size of ovules may, in part, determine method of enclosure. Conduplicate closing perhaps provides better enclosure where there are many ovules scattered over the lamina; involute, better space for ovules massed in longitudinal rows near the margins. Carpels that are clearly conduplicate usually have laminar or sublaminar placentation — an association of primitive characters, but not necessarily evidence that involute carpels also may not be primitive. The majority of carpels appear involute, with the sterile border greatly reduced or absent, and the margins meeting, edge to edge. Though the existence of marginal meristems in carpels is well known, little attention has been given to late stages in ontogeny, where position of the marginal meri- stems should aid in determination of the basic type of closing. Avail- able information indicates that, in at least some taxa, the marginal meristems meet directly — evidence of involute closure. Closure of the carpel, where ontogenetic, results in the formation of a more or less distinct ventral suture. The suture varies greatly in form and distinct- ness with the degree of inrolling and of completeness of histological fusion. The openings may extend the full length of the ovary or be restricted to the median or distal parts; in Platanus, the opening is very short. Closure of the carpel is clearly still in process of establishment in the angiosperms. The closed carpel is not a fixed, universal character in "angiospermy," though it serves well as one of a group of distinguishing characters. THE CARPEL 195 Closure by Adnation to Receptacle. Carpels that stand obliquely on the receptacle may be closed at the base by adnation of the margins to the receptacle. In these carpels, the locule is enclosed, in part, by tis- sues of the axis. Apparently, the ontogeny of these carpels has not been described, but the adnation is pi-obably congenital. The crescent-shaped carpel primordia and the primordium of the receptacle develop together; the flanges of the carpel primordium arise united with receptacular tissues. Higher up, the carpel margins unite with each other. Closure by Connation. In some syncarpous gynoecia, carpels open at the base are united by their lateral walls. Their locules unite, forming a common ovarian chamber — Phytolacca, Sabal and other palms. This type of gynoecium is part of the evidence that connation and adnation were present in the gynoecium before the carpel closed; that the closed carpel was not a character of the earliest angiosperms. Closure by Ontogenetic and Phylogenetic Fusion. The closure of the carpel is commonly ontogenetic; the margins of a primordium, cres- cent- or horseshoe-shaped in cross section, develop toward one another and unite. But the closure may be congenital; the carpel arises closed from a ring-shaped primordium. A primordium that is at first crescent- shaped may become ring-shaped at the base and, with continuing growth from the base, form a carpel that is closed distallv during ontogeny but arises fused proximally. In many advanced families, the carpel borders have become congenitally concrescent; the closed carpel is phylogenetically established. (Failure to recognize the nature of the ring-shaped carpel primordium has been largely responsible for the peltate theory of basic carpel form.) A common comparable condition in the corolla helps to make the complex carpel development clear. The typical gamopetalous corolla arises from a whorl of separate primordia, each representing one of the ancestrallv free petals. After elongation has continued for some time, the petal primordia fuse into a ring, which forms the corolla tube. The bases of the petals, evident by their inde- pendent vascular supplies, are fused congenitally (or phylogenetically); they arise fused. The Complex Carpel The differentiation of ovary, style, and stigma from the simple, primi- tive carpel was a gradual one, which took place in many lines, with elaboration of style and stigma proceeding at different rates. The stigma is sometimes sessile, the style undeveloped in some of the Winteraceae and Euptelea; the style may be well developed, the stigma still a primi- tive stigmatic crest, decurrent on the style, as in Cercidiphylhim. Mor- phologically, the style is usually the distal part of the primitive carpel. 196 MORPHOLOGY OF THE ANGIOSPERMS with its ovules lost; several families show gradual transition from ovary to style, with vestigial ovules and traces for lost ovules in the transi- tional region. Elongation of the style is characteristic of many taxa. The nature of the stigma is evident in the stigmatic crests of primitive carpels, especially those of the woody Ranales, such as Degeneria and the Winteraceae. The pollen-receiving surface of the stigma is char- acteristically papillose and often secretory. Where this area extends the full length of the carpel, forming an undifferentiated stigma, a stigmatic crest, the papillae cover the margins of the carpel and narrow adjacent bands of the lamina surface. Where the marginal areas are merely approximated or appressed, the carpel still open, the papillae fill the slit loosely or compactly. In some conduplicate carpels, the margins of the carpels flare (Fig. 72C, F, H), exposing a narrow band of the adaxial carpel surface and making the stigmatic crest two-lobed (Degeneria, Drimijs). The two-lobed form of crest, conspicuous in the most primitive carpels, is carried over into some advanced types and doubtless explains the two-lobed stigmas of some taxa. (The double lobing of stigmas of other taxa may be morphologically different.) In stigmatic crests, the papillae may occur not only on the carpel margins but on adjacent narrow bands of both dorsal and ventral surfaces ( Degeneria, Winteraceae ) . In the Winteraceae, they may extend inward on the ventral surface to the placental ridge, around the ovule bases, and, in Degeneria, farther in over much of the carpel surface, forming an internal continuation of the stigmatic area, even into sealed parts of the carpel. A similar "internal stigma" is present in Eupomatia, where, in a carpel completely closed by connation, stigmatic papillae extend downward from a distal pore to and around the ovules. In the Annona- ceae (Artahofrys) also, the papillate tissue extends downward along the carpel margins to the ovule bases, as a transmitting tissue for the pollen tubes. Internal stigmatic tissue in a sealed carpel is doubtless a remnant of the more extensive stigmatic area of primitive open carpels. The carpel, as a simple, folded, or involute laminar appendage, with several to many ovules, varies greatly in form: with or without clear distinction of ovary, style, and stigma; with or without style; stipitate or nonstipitate; completely or incompletely sealed. Evolutionary changes in tliese characters have progressed unequally, and primitive form in one character may accompany advanced form in another. The carpel of this type is follicular and matures into the fruit type termed a follicle. Modification of the Follicular Carpel Specialization of the follicular carpel has been primarily in reduction in size and in ovule number, simplification of vascular supply, and loss of dehiscence. In extreme reduction, only one ovule persists. The fruit THE CARPEL 197 formed from such a carpel as this, when indehiscent, is an achene. All stages in the development of the achene type of carpel are present in the Ranunculaceae and Rosaceae. The change is evident, superficially, in the shortening of tlie ovary and the presence of abortive or vestigial ovules. That the achene is a reduced follicle is shown in Fig. 42; the number of traces is reduced by loss and by fusion, and traces to lost ovules persist in the ovary wall. Clematis, Caltha, Calycanthus, and Adonis also show derivation of the achene from the follicle. Reduction and simplification of the carpel have occurred independ- ently in many families and apparently from follicular types that were at various stages of specialization: from those with well-developed style- — the style retained — as well as from those with sessile stigma; from stipitate carpels — the stipe retained — and from sessile carpels. Re- duction may proceed beyond functional form, with the last ovule lost and the locule nearly or quite closed — the "solid" carpel. The SoLm Carpel The term solid, as applied to carpels, became prominent in the 1920s and 1930s when the theory of carpel polymorphism aroused consider- able discussion. But the term was used in the older literature, much more soundly, for sterile carpels with locule greatly reduced or absent (Fig. 76), carpels that are vestigial or abortive. In its early use, the term was descriptive, and its nature, a reduction form, correctly in- terpreted. Sterile carpels are frequent in many taxa and show all stages in structural reduction — compressed and "consolidated" in various ways and degrees; rodlike, free or adnate to normal carpels; short stubs. In syncarpous gynoecia, they may be represented by vascular supplies only. Vestigial carpels are most readily recognized in staminate flowers of monoecious and dioecious genera. In some unisexual flowers that have no external remnants of reduced carpels, stubs of the carpel traces are present in the receptacle. In syncarpous taxa that show series in gynoecial reduction, solid carpels are often apparent as remnant struc- tures. Triglochin has obvious solid carpels; within the genus, carpel number is reduced from six to three; species with three normal carpels have three sterile carpels, with little or no locule (Fig. 76F). (The solid carpel as a reduction form is discussed further under Syncarpy. ) The Stigma and Transmitting Tissue In the more primitive carpels, the stigma is not clearly set apart: it merges into the style or ovary; in more specialized forms, it is usually distinct. In such genera as Degeneria and Drimijs, it is represented by the stigmatic crest, consisting of proliferated and papillose marginal bands extending the full length of the lamina (Fig. 72C). These bands 198 MORPHOLOGY OF THE ANGIOSPERMS form a pollen-receptive area, which is doubtless the primitive stigma. Restriction of the crest to its distal part, with reduction and sharper delimitation, has formed the specialized stigma of the higher taxa. In- termediate forms occur — Butomus, Calfha, TroUiiis. Some decurrent stigmas are probably specialized, rather than primitive, types, as in anemophilous taxa — Cercidiphijlhtm (Fig. 146), Juglandaceae, Thalic- trum, some Restionaceae (Anarthria). Fig. 76. Semisolid and solid carpels in syncarpous gynoecia. A, B, Glaucium flavum; C, Chclidonittm majus, one of die two pairs of carpels without loculus, one carpel showing ovules; D, Valeriana sp., two solid carpels, each with vestigial loculus and one normal carpel; E, Aclihjs sp., two carpels, one fertile and one solid (without loculus); F, Triglochin palustris, three normal, three solid carpels; G, T. maritima, six normal carpels; H, Triostcinn po-foliatiim, three normal carpels with large loculi, one solid, sterile, with narrow loculus; /, Linnaea borealis var. americana, three carpels with two adnate bracts, one normal, two solid. (A, B, C, after Van Tieghem; D, after Dijal; E, after Chapman; F, G, after Uhl; H, I, after Wilkinson, 1949.) The stigma and the conducting tissue — much better termed the transmitting tissue, because vascular tissue is commonly called the con- ducting tissue — represent the surviving and specialized parts of the stigmatic crest and the accompanying papillose carpel surface, which, in primitive carpels, extended between the carpel margins, over their internal surface, about the ovules, and even farther in on the laminar surface. In carpels with a stigmatic crest, the pollen tubes pass directly to ovules that are close to the place of germination. Restriction of the THE CARPEL 199 pollen-receiving surface involves, for the pollen tubes, a much longer course, but one still largely through the transmitting tissue. The transmitting tissue is commonly thought of as filling the stylar canal only, but it was shown as early as the 1870s and 1880s to extend. Fig. 77. Transmitting tissue. A, B, C, D, Verbascum thapsus: A, longitudinal section of style and stigma; B, section at o-a in A, c, stylar canal; C, D, cross sections of middle and base of style, respectively. E, F, longitudinal sections of ovaries of Adoxa moschatellina and Fedia conmcopioides, respectively, showing extent of trans- mitting tissue (shaded) in style and ovary; G to K, longitudinal section and cross sections of stigma of Heliotrophim pcnivianum: H, I, ], K, cross sections at levels 1, 2, 3, 4, of G, respectively. L, Philodcndwn coidatum, papillose transmitting tissue on receptacle and filaments; M, Vinca minor, papillae with drops of oil at base; N, Heliotrophim pcnivianum, vertical section of border of stigma, shown in G. (A to K, N, after Guegen; L, M, after Capus.) in the great majority of families, over the placenta and considerable parts of the ovary wall (Fig. 17 E, F). The presence of a papillose sur- face over areas of the wall where ovules have been lost is probably evidence of primitive laminar placentation. The transmitting tissue is formed from superficial carpellary — rarely ovular — tissues, from the epidermis, or from the epidermis and hypo- 200 MORPHOLOGY OF THE ANGIOSPERMS dermal tissues (Fig. 78). Cell modification, the development of papillae or hairs, and proliferation of tissues form the stigmatic surface. Cell divisions in the epidermis are chiefly tangential; those of inner layers are also largely tangential but may be in all planes. Papillose cells form the most common stigmatic surface. Cells of the transmitting tissue are delicate, more or less elongate, thin-walled cells, sometimes loosely arranged. They are glandular in function, with dense cytoplasm, large nuclei, and often considerable starch. The fluid secreted by them is said to be much like that of pollen-grain cytoplasm. Oil and other Fig. 78. Dcgcneria vitiensis. Micropylar part of ovule and adjacent carpellary tissue at pollination, showing transmitting tissue and course of pollen tube to micropyle. (After Swamtj.) fatty substances are frequently present, mingled with mucilaginous substances formed by cell-wall breakdown. Chlorophyll is common in all parts of the tissue. Some transmitting cells are simple or compound hairs, as in Zea. Plumose stigmas, such as those of many grasses and other anemophilous taxa, have numerous delicate branches that consist of papillose cells surrounding a single protoxylem element. Pollen grains reaching the stigmatic tissue are retained by entanglement among the projecting cells or are held on cell surfaces by their rough exines or oily surfaces, or by the mucilaginous or sugary secretions of the stigmatic cells. Cells of the transmitting tissues, especially those of the stylar canal and of the upper ovary wall, may become separated by the dissolution THE CARPEL 201 of the middle lamellae and float free in a mucilaginous fluid (Fig. 79E, F, H). This condition has been considered rare — "found in orchids only" — but was long ago reported for many taxa. Function of the Transmitting Tissue. The discovery, in 1823, of pollen germinating on the stigma aroused renewed discussion of the mystery of the method of fertilization. Two theories had been held in the pre- Oir^^^M^ .-^ MUmi^ Ohb' Fig. 79. Transmitting tissue. A, Vcrhascum vermile, transverse section of placental tissue, parenchyma cells swollen and filled with starch; B, Pittosporum sinensc, papillose surface of ovary wall near ovules; C, Solamtm glaucophijllum, transverse section of surface of placenta, epidermal cells filled with starch; D, Riibiis odoratus, transmitting tissue of the carpel margins at top of ovary; E, cross sections of pollen tubes filled with starch grains in the gelatinous substance formed from the break- down of the middle lamellae of the transmitting tissue; F, Deherainia smaragdina, cross section of part of the style showing stylar canal filled with mucilage in which the pollen tubes run; G, Cheirantlius cheiri, transverse section of ovary, showing central transmitting tissue; H, Fumaria major, cross section of stylar canal filled with granular mucilage and lined with a conducting epidermis against which the pollen tubes lie. {After Capus.) ceding century: that the "fertilizing dust" (pollen) contained an "esprit vohtil," which entered the vascular tissue of the stigma, passed down to the placenta, entered the ovules, and fertilized the "embryo"; that the pollen grains themselves "descended" into the ovary through open stylar canals (then believed to occur in all plants) and formed the "basis of the embryo." The discovery that, in many plants, the stylar canals were closed completely, or for part of their length, and that the 202 MORPHOLOGY OF THE ANGIOSPERMS pollen tube was seen to enter the stigma and its tip to be present at the ovule ended the pollen-grain-descent theory. The distribution of the transmitting tissue probably led to the discovery of its function. It had been noted also that considerable time elapsed between the presence of pollen on the stigma and the presence of an embryo. Finally, the pollen tube was followed from stigma to ovule, and a "double role" was assigned to the transmitting tissue: "to nourish the tube in its long course" and "to guide it to the ovule." Types of Transmitting Tissue. Transmitting tissue is apparently an elaboration of an area of simple, papillose epidermis; the cells separate and break down in part, forming a complex, often deep-seated, tissue. The nonpapillose transmitting tissue of water plants and of some lower monocotyledons probably represents a loss of papillose form in the cells; and the haii'-coated stigmas and styles of the grasses and sedges, a modification related to anemophily. Small, isodiametric cells form the stigmatic surface in many aquatic genera — Najas. In many wind- pollinated plants, the stigmatic surface is extended by hairlike processes; the important "receptive haiis" may be true hairs or merely tips of papillose epidermis cells. In the Liliales and some related taxa, the transmitting tissue is a simple epidermal layer, with few or no papillae. A most extreme form of the nonpapillose type is the multilayered, disintegrating tissue of the orchids. In dicotyledons, the transmitting tissue is usually papillose and reaches its greatest complexity in the gamopetalous taxa, where carpel margins are intimately fused and proliferated. Modifications of vascular tissue accompany unusual forms of the stigma. Where the stigma is simple and not well defined, as in Zea, the vascular tissue becomes gradually weaker, distally; where the stigma is large or lobed, the vascular tissue branches peripherally into many delicate strands. Form of the Stigma. Variations in form of the stigma are largely re- lated to the morphological structure of the carpel and gynoecium. A more or less distinctly two-lobed stigma, retaining, in part, the two- lipped form of the primitive ancestral crest, is frequent in less advanced taxa. Anatomically, two-lobed stigmas are of two types: the median vascular bundle dichotomously forked, the lateral bundles shorter; the median bundle short, the laterals prolonged. These vascular variations reflect the fundamental structure of the angiosperm node and ap- pendages — a median trace, double in nature, with a pair of laterals (Fig. 6C) (the "three-trace supply," formerly considered basic for carpels). Two-lobed stigmas seem to be primitive types; the highly specialized stigma is simple — globose or cylindrical — or elaborate in form — plumose or dissected. Greatly increased stigmatic surface is THE CARPEL 203 associated with wind pollination. A well-differentiated stigma — in con- trast with the stigmatic crest — and a sealed carpel have been called characters associated with insect pollination. The commissural stigma is discussed later in this chapter. The Stylar Canal. The stylar canal is a space enclosed by the uniting margins of a carpel — in syncarpous gynoecia, of all the carpels. It is usually closed for part or all of its length by ti-ansmitting tissue. In highly specialized syncarpous pistils, it may be complexly branched; the branches may extend as strips of transmitting tissue beyond the bases of the styles along diverse courses to the placentae. The transmitting bands may vary in color from the ovary wall and, in the dicotyledons, may be somewhat collenchymatous and easily dissected from the ovary walls. A hollow style is occasional — Viola, Campanula, Reseda, Liliiim, Yucca, Enjthwnium, Butomaceae. Pollen grains have been found in the open canal of Biitomopsis and HiUebrandia but have not been seen germinating there. In syncarpous gynoecia, the term stylar canal is often applied to the transmitting tissue of the style only. It has been suggested that it be applied to all parts of the branched transmitting system, but the distal parts may be surface areas; where there is com- plex extension of the tissue, it could well be termed the stigmatic- tissiie sijsicm. The Terms Stigma, Transmitting Tissue, Stigmatic Tissue, Stigmatic Crest. The stigma has been variously defined: morphologically, as the distal part of the carpel, modified in form in adaptation to pollen re- ception; functionally, as the tissues of the carpel, external and internal, concerned with receiving and transmitting the pollen tube. The trans- mitting tissue within the style and ovary has been called the internal stigma. Where carpel borders with papillose epidermal cells are ap- pressed and sealed ontogenetically, the external hairs are lost, and the internal hairs form an internal stigmatic line that extends downward along the placental line and around the ovules. In the primitive carpels of Etipotnatia, which are without style and with an external stigma of a mere tuft of apical hairs, connation has resulted in the loss of the stigmatic crest, except for tlie internal band of papillose cells (Fig, 72A). The genus Drimijs shows stages in the restriction of the stigmatic area to the distal part of the carpel and the internal surface. In the section Tasmannia, the carpel borders are only appressed, and closure is by glandular hairs; in the section Wintera, the borders are sealed, and the crest reduced to a distal area. If the terms transmitting tissue and conducting tissue were discarded, all the tissues that nourish and control pollen-grain germination and pollen-tube growth could well be 204 MORPHOLOGY OF THE ANGIOSPERMS called the stigmatic tissues. The stigma could then be loosely defined as the specialized distal part of the carpel, covered wholly or in part by stigmatic tissue. The stigmatic crest is a primitive form of stigma where the stigmatic tissue is present along the carpel margins, below as well as at the apex (Figs. 72C, 83). It has been claimed that, morphologically, the stylar canal represents the distal parts of modified or vestigial ventral carpellary bundles, but this theory is based on misinterpretation. Vestigial bundles, representing lateral veins, are often present in the style and may lie beside the stylar canal. Both ventral bundles and the canal are present even in one of the examples cited to support the theory. (The Boraginaceae show well the strands of transmitting tissue beside and distinct from the ventral carpellary bundles. ) Placentation Placentation in Free Carpels. Ovules are borne on the adaxial surface of the carpel lamina. Evidence from comparative studies and from anatomy demonstrates that there are no cauline ovules in the angio- sperms; all are borne on carpels, though some appear to be terminal on the receptacle. The ovules may be distributed over most of the surface of the lamina, with only a narrow marginal strip and sometimes the midvein line sterile, or they may be greatly restricted in location in several ways. The pattern of ovule arrangement on the carpel con- stitutes placentation. The area where the ovules are attached, usually more or less enlarged as an emergence of the laminar tissues, is the placenta. In primitive taxa, there may be little or no modification of the region where the ovules are borne; in these groups, the placenta is merely a location. But position of ovule attachment is important both morphologically and taxonomically. Absence of an emergent placenta is probably the primitive condition; Drimijs and Dcgeneria have only slight placental ridges. The ovule-bearing projection may be simple, a mere low cushion or ridge, but it is frequcndy large and complex in form and structure, and, together with the ovules, may nearly fill or even divide the locule (Fig. 80). It may form a major part of the ripe fruit and even project from the ovary walls. The form of the placenta varies greatly in number of ovules, type of ovule arrangement, and form of locule. When the placenta is close to the margin and where, in spe- cialized carpels, the marginal band is reduced or lost, interpretation of ovule position may be difficult. Where the fertile area is all or most of the ventral carpel surface, the entire surface is the placenta. Where the fertile area is restricted to sub- marginal areas and the margins are fused so that the fertile areas are brought together, the placenta is double in nature, consisting of strips THE CARPEL 205 from two lateral areas. Two-lobed form ( Fig. 80C, D, E ) may give evi- dence of this union, but often the fusion is anatomically complete and the placenta is simple in form and structure. It has been suggested that the term half placenta be used for each of the parts of this double structure, but this would involve unnecessary complications in descrip- tion, especially in syncarpy, where fertile areas from adjacent carpels Fig. 80. Chart showing placentation in the Gentianaceae. The variety of types show theoretical modifications of basic form and possible evohitionary relationships of types. Placenta merely a position in A, F, I, J- a ridge in H; a carpellary flange in B, C, D, E, G. A, Gcntiana chinensis; B, Chironia baccifera; C, C. palustris; D, C. dcnsiflom; E, Sabatia; F, Crawfordia fasciciilata; G, Bartonia virginica; H, Frascra speciosa; I, Gentiana linearis; ], G. crinita. {After Lindsey.) unite to form a common placenta. Placentae would have to be described —as were carpels under the polymorphism theory — as "one-half plus one plus one-half" in nature. The term placenta should be used loosely for the fertile area of a carpel or united carpels. Types of Placentation. In both morphology and taxonomy, the use of terms describing placentation is inconsistent and confused. A term given to placentation in free carpels has been applied to a morphologically different type in connate carpels. And differences in interpretation of the nature of the syncarpous gynoecium — whether it is wholly carpellaiy or 206 MORPHOLOGY OF THE ANGIOSPERMS partly cauline in nature — have greatly complicated the terminology of placentation. Morphological consistency requires interpretation of placen- tation in the simple, free carpel as the basis for that in the fused-carpel ovary. The follicle type of carpel, with the carpellary margins approximated or united, is generally accepted as primitive. In this carpel, there are two types of placentation. Commonly, the ovules are borne in rows near B Fig. 81. Diagrams showing subbasal placentation. A, B, Scirpus robustus, showing reduction from free central (one lateral ovule surviving on original central column); C, D, Saccharum and Bromus showing reduction from submarginal placentation: Saccharum in longitudinal section of ovary showing vascular supply to ovule, Bromus in cross section of ovary at level of ovule attachment showing three carpels (one fertile), three dorsal and three (double) ventral bundles, the ovule supplied by the large ventral; E, diagram of typical vascular supply for flower in Cyperaceae; F, series of cross sections of flower of Cijpcrus dcntatus showing origin and path of traces to carpels and stamens: three carpels, each with dorsal and two ventral traces, the pairs of ventrals uniting in a central column, which supplies the ovule traces. ( A, B, E, F, after Blaser. ) the margins — submarginal placentation; infrequently, they are dis- tributed broadly over the lamina — luminar placentation (also termed diffuse, scattered, superficial, reticulate ) . Laminar seems morphologically preferable. Reduction in ovule number in the carpel with submarginal placentation, together with modification of carpel size and form, brings about changes in apparent ovule position; the one, two, or few surviving ovules are usually borne near the base or apex of the locule, and the placentation is termed basal (Fig. 81) and suspended (apical) (Fig. 82), respectively. Basal placentation, derived directly from laminar emb Fig. 82. Sarcandra (Chlorantluis) . A to E, longitudinal diagrams showing vascular structure of ovary with solitary suspended ovule. Vascular supply to ovule from both ventral bundles in A, B, or from one only in C, D, E; F, cross section of fruit, hr, bract; emh, embryo; end, endosperm; int, inner integument; ni, dorsal bundle; per, pericarp; scl, stony layer; st, stamen traces; stg, stigmatic end of fruit; V, ventral bundle. ( After Swamy and Bailey. ) 207 208 MORPHOLOGY OF THE ANGIOSPERMS placentation, is apparently rare — Helobiales, Nymphaeaceae. Basal and suspended placentation in the syncarpous ovary may have a more complex origin but represent derivation indirectly from submarginal. It has been urged that the term parietal be applied to placentation in free carpels of the follicular type, here called submarginal, but parietal has long standing as applied to a type of placentation in syn- carpous ovaries and, so used, is a good descriptive term. Parietal, in its meaning of "on the wall," is applicable to both submarginal and laminar placentation, but seems best restricted to syncarpous ovaries, where it conti-asts well with placental positions of markedly different morpho- logical types. It has also been used as a synonym for laminar, especially where the ovules are very few and are isolated on the carpel wall be- tween the median and venti'al veins. Though "marginal" has long been generally accepted as a simple and primitive ovule position, the term submarginal — now commonly used — was applied to this type of placentation as early as 1850, but it did not continue in use. "Marginal" is especially undesirable, because it may suggest a nonlaminai" position, a position on the edge of the carpel. The critical study of the highly primitive carpels of the woody Ranales, together with ontogenetic studies in many taxa, has demonstrated that the primitive position of the so-called marginal ovules is submarginal, and "marginal" is incorrect, morphologically (Fig. 83). Study of the placentation of free carpels in many families shows that the submarginal position is characteristic of most taxa. In some taxa, die ovules are apparently borne on the edge of the lamina, but these are specialized carpels with lamina reduced and the strip of blade between the ances- tral ovule position and the margin narrowed or lost. This reduction is frequent in syncarpous ovaries (Fig. 84). These "marginal" ovules are shown by position of their primordia to be submarginal. Laminar placentation is probably present only in families generally accepted as primitive — Nymphaeaceae, Cabombaceae, Butomaceae, probably all of the Helobiales, Lardizabalaceae. There is strong evi- dence that laminar placentation is the ancestral type. In this type, the ovules, typically, are distributed over the entire lamina, as in Butomus, Hydrocharis, Nijmphaca, and derive their vascular supply chiefly from the smaller meshwork bundles of the lamina, rarely directly from the median or major lateral veins. In reduction in ovule number and re- striction to the submarginal position, a marginal sti'ip of the lamina and the median line become sterile, and the ovules derive their vascular supply from two major lateral bundles, one on each side, not from the laminar vascular meshwork. The Degeneriaceae and Winteraceae show evidence of transition from laminar to submarginal placentation. In Degeneria, though most of the ovules derive their vascular supply THE CARPEL 209 from the ventral veins, the distal ovules are connected with branches from the dorsal bundle, and a few ovules derive their supply from a meshwork of veins connected with both dorsal and ventral bundles (Fig. 83). In Drimt/s, the ovules are in two submarginal rows, with the ovule traces derived, in part, from the ventral bundles and, in part, from anastomosing branchlets from both dorsal and ventrals. These ■s^' V Fig. 83. Diagrams showing placentation transitional from laminar to submarginal as shown by vascular structure. Ovules in loose, submarginal rows with traces derived from dorsal and ventral bundles and from the small, connecting strands. A, B, Degeneria, showing variations in ovule supply; C, D, Drimys (Tasmannia type): C, lateral view of carpel; D, carpel split ventrally and spread open. For clarity, ovules are omitted, small circles and black dots indicate position of micropyles; dotted lines indicate position of stigmatic crest. (A, B, after Swamy; C, D, after Bailey and Nast.) 210 MORPHOLOGY OF THE ANGIOSPERMS two primitive families, with placentation transitional from laminar to submarginal, seem to show that the primitive ovule position was laminar and that the submarginal type has been derived from the laminar by restriction of ovules to the near-marginal areas. A similar evolutionary step is evident in the Nymphaeaceae, Cabom- baceae, and Ceratophyllaceae, where ovule number is reduced in the more specialized genera — Niipliar, Brosenia, Cobomba, CeratophijUum, Nelumbo. In most of the genera of the Nymphaeaceae, the ovules are numerous and distributed over the lamina; in Brasenia, there are only Fig. 84. Diagrams of cross sections of syncarpous ovaries showing varieties of placentation by differences in degree of closure of constituent carpels. A, C to F, forms of parietal placentation, by union of open or partly closed carpels; fusion between adjacent carpels by margins only in C, E, F; by margins and sides in A, D; B, axile placentation, by union of closed carpels. A, Tofieldia calyculata; B, Lilium regale; C, Platystemon sp.; D, Aristolochia clematitis; E, Argemone mexi- cana; F, Reseda lutea. {After Jtihnke and Winkler.) two ovules, described in taxonomic treatments as "dorsal," because they are borne close to the midvein. But they derive their vascular supply from anastomosing branchlets from both ventral and dorsal bundles (Fig. 85A), as do most of the ovules in typical laminar placentation. Cabomba has four, sometimes three, ovules (Fig. 85B). Where there are four, all are attached between the dorsal and ventral bundles, as in Brasenia. Where there are three, the third is on the ventral side and above the others, and the vascular traces of all three come from the ventral bundles: those to the lower ovules as laminar branches across the carpel sides, that of the upper ovules, dii-ectly. Ovule position and vascular supply in these two genera clearly indicate derivation from laminar placentation, as would be expected in these highly specialized genera of the nymphaeacean line. The carpel of CeratophijUum, with THE CARPEL 211 only one ovule, has a reduced vascular supply, consisting of simple, unbranched dorsal and ventral bundles (Fig. 85C). The ovule is de- scribed as "borne on the dorsal bundle." Though illustrations show no vascular trace, the ovule lies directly over the dorsal bundle and seems to represent a last stage in reduction from laminar placentation. These three genera apparently show derivation of few-ovule placenta- tion directly from the laminar type. In contrast, in carpels with one or two ovules in apocarpous Rosaceae and Ranunculaceae, basal placenta- tion is clearly reduced from submarginal (which, in turn, has been derived from laminar placentation) (Fig. 42). Fig. 85. Diagrams of ovaries showing reduced laminar placentation. A, Brasenia, two ovules at union of midvein and lateral veins; B, Cabornha, three to four ovules on transverse veinlets; C, CciatopJujUum, vascular supply reduced, remnants of dorsal and ventral bundles; D, Nelumbo, ovule solitary, attached at union of lateral veins. (A, B, D, adapted from figures arid descriptions of Strasburger and Saunders; C, after Eckardt.) Further evidence that laminar placentation is primitive is seen in the Butomaceae and Hydrocharitaceae, families in which this type of placentation is associated with open carpels and, in the Butomaceae, with decurrent stigmas. Laminar placentation, though little discussed, has been considered a specialized, rather than the primitive, type, a form derived from "marginal" by increase in the fertile area of the sporophyll. But the presence of the laminar type in primitive families of both dicotyledons and monocotyledons and of transitional forms in the Ranales and Cabombaceae supports the view that placentation in the angiosperms was primitively laminar. In greatly reduced placentation, interpretation of ovule position may be difficult; evidence from anatomy, ontogeny, and comparison with 212 MORPHOLOGY OF THE ANGIOSPERMS ovule position in related taxa are all necessary where a carpel has only one or two ovules. These ovules may be survivors of the many ovules of either laminar or submarginal clusters. The terms basal, median, distal, suspended are descriptive, rather than morphological. "Basal" should be interpreted as morphologically subbasal. The "basal" ovules of many taxa, such as Ariemonc and Potcntilla, which have abortive ovules (Fig. 42N) and traces to lost ovules along the ventral suture above the normal ovule, show that the functioning ovule is the lowest in the ancestral row. (These genera are achene-bearing taxa, with the achene obviously Fig. 86. Longitudinal sections of flower of Peperomia showing solitary, "basal" ovule with fomi of ovule trace, hooked, indicating ovule to be lateral, not basal and cauline. A, P. argijreia; B, P. cniapas. ( After Murty. ) reduced from a follicle.) That the apparently basal position in achenes is morphologically submarginal is evident in many taxa by the course of the ovular trace, which shows that the apparent basal position has been obtained by differential growth, a downward "migration of the ovule"; the trace descends to the so-called basal position from an originally more distal position (Fig. 86) — Ranunculus, Piper. (Solitary, so-called basal ovules in syncarpous ovaries may show similar evidence of change in position.) OlDJection has been raised that this roundabout course of the ovule trace has no significance, that it is the result of physiological demand, but vascular bundles, developing in relation to function, as in the fleshy tissues of fruits, do not follow circuitous routes. The term basal is perhaps more often applied to the solitary ovule of a syncarpous ovary, where it is considered to be cauline (Fig. 81). These ovules, as shown by anatomy and comparison with the placentation of THE CARPEL 213 closely related groups, are always appendicular. Interpretation of the basal ovule in free carpels as cauline is nonsensical, especially when applied to stipitate achenes; minute projections of the stem must be assumed to pass through the stipe of each carpel. The solitary surviving ovule of a submarginal row is usually the proximal member but may be the distal member, as in Leitneria. Rarely, if ever, is it a median one. The term medion placentation has been used occasionally — apparently always loosely and in a descriptive sense — for ovules borne "in the middle of the lamina," on the dorsal side, on the ventral side, or between these, as in Brasenia and Cabomba. Median has been suggested as a replacement for submarginal and, in "peltate" carpels, as a term to refer to the condition where the solitary ovule is borne on the "cross zone." Median, as a descriptive term for placenta- tion, is hopelessly loose morphologically and should not be used. It has been combined with an even poorer term, lateral — in "median lateral." Both "lateral" and "median" have been applied to ovules borne on the dorsal and ventral edges of the folded carpel and on the lamina between. The vestigial ovules in achenes which are obviously submarginal and ventral have been called "lateral median," and the ovules in achenes of the Ranunculus type have been described as "median lateral." These terms are morphologically meaningless and of doubtful descriptive value. Axial, parietal, and free central placentation are discussed under Syncarpy. The U-type Placenta. As a part of the demonstration of carpel mor- phology according to the peltate theory, a form of placenta called the U-type has been described and interpreted as "the placentation type of the angiosperms." The interpretation has been the result of the exten- sion of the theory of the underlying peltate nature of floral appendages to greatly reduced carpels, with emphasis on the "cross zone" as a morphologically important part of the organ, especially of the carpel. The U-shaped placenta has been given this name because it surrounds the base of the ventral furrow or slit of a so-called peltate carpel. It is described as present in its typical form in achenes of the Rosaceae and Ranunculaceae, where the ovule is borne at the base of the "slit" — and of the "U" — on the "cross zone," with the arms of the placenta extend- ing upward along the united margins of the carpel. The arms of the placenta, sterile in most of these taxa, may be fertile. The placenta, in these achenes, is merely a position, or area, where the ovule is borne. A consideration of placentation in general and the ontogeny and basic morphology of the carpel shows that the U-type placenta is merely a modified form of the submarginal type, a form resulting from advance in carpel closure from ontogenetic to congenital fusion, together with 214 MORPHOLOGY OF THE ANGIOSPERMS reduction in ovule number to one. The primordium of the more primi- tive carpels is crescent-shaped in cross section, and closure is onto- genetic; that of advanced carpels is ring-shaped in cross section, and the maturing carpel is tubular and its margins are united congenitally. The ontogeny of many carpels combines these two methods. The primordium, at first crescent-shaped, becomes ring-shaped later. The carpel, so formed, is distally open, and closed only in late stages; it is proximally tubular, arising closed. (A tubular corolla with petallike lobes has similar ontogeny.) The so-called U-type placenta is far from primitive and cannot be considered in any way a basic type. It is a high type of placenta, the result of extreme reduction in carpel length and ovule number. Ac- cording to the theory that the U-type is the basic kind of placenta, the typical double-row, submarginal placenta is considered derived from the U-shaped type. This is an example of reading a series from the apparently simple to the complex, because the simple was accepted as always primitive. The series obviously runs in the other direction; achenes are surely highly specialized carpels. In carpels in which the primordium changes during ontogeny from crescent-shaped to ring-shaped, the level of transition is a point or line where the margins meet. This transverse connecting bit of margin has been termed, under the peltate theory, the cross zone. It varies in width with the form of the crescent-shaped primordium but is usually very narrow. The peltate theory stresses the cross zone as of much morphological importance, but its existence is incidental to the change in primordium shape, the beginning of congenital fusion of the carpel margins. Recognition of the zone as significant of basal form in carpel morphology has led to far-reaching misinterpretations in structure. The cross zone has been considered a placenta because, as seen in achenes, the solitary ovule is attached at that level, apparently on the connecting bit of transverse margin. But the ovule is the lowest member in the submarginal placenta of the ancestral follicle, as shown by anatomy and by the presence, in Clematis and Anemone, of vestigial ovules above the normal ovule. Placentation called U-form has also been called lateral, median, median laminar, median lateral, ventral median, and submarginal lateral. Obviously, it is a reduction form of submarginal. In both apocarpous and syncarpous gynoecia, reduction in ovule number introduces much difficulty in the interpretation of ovule position and the terminology of placentation; comparative, ana- tomical, and ontogenetic studies are essential in interpretations. Reduction in Laminar Placentation. The evolutionary history of placen- tation is clearly one of reduction of ovules from many and indefinite in number to much smaller numbers and, ultimately, to two or one. THE CARPEL 215 The primitive carpel has numerous ovules distributed over the adaxial surface, except perhaps for a narrow marginal band. Ovules are occa- sionally borne even on the midvein area, though less abundantly than elsewhere. There seems to be no definite pattern of position or of origin of vascular supply. Each ovule is supplied by a slender trace, derived largely from the meshwork of small bundles of the lamina; the traces represent the tips of more or less free veinlets, or shorter strands derived at points of anastomosis in the meshwork and involving small veins only, or small and larger veins, even the midrib. With reduction in number, the ovules are first restricted to rows near the margins, with their traces derived from ventral veins — submarginal placentation. The rows are at first rather broad and loose, with traces of some of the ovules derived from the lamina meshwork {Degeneria, Drimijs; Fig. 83D); with further reduction, the rows become linear, with all traces from the ventral veins; further reduction of ovules in the submarginal lines leaves few, and ultimately two or one. The persisting ovules are commonly the proximal members of the row — often with vestigial, distal ovules above. Persisting distal ovules — two or one — are uncommon or rare ( Proteaceae ) , and median ovules are rare. "Basal," "suspended," and "median" solitary ovules, especially those of syncarpous ovaries, represent various morphological situations and require individual in- terpretations (Figs. 81 and 82). Solitary ovules may have more than one trace, and the traces mav come from the two ventral bundles or even, very rarely, in syncarpous ovaries, from the ventrals of different carpels. Fusion of the major carpellary veins — characteristic of many achenes (Fig. 42) — may suggest the derivation of ovule traces directly from the dorsal vein. The Nymphaeaceae and Cabombaceae provide excellent examples of reduction to few, two, and one ovule directly from lamina distribu- tion. The vascular structure of the carpels shows that, in reduction in ovule number, intermediate submarginal stages have been omitted. Most of the genera in the Nymphaeaceae have the numerous ovules scattered widely over the carpel surface; Neltimbo has only one; in the Cabomba- ceae, there are three (rarely four); and in Brasenia, two (rarely three) (Fig. 85). Ovular position in these genera has long been a morphologi- cal puzzle, but their positions on the carpel walls and the derivation of their traces show that they represent scattered members of ancestral laminar distribution. In taxonomic treatments, ovule position in these taxa has been described as "parietal" and "on the sides of the carpel" — more speci- fically, in Cahomha, "on the lateral walls" and, in Brasenia, "along the dorsal suture"; these locations are also called by the morphologically better terms "laminar lateral" and "laminar dorsal," respectively. In 216 MORPHOLOGY OF THE ANGIOSPERMS Cabomba, two of the three or four ovules are borne on the side walls, near or below the center of the carpel, the others similarly above (Fig. 85B). Horizontal veinlets extend from the midrib and from the laterals across the carpel side just above the ovules, and the ovule traces are derived at the point of anastomosis of each pair. This has been de- scribed as "suspended by the sling method." In Brasenia, the vascular system is similar, but the ovule trace is derived at the point of union of the midvein and a transverse veinlet; the ovules, therefore, are borne along the midvein. In Nelumbo, the ovule is attached along the ventral suture, deriving its trace in the "sling" fashion at the point of union of horizontal veinlets which connect two "posteriolaterals." Cerafophijllum has a solitary ovule attached along the dorsal vein. In these four genera, the placentation is clearly modified laminar (Fig. 85). Some of the Magnoliaceae (Michclia) also show reduction of laminar placentation — ovule supply from both dorsal and ventral veins. Morphological Nature of the Carpel The problem of the fundamental nature of the carpel — axial or ap- pendicular — leads far back in the history of vascular plants, back to the differentiation of the ancient, thalloid plant body. Because of similarity in form and structure, the leaflike nature of the carpel has been recognized since the earliest days of plant study. Detailed com- parative studies in ontogeny and anatomy, especially those of nodal structure, have shown that the similarity is much closer than had been believed; that many supposed differences in position, origin, and de- velopment of these organs are based on misinterpretations. Discussion of the possible homology of carpel and leaf has been prominent in botanical history. The carpel has been called a fertile leaf, and the leaf, a sterile sporophyll. Leaf and sporophyll have been called organs sui generis. The theory of phyllospory and stachyospory has added the new term stegophylls for carpels that "surround" and "protect" stachyo- sporous ovules. Interpretation under this theory implies that the ap- pendages commonly called carpels are of two basically different types — foliar and cauline — an interpretation that is unacceptable by morphol- ogists. In 1956, the interpretation of the carpel as a leaflike appendage was challenged again; it was considered axial, a part of the receptacle, on which are borne "buds," the ovules. In support of this view, the phyllospory-stachyospory theory maintains : 1. The inversion of bundles has no significance. The inversion is merely the "aberrant result" of the "moving inward" of a vascular bundle to a part radially inward from its point of origin. 2. Pairs of normally oriented bundles in the ovary wall — as in parietal placentation — are merely "two adjacent bundles." THE CARPEL 217 3. The vascular systems supplying ovary wall and placenta are independent. 4. The recent recognition of the two-trace nature of the primitive angiosperm leaf breaks down the homology of carpel and leaf, for the carpel is a three-trace organ. 5. The sessile nature of carpels is evidence that the carpel is axial, a part of the receptacle. 6. The placenta is an independent structure, also of axial nature, bearing "terminal buds" (ovules). In support of the appendicular theory, are the following facts: 1. The inversion of the lateral bundles is clearly the result of up- turning and inrolling of the sides of laminar organs; they do not arise inverted but are inverted during their course to the carpel or within the carpel base. Similar inverted bundles are common in petioles, where there has been a rolling of the margins and a flattening. Inverted bundles are also found in many veins in cylindrical and rolled leaves of xero- phytes. 2. The pairing of normally oriented bundles in a syncarpous ovary wall is not a mere "happenstance." The pairs are the ventral bundles of adjacent carpels, fused margin to margin, as open and partly upfolded laminar carpels. The bringing together is ontogenetic in some taxa; congenital in others. 3. Independence of vascular systems of ovary wall and placenta is normal; the traces which supply these parts arise at different levels on the stem and wholly independently. (They may be freely connected by small veins.) If carpels are to be interpreted as axillary on this basis, so are leaves and all other floral organs in which there are several traces. In laminar placentation, there is no distinction of vascular "systems." 4. The dorsal trace of the carpel is frequently double; the stamen, in some families, has two traces; the cotyledon, in most families, has two or four traces and shows the origin of the three-trace condition by fusion of the median two (Chap. 9). Anatomically, carpel and leaf are now brought even closer together. (Recognition of the two-trace node as primitive has replaced the long-held theory tliat the three-trace node was primitive. ) 5. The carpel in primitive families, both monocotyledonous and dicotyledonous, is stipitate; the sessile carpel is specialized. 6. The placenta is merely a location on the carpel where ovules are borne and where, often, the fertile area becomes enlarged. It appears to be an independent structural part only in highly specialized ovaries where greatly enlarged or where lateral-wall connections have been lost, as in free central and basal placentation. Where enlarged, it has no uniform pattern of vascular supply, as an independent axis would 218 MORPHOLOGY OF THE ANGIOSPERMS have. If a carpel be considered an axis, it would be a hollow structure containing other axes, the placenta and its branches, the ovules. The claim of homology of carpel and axis fails to explain the existence of open carpels and the ontogenetic closing of carpel primordia. It com- pletely disregards anatomical structure and ontogeny, and is valueless in morphological interpretations. When evidence from all fields is considered, none of the twentieth- century concepts of t