The Evolutionary Origin of the Integument in Seed Plants

Home , Aril, Catha (plant), Celastrus orbiculatus, Celastrus paniculatus, Comastoma, Euonymus europaeus

The evolutionary origin of the integument in seed plants

Anatomical and functional constraints as stepping stones towards a new understanding

DISSERTATION

to obtain the degree Doctor Philosophiae (Doctor of Philosophy, PhD) at the Faculty of Biology and Biotechnology

RUHR-UNIVERSITÄT BOCHUM

International Graduate School of Biosciences Ruhr-Universität Bochum

Evolution and Biodiversity of Plants

submitted by Xin Zhang

from Urad Qianqi (Inner Mongolia, China)

Bochum October 2013

First supervisor: Prof. Dr. Thomas Stützel Second supervisor: Prof. Dr. Ralph Tollrian

Der evolutionäre Ursprung des Integuments bei den Samenpflanzen

Anatomische und funktionale Untersuchungen als Meilensteine für neue Erkenntnisse

DISSERTATION

zur Erlangung des Grades eines Doktors der Naturwissenschaften an der Fakultät für Biologie und Biotechnologie

RUHR-UNIVERSITÄT BOCHUM

Internationale Graduiertenschule Biowissenschaften Ruhr-Universität Bochum

angefertigt am Lehrstuhl für Evolution und Biodiversität der Pflanzen

vorgelegt von Xin Zhang

aus Urad Qianqi (Innere Mongolei, China)

Bochum Oktober 2013

Referent: Prof. Dr. Thomas Stützel Korreferent: Prof. Dr. Ralph Tollrian

Contents I

1 Introduction 1

1.1 The ovule of gymnosperms and angiosperms 1

1.2 The theories about the origin of the integument in the nineteenth century 1

1.3 The theories about the origin of the integument in the twentieth century 1

1.4 The pollination drop 5

1.5 Ovule and pollination in Cycads 6

1.6 Ovule development in Magnolia stellata (Magnoliaceae) 6

1.7 Aril development in Celastraceae 7

1.8 Seed wing in Catha edulis (Vahl) Endl. (Celastraceae) 8

1.9 Ovule development in Homalanthus populifolius Graham (Euphorbiaceae) and differentiation of caruncula and aril 10

2 Material and methods 12

2.1 Material collection and preparation 12

2.2 Scanning Electron Microscopy (SEM) 12

2.3 Anatomical studies 13

3 Results 14

3.1 The morphology of Zamiaceae 14

3.2 Ovule development and seed anatomy in Zamia L. 16

3.3 Ovule morphology and anatomy in Cycas revolutaThunb. 22

3.4 Ovule development, seed morphology and anatomy in Magnolia stellate (Siebold & Zucc.) Maxim. 26 Contents II

3.5 Aril development in Celastraceae 32

3.5.1 Celastrus orbiculatus Thunb. 32

3.5.2 Euonymus europaeus L. 32

3.5.3 Euonymus planipes Koehne 33

3.6 Ontogeny of the ovule and seed wing in Catha edulis (Vahl) Endl. (Celastraceae) 41

3.7 Ovule development in Homalanthus populifolius Graham (Euphorbiaceae) and differentiation of caruncula and aril 45

3.7.1 Homalanthus populifolius Graham 45

3.7.2 Passiflora citrina J.M. MacDougal 46

4 Discussion 52

4.1 Integument and aril or aril-like structure in gymnosperms 52

4.2 Evolution of the (only) integument 53

4.3 The structure of the ovule within angiosperms 55

4.4 The envelope of the ovule within angiosperms 55

4.5 The function of the second envelope of seed plants 57

4.6 Origin of ovule and integument 59

4.7 Ovules in Lyginopteridatae 59

4.8 The pollination drop 60

4.9 Pollination in Cycads 61

4.10 Aril development in Celastraceae 62 Contents III

4.11 Seed wing development in Catha edulis (Vahl) Endl. 64

4.12 Caruncula and aril 67

4.13 Stomata on the outer integument 68

5 Summary 70

6 Zusammenfassung 72

7 Bibliography 74

8 Appendices

Curriculum Vitae

Published article

9 Acknowledgments

List of figures IV

Fig.1: Zamia amblyphyllidia D.W. Stev. in the Botanical Garden Bochum 14

Fig. 2: Male strobilus of Zamia amblyphyllidia D. W. Stev. 15

Fig. 3: Female strobilus of Zamia amblyphyllidia D. W. Stev. 15

Fig. 4: Microsporophyll of Zamia amblyphyllidia D. W. Stev. 16

Fig. 5: Ovule development of Zamia amblyphyllidia D. W. Stev. I 18

Fig. 6: Ovule development of Zamia amblyphyllidia D. W. Stev. II 19

Fig. 7: Ovule development of Zamia amblyphyllidia D. W. Stev. III 20

Fig. 8: Seed anatomy of Zamia amblyphyllidia D. W. Stev. 21

Fig. 9: Cycas revoluta Thunb., young leaves and female cone 22

Fig. 10: Megasporophyll and ovule of Cycas revoluta Thunb. 23

Fig. 11: Anatomy of the ovule of Cycas revoluta Thunb. 25

Fig. 12: Magnolia stellata (Siebold & Zucc.) Maxim. in the Botanical Garden Bochum 26

Fig. 13: Flower buds, flower, androecium, young fruit, mature fruit and seeds of Magnolia stellata (Siebold & Zucc.) Maxim. 27

Fig. 14: Ovule development in Magnolia stellata (Siebold & Zucc.) Maxim. 29

Fig. 15: Successive processes of the micropyle development of Magnolia stellata (Siebold & Zucc.) Maxim. 30

Fig. 16: Seed morphology and anatomy of Magnolia stellata (Siebold & Zucc.) Maxim. 31

Fig. 17: Celastrus orbiculatus Thunb.: Ovule, successive developmental stages 35

Fig. 18: Celastrus orbiculatus Thunb.: Caruncula, successive developmental stages 36

Fig. 19: Euonymus europaeus L.: Ovule, successive developmental stages 37

Fig. 20: Euonymus europaeus L.: Caruncula, successive developmental stages 38

Fig. 21: Euonymus planipes Koehne: Ovule, successive developmental stages 39

Fig. 22: Euonymus planipes Koehne: Caruncula, successive developmental stages 40

Fig. 23: Catha edulis (Vahl) Endl.: Flower and ovule, successive developmental stages 42

Fig. 24: Catha edulis (Vahl) Endl.: Caruncula, successive developmental stages 43

Fig. 25: Catha edulis (Vahl) Endl., younger and mature stage of the fruit 44

Fig. 26: Homalanthus populifolius Graham: Carpel, successive developmental stages 47 List of figures V

Fig. 27: Homalanthus populifolius Graham: Ovule, successive developmental stages 48

Fig. 28: Homalanthus populifolius Graham: Caruncula, successive developmental stages 49

Fig. 29: Homalanthus populifolius Graham: Stomata on the outer integument and abaxial side of the leaf 50

Fig. 30: Passiflora citrina J. M. MacDougal: Aril, successive developmental stages 51

Fig. 31: Seed envelopes in gymnosperms. 52

Fig. 32: Ginkgo biloba L.: Ovule. 53

Fig. 33: Pinus parviflora Siebold & Zucc.: Ovule. 53

Fig. 34: Podocarpus macrophyllus (Thunb.) Sweet var. marcophyllus: Ovule. 53

Fig. 35: Structures of angiosperm ovules. 54

Fig. 36: Liriodendron chinense (Hemsl.) Sarg.: Two integuments. 55

Fig. 37: Actinidia arguta (Siebold & Zucc.) Planch. ex Miq.: Unitegmic ovule. 56

Fig. 38: Typical aril and caruncula in Celastraceae. 56

Fig. 39: The telome hypothesis after Andrews (1961). 59

Fig. 40: Structures of the ovules in Lyginopteridatae. 59

Fig. 41: Pollination drops of Zamia amblyphyllidia D.W. Stev. and Cupressus arizonica Greene var. glabra (Sudw.) Little 60

1 Introduction

1.1 The ovule of gymnosperms and angiosperms

The ovule of gymnosperms and angiosperms is the female structure with at least one or two envelope structures covering the central part called nucellus or megasporangium. Within the nucellus, a megaspore develops into a haploid megagametophyte and produces the egg that is fertilized by the microgamete and forms an embryo. The envelope structure is called integument and surrounds and protects the nucellus.

1.2 The theories about the origin of the integument in the nineteenth century

During the later part of the nineteenth century a number of conflicting theories based on anatomical, developmental and teratological evidence from extant plants were put forward but no definite conclusion was reached. Worsdell (1904) discussed the three principal theories for ovule origin advanced during the nineteenth century, namely the axial theory, the foliolar theory and the sui generis theory. The axial theory regards the nucellus as a shoot axis bearing foliar appendages fused to form the integument. This is supported by the morphology of gymnosperms, especially the strobilus structure of conifers. The foliolar theory held the ovule to be morphologically a phyllome; it relates the ovule to a three-lobe leaflet of the female sporophyll (or carpel in the angiosperms); the integument comes from the lateral ones; the nucellus is an emergence borne on the cup-shaped terminal lobe. The foliolar theory has various forms. It is supported by the angiosperm ovule particularly (Eames 1961). The sui generis theory is that the ovule is a new structure which is not homologous to either caulome or phyllome. These three theories generally lost influence. However, one can also find their shadow in some of the botanical text books today.

1.3 The theories about the origin of the integument in the twentieth century

In the 20th century, the consideration of the ovule origin has focused largely on Pteridospermophyta, which are generally accepted to include the oldest seed plants. Benson (1904) noticed that the synangium of Telangium scottii Benson and the 2 ovule of Lagenostoma Williamson are very similar to each other. She hypothesized that the integument evolved from a sterile outer ring of sporangia in a radial synangium. This is the first theory of the pteridosperm ovule based on fossil evidence. According to her theory, the number of spores in the central sporangium was reduced to one which finally produced the female gametophyte.

Oliver and Scott (1904) hypothesized the evolution of the integument from a cup-like indusium in their monograph on Lagenostoma lomaxii Williamson. With regard to the origin of the ovule they stated that “we have in Lagenostoma Williamson a megasporangium which has been enclosed by two successive, concentric, indusium- like structures of which the inner has become an integral part of a new organ, the seed; the outer is probably of later origin…" It is quite possible that the two enclosures have originated very similarly, i.e. as peltate, lobed structures, and that the present integument was once a comparatively unspecialized, cupule-like indusium”. In the monograph on Lagenostoma Williamson they state “a comparison of the seeds of Cycads with Lagenostoma Williamson is inevitable”. Stopes (1904, 1905) and Matte (1904) supported this idea with their work on the integument structure of the Cycads. They showed independently that the inner vascular system of the cycad ovule supplies the inner part of the integument but not the nucellus as previously thought, and that the integument itself has a double structure in respect to vascularization. Stopes considered that the integumentary vascular system of Lagenostoma is equivalent to the inner system of Cycads and that the cupular system is equivalent to the outer Cycads system. The integument of Lagenostoma Williamson is thus regarded as homologous with the inner layer of the Cycads integument and the cupule is homologous with the outer part. As the three papers (Oliver and Scott 1904; Stopes 1904; Matte 1904) dealing with Cycads and Lagenostoma Williamson appeared simultaneously, the comparison does not cover all details. Stopes (1905) stated “I will not attempt to do this now in detail, but there are one or two points about which I should like to add something to my published view”. Work on the anatomy and morphology of living Cycads revealed that their integumentary structure is more complex than was generally supposed (Stopes 1904). In fact that this comparison of Cycads and Lagenostoma Williamson has not been done until now. 3

The investigations concerning the ovule started from the conditions of the angiosperms, and in order to state the nature of nucellus and integument, five different approaches were used, ontogeny, anatomy, topographical morphology, phylogeny and teratology (de Haan 1920). De Haan (1920) reviewed the current views on the origin of the ovule and put forward four important questions:

1. Are integuments of the pteridosperms, gymnosperms and angiosperms comparable, or is their origin polyphyletic?

2. If the integuments are homologous organs, are they also homologous with the indusium of ferns?

3 Is the integument composed of several units, which may be evident as ribs or sutures or as slips at the micropyle?

4. If the integument is composed of units, to which bauplan elements are they homologous?

After surveying the ovules of all groups of seed plants except the angiosperms, de Haan reached four conclusions at the end of his paper:

1. The integuments of the pteridosperms and gymnosperms are homologous and most probably also with those of the angiosperms.

2. The integument is not homologous with the indusium of the ferns and lycopods; “indusium” should not be used for these organs.

3. The integument is originally composed of units.

4. The value of the units is not yet certain.

These units are now referred to as telomes based on the idea of de Haan (Smith 1964; Herr Jr. 1995). Zimmermann (1938, 1952) developed this concept of telomes which conceived a dichotomously branched axial system bearing terminal sporangia as the evolutionary starting point. The ultimate axes are termed telomes, either sterile or fertile. Parts between two branching points are called mesomes. Through time there is a gradual reduction of some of the axes so that a single sporangium becomes surrounded by an aggregation of sterile processes. A number of Devonian- Mississippian seeds exhibit some of the structural modifications suggested by the 4 telome theory, as the earliest integuments are formed of unfused axes surrounding the nucellus. This theory has been revised by several investigators (Walton 1953; Andrews 1961; Smith 1964; Long 1966; Gensel and Andrews 1984; Taylor and Taylor 1993; Herr Jr. 1995; Taylor et al. 2009), and the concept is regarded as the most plausible to date (Herr Jr. 1995). Smith (1964) considered it to be the only theory to account for the structural variations in pteridosperm ovules. However, there are two things that should be noticed. The first step of this concept is hypothetic; second, what is the structure of the telome and how this structure was evolved to fit the function of pollination and fertilization is not clear.

The telome concept is widely considered the most plausible one. Andrews (1961) reviewed the theories on the evolution of the ovule and considered that except the telomic concept there is another principal theory, the nucellus modification concept. Andrews (1961) starts the evolutionary series from the megasporangium of the Lower Carboniferous fern Stauropteris burntislandica Bertrand which has only two functional megaspores. The tapering apex of the sporangium often has a small pore, and the sterile lower half is traversed by a central vascular strand. It seems probable that the spores were not shed from the sporangium.

Andrews conceived the following sequence of events:

1. A reduction to the one megaspore.

2. The ‘sinking’ of the megaspore towards the base of the sporangium.

3. The division of the vascular strand into a number of branches which ‘grow up’ round the megaspore.

4. Slight modification of the sporangial wall and apex. This would result in an organ similar to the vascularized nucellus of the Trigonocarpalean ovule.

5. Further modification, presumably growth and lobing at the apex of the sporangium, resulting in an ovule of the lagenostomalean type.

Walton (1953) hypothesized that a portion of a frond or a leaf became enrolled, based on the presence of vascular tissue in the peripheral nucellus of trigonocarpels, which he considered as an integument fused to a strongly reduced nucellus. He 5 regarded the previously designated integument as equivalent to the cupules of Lagenostoma Williamson.

Camp and Hubbard (1963) proposed the double integument theory by which Paleozoic ovules were regarded as having two adnate integuments derived from sterile dichotomous lateral branches subtending a terminally borne megasporangium. The inner set of branches fused with the nucellus so that the ovules appear to have only one integument. From a comparison of pteridosperm and Cycas L. ovules with those of angiosperms, they suggested a greater homology among ovules of different groups that all ovules have two integuments whether separate or fused. Meeuse and Bouman (1974) also set a similar hypothesis.

The morphologic unit that has received the most attention in discussions on the evolution of the seed is the integument (Taylor et al. 2009; Herr Jr. 1995; Smith 1964). The only point of agreement in the different theories on the origin of ovule is that the entire nucellus is a megasporangium that retains a single megaspore and the endospermic female gametophyte (Herr Jr. 1995). However, the detailed structure of the nucellus itself is ignored.

1.4 The pollination drop

The pollination drop was present even in the Paleozoic times (Singh 1978). The most obvious adaptive significance of an integument appears to be an increased protection of the megagametophyte and embryo, but the free integumentary lobes of the earliest seeds may also have served to increase pollen capture (Taylor et al. 2009). Fossil Cycads are known from the Lower Permian in China 270-280 million years ago, and the group is thought to have arisen from within the ancient seed ferns, of the later Paleozoic era, which ended 250 million years ago.

Normally, fertilization is the key point in the development from the ovule to seed. Prior to fertilization, the ovule develops. The covering sheath is called integument. After fertilization, the ovules mature to form seeds. The integument becomes the seed coat or testa. However, compared to seed development fertilization is just a very short time, and in the development from the ovule to the seed there is no obvious interruption. So here the development of the ovule is the development from 6 the primordium to the immature seed. The covering structure is the integument even in young stages of a seed after fertilization.

1.5 Ovule and pollination in Cycads

Cycads are well studied in respect of morphology and anatomy. However, in conifers freely exposed pollination drops are well documented while similar records are lacking for Cycads. There must be a liquid phase involved as spermatozoids would not swim otherwise. On the other hand artificial pollination under greenhouse conditions is performed in Cycadopsida by rinsing water with dispersed pollen through the strobilus. Freely exposed pollination drops would be washed away applying this method. The method is working successfully, but this is in conflict with our present knowledge of the pollination procession of Cycads. Thus, a more detailed analysis is essential.

“Interest in the morphology of the ovules of living Cycads has been much stimulated by the recent appearance of several papers dealing with the structure of Cycad-like seeds of Carboniferous age. The structures found in these fossils are not only of great interest in themselves, but also seem to throw light on some of the difficult points in the morphology of recent seeds, and it became clear that a detailed examination of the living forms might be undertaken with advantage” (Stopes 1904).

1.6 Ovule development in Magnolia stellata (Magnoliaceae)

The structure of Magnolia seeds is very special in angiosperms. In 1848, Gray first described, in the ‘General of North American Plants Illustrated, ’vol. I., that the seeds of Magnolia were not arillate, but baccate, or in other words that the fleshy coat was the testa. From 1856 on, there is a debate between Miers and Gray, the two most famous botanists at the same period of Darwin, whether the fleshy structure of the seed is an aril or a testa. Miers (1856) first criticized Gray that “the external fleshy coat of the seeds of Magnolia is described as testa and its thick bony shell as tegmen, or inner integument, the true integument having escaped the notice of that excellent botanist”, and “there is no reason to doubt that in Magnolia the scarlet envelope is due to a subsequent growth over the primine”, and the fleshy coat is therefore an arillus. In 1857, Gray settled three points by direct observation. First, no 7 accessory cover or arillus was developed over or upon the primine of the ovule. Second, the fleshy envelope of the seed represents the primine or outer coat of the ovule. Third, the bony coat of the seed was represented in the ovule only by the innermost layer of young cells, lining the primine; which cells, multiplying by meristematic division during the growth of the seed and their walls at length thickening and hardening irregularly, form the crustaceous or bony coat; so that the character of the seed is best expressed by the term ‘drupaceous’. These results had been substantiated by Maneval (1914) and Earle (1938). Kapil and Bhandari (1964) found the same structure as Earle (1938) in different species. Boer and Bouman (1972) made a very detailed observation of the integuments development of Magnolia stellata and Magnolia virginiana. They found that the inner integument only became thicker by “secondary” cell divisions in the micropylar region and that it does not contribute substantially to the formation of the testa. It persists as a “papery” layer. However, Corner (1976) pointed out that the lignified cells of the endotesta are peculiar and will need study by electron-microscopy for elucidation which is beyond the resolution of the ordinary microscope. Matsui et al. (1993) investigated the ovular development and morphology in some Magnoliaceae species with SEM and microtomy. They found that the outer integument and the funicular outgrowth together form an envelope complex. In 2003, Yamada et al. rechecked the outer integument and funicular outgrowth complex in the ovule of Magnolia grandiflora. The interpretation of a single cupular outer integument is not supported.

Here Corner’s advice is followed to check the later lignified cells of the endotesta and the inner integument around the micropylar region in a mature seed and to compare the outer integument of Magnolia to the only integument of Cycas, which develops into sarcotesta and sclerotesta, too.

1.7 Aril development in Celastraceae

To learn more about the evolution of secondarily intercalated seed envelopes, developmental studies of arillate seeds in gymnosperms and angiosperms should be undertaken. The goal is to test whether the second (outer) integument could be derived from an aril of gymnospermous ancestors. 8

The aril is a very conspicuous and taxonomically important character within Celastraceae. Planchon (1845) described the putative arils of Celastrus scandens and Euonymus latifolius as arillodes – false arils. Planchon cited these false arils as derived from the exostome of the integument rather than from the funiculus. However, Miers (1856) disputed Planchon’s conclusion and, based on his own investigation of Euonymus europaeus, he concluded that the aril is derived from the funiculus and is therefore a true aril. Pfeiffer (1891) described the arils of Celastrus, Euonymus and Gymnosporia cassinoides as derived from the exostome and the hilum. Corner (1976) described the aril of Euonymus glandulosus as derived entirely from the funiculus, and the aril of other species (Catha edulis, Celastrus paniculatus, Sarawakodendron filamentosum) as derived from the exostome and the funiculus. Van der Pijl (1972) concluded that Euonymus has an arillode, not an aril, but a detailed study on the seed development is still wanting. This study focuses on the different stages of development and intends to show the whole process of aril development in Celastraceae, and thus may help to understand the evolution of this family and the order Celastrales sensu lato which is a loose assemblage of probably not closely related taxa (Shisode and Patil 2011). It may furthermore be helpful and serve towards an evolutionary understanding of the formation of different types of fleshy seed appendages.

1.8 Seed wing in Catha edulis (Celastraceae)

The seed wing of Catha edulis (Vahl) Endl. (Celastraceae) has been described as an aril derived from the funiculus, while the aril of Celastraceae is considered arising from the micropyle instead of the funiculus. So as a special species in Celastraceae, Catha edulis (Vahl) Endl. is selected for study the ontogeny of the ovule and the seed wing.

Wings as a means of dispersal occur not only on the level of entire fruits but also on the level of seeds. In both cases, structures which are equivalent in respect to function have evolved in different ways from different origins. For understanding the structure of a seed, an adequate knowledge of the development of the ovule is essential (Boesewinkel and Bouman 1984). Here we focus on a seed wing in a group which is generally known for fleshy appendages (Simmons and Hedin 1999; 9

Simmons 2004; Simmons et al. 2008). The aril development in Celastraceae was studied by Zhang et al. (2012), and they showed that the fleshy structure of Celastraceae is rising from the micropyle instead of the funiculus. From the morphological and developmental perspective, the difference between aril and caruncula depends on the position of the micropyle. An aril develops from the funiculus, so the micropyle in the later development is covered by an aril. As the caruncula initiates from the micropyle, it can always be recognized by the micropyle in the centre of the caruncula (Kapil et al. 1980). However, apart from fleshy seed appendages dry seed wings occur in Celastraceae, too (Simmons and Hedin 1999; Simmons 2004; Simmons et al. 2008). Seed wings may be very diverse in origin, structure and distribution within angiosperms, and they have rarely attracted the attention of morphologists (Kapil et al. 1980). An early anatomical study on seed wings was done by Wahl (1897, cited after von Guttenberg 1971), later followed by von Guttenberg (1971) who distinguished six types of wings depending on their number, position on the seed and their functioning. Later, Boesewinkel and Bouman (1984) quoted von Guttenberg’s work and defined four types of seed flyers. Detailed studies are needed addressing the ontogeny of seed wings, in order to characterize them as homologous or analogous structures, respectively, when comparing such organs of different taxa on the one hand and relating them with other appendages of seeds that serve seed dispersal, on the other hand. Here, the monotypic genus Catha is selected to study the ontogeny of the ovule and the seed wing in Celastraceae for comparison with earlier studies on fleshy appendages.

Catha edulis (Vahl) Endl. is an evergreen shrub or small tree within the Celastraceae (Simmons 2004). Detailed studies of the anatomy and the morphogenesis of the macroscopic and microscopic features of Catha edulis leaves, flowers, fruits and seeds are given by Paris and Moyse (1958), Nordal (1980) and Revri (1983). The seed wing of Catha edulis has been described as an aril derived from the funiculus (Loesener 1942; Corner 1976; Simmons 2004). The winged seeds of other Celastraceae have also been interpreted as homologous to arils or arillodes (van der Pijl 1972; Corner 1976; Simmons and Hedin 1999). The processes of development from an ovule to a seed are essential for understanding the structures of the seed, especially if structures of the same origin are serving different functions. The goal of 10 the present study is therefore to check whether the seed wing corresponds to earlier descriptions.

1.9 Ovule development in Homalanthus populifolius (Euphorbiaceae) and differentiation of caruncula and aril

The structure and development of seed appendages presents a variety of different patterns and biological functions (Kapil et al. 1980). If these structures are used in an evolutionary framework, a clear delimitation and distinction of the different types is essential, especially for taxa in which the existing descriptions are incomplete or even contradictory. The genus Homalanthus is a good example. Corner (1976) documented the ovule and seed morphology of Homalanthus populneus (Geiseler) Pax and stated that the seed lacks a seed appendage, while Esser (1997) in his revision of the genus Homalanthus in Malaysia referred to the seed caruncula as an arillode. Later Gardner (1999) described the fruit and seed of Homalanthus populifolius Graham from Australia. He stated that the seeds have a papillose testa and a well-developed aril and noted that the conflicting observations of Corner and Esser remain to be reconciled. He pointed out that a detailed developmental study was needed. Tokuoka and Tobe (2002) reported the seed of Homalanthus as having a fleshy aril (or caruncula), making the distinction still more confusing.

For the understanding of the seed structure, an adequate knowledge of ovule development is essential (Boesewinkel and Bouman 1984), though often difficult to obtain from dried material (Gardner 1999). We will study the entire development from ovule to seed of Homalanthus populifolius Graham and compare it with the aril structure of Passiflora citrina J. M. MacDougal.

Homalanthus populifolius Graham, the Bleeding Heart or Queensland Poplar, is a shrub or small tree distributed in Northeast and East Australia, the Solomon Islands and New Guinea (Esser 1997). It grows very fast and can be used as a pioneer species in rainforest regeneration. In Australia it is used as a pioneer tree in 11 landscape rehabilitation projects (Floyd 2008). If more complex structures originating from the integument or the funiculus are present, it is very likely that the surface structures of the ripe seed are formed directly from the epidermal layer. Therefore, it is of some interest to see which properties of a normal epidermis are present in Homalanthus populifolius Graham and what modifications the epidermis underwent during ripening.

2 Material and methods

2.1 Material collection and preparation

All the material was collected at the Botanical Garden of the Ruhr-Universität Bochum (Germany); collections were done weekly from 2009 to 2013. The different developmental stages were collected from living plants and dissected in 70% ethyl alcohol. Selected structures were analyzed under the dissection microscope (ZEISS Stemi SV 11) and photographed with a KEYENCE VHX-500F digital microscope. After that, the material was fixed in FAA (formalin: acetic acid: ethyl alcohol 70% = 5: 5: 90) and kept in the fixative under moderate vacuum for at least 30 min. After 2 days, the FAA was replaced by 70% ethyl alcohol for further storage.

2.2 Scanning Electron Microscopy (SEM)

The structures of interest were transferred from ethyl alcohol to dimethoxymethane and stored at 4°C for at least 48 hours. Dimethoxymethane chemically dehydrates the plant tissue and serves as intermedium in the critical point drying process (Gerstberger and Leins 1978). Critical point drying was performed using a CPD 030 (BALZERS). Depending on size and structure of the material, the dried tissue was mounted on aluminum stubs either with conductive pads (Leit Tabs, PLANO) or conductive carbon cement (Leit-C, PLANO) and then stored in a desiccator with silica gel.

The samples were sputter-coated with gold for 200-400 s at 42-43 mA (BAL-TEC SCD 050). Scanning electron microscopy was performed with a DSM 950 (ZEISS). For documentation, a digital image processing system (DIPS 2.2, POINT ELECTRONIC) was used, which allows the storage in the tiff-format (2000 × 2000 pixels). The size was adapted to the plate format using Adobe Photoshop®; other image processing was not performed.

2.3 Anatomical studies

Anatomical studies were done using classical paraffin serial sectioning. The fixed material was transferred from FAA to 70% ethyl alcohol for at least 24 hours and then gradually transferred to tert-butyl alcohol and to Paraplast Plus® (McCORMICK SCIENTIFIC) according to Gerlach (1984, modified).

The embedded samples were mounted on wooden stubs and cut with a rotary microtome (LEICA RM 2065). Serial sections (5-12 µm) were mounted on microscope slides with protein glycerol according to Mayer and stored at 40°C for at least 12 h. After staining with Safranin/Astra blue according to Gerlach (1984, modified), slides were covered with Entellan®NEW (MERCK) and cover glasses. Photographs were made with a ZEISS Axioplan supplied with a Color View II camera (OLYMPUS). Plates were made using PHOTOSHOP 6CS. For multiple image alignment, the software Cell^F (OLYMPUS) was used. For the schematic drawing, the software CorelDRAW® X4 was used.

3 Results

3.1 The morphology of Zamiaceae

Zamia amblyphyllidia D. W. Stev. (Fig. 1) is a representative of the Cycadales, an evolutionarily ancient gymnosperm order. They are generally known as Cycads having the original features of fossil gymnosperms, like all other representatives of the order. First, the megasporophyll is a leaf-like structure (Fig. 1). Second, the sperm is a spermatozoid, which in addition to Cycadales is only found in Ginkgo biloba L. Third, like many Cycadales, they strongly resemble tree ferns in appearance because of their leaf Fig.1: Zamia amblyphyllidia D. W. Stev. crown with large fronds (Fig. 1, Fig. 9). in the Botanical Garden Bochum Z. amblyphyllidia is pachycaul due to the branching directly above the ground. Distally at each end of the branched trunk the leaves form a crown with a maximum of 15 leaves that can reach a length of 150 cm and up to 40 pairs of uniformly distributed leaflets. The individual leaflets are narrow-lanceolate and slightly serrate in the distal quarter. Samples of Z. amblyphyllidia were collected at the Botanical Garden Bochum. The plants are kept in pots in a temperate greenhouse (Fig. 1). There are several female and male individuals of Z. amblyphyllidia. In the vegetative stage both sexes cannot be distinguished, unlike Cycas the strobilus is grouped in 2-5 and is not proliferating. The slender strobilus of male plants (Fig. 2) is cylindrical in shape and can grow up to 20 cm long. For pollination the red-brown sporophylls diverge. After pollen shed the male strobilus dies.

The much more compact, cylindrical female strobilus (Fig. 3) is about 15 cm long and reaches a diameter of 4-6 cm. The megasporophylls form a compact strobilus and have a well-defined stalk which expands into a peltate head. The cone has a slightly ridged apex (Fig. 3, arrows). At pollination the apex is dark brown (Fig. 3, green arrow) and later rather grayish (Fig. 3, red arrow). The individual sporophyll diverges for the purpose of pollination just below the cone apex and

Fig. 2: Male strobilus of Zamia near the cone basis. amblyphyllidia D. W. Stev.

Fig. 3: Female strobilus of Zamia amblyphyllidia D. W. Stev., younger stage and mature stage.

After pollination, the female strobilus closes again and remains stays compact during the entire maturation. The cone scales are forced apart only by the growing seeds. 16

As in the Botanical Garden Bochum several female specimens are available, cones were collected depending on the desired size from different plants similar in size and age. To obtain a complete developmental sequence, fixed material was used collected from 2009 to 2013.

Normally two ovules are borne on each peltate head which is elongated transversally, and an ovule is borne on either side in the transversal plane (Fig. 4). Occasionally a third ovule may occur. The two ovules are situated one on either side of the short stalk, and their micropyles point radially towards the axis of the strobilus. The ovules are clearly not marginal, but each is situated slightly within the margin of a small area apparently corresponding to one of the areas of the lower surface of a female sporophyll. It is impossible to obtain a strobilus which is young enough to show the first stages of the development of the ovule without destroying the plant.

Fig. 4: Microsporophyll of Zamia amblyphyllidia D. W. Stev.

3.2 Ovule development and seed anatomy in Zamia

The longitudinal growth of the ovule takes place predominantly in the distal region of the ovule, where the integument is free. From this growth results that the nucellus is fused only to about half its length to the integument before the development of the mega-prothallium begins. Reinforced by a previous complete extension of the integument growth, the integument now encloses the nucellus (Fig. 5A, B). There is 17 a protruding beak on top of the nucellus pointing to the micropyle. It is called nucellar beak. This nucellar beak grows into the micropylar channel and step by step fills it completely (Fig. 5, A, C, E). In parallel the micropylar channel gets closer and closer and the opening finally appears irregularly lobed in top view of the ovule. At this stage, the further differentiation within the nucellus goes on with a size increase of the integument macrospore (Fig. 5E, F).

The integument forms a micropylar tip (Fig. 6A-C). The nucellar beak with longitudinal cells reaches from the tip to the macroprothallium (Fig. 6D, F). The protruding nucellar beak first reaches the middle of the micropylar channel and fills it completely without leaving any space between nucellus and micropylar tissue (Fig. 6E). By further growth the nucellar beak extends to the distal end of the micropylar channel and can be seen from outside in top view of the ovule (Fig. 7B).

At about this stage the lysis of the tissue of the nucellar beak begins starting from its distal end and proceeding gradually down to the macroprothallium. At early stages the nucellar beak seems to be still closed (Fig. 7A). An exposed pollination drop has never been observed despite seeking carefully for such a stage. Only when the lysis has reached the main body of the nucellus, a widening towards the pollination chamber can be seen (Fig. 7D). At this stage the peripheral wall of the nucellar beak is still present forming a kind of chimney that can be detected as a dark blue line in longitudinal sections. It is unclear whether the beak tip is still closed or already open at this stage. In longitudinal sections the distal end of the beak appears blue due to astra blue staining and no clear cell margins are visible. Slightly later the distal end shows a clear opening in SEM samples (Fig. 7C, E).

After pollination, the strobilus is closed again. The pollen inside the nucellus forms the pollen tube and passes the tissues of the chamber between the nucellar beak and macroprothallium until mid-February in the nucellus (Fig. 8A). At this moment, there are several microprothallium developed in the pollen chamber in different stages (Fig. 7A, red arrows). There are two archegonia within the section, each forming neck cells (Fig. 8B). We found a special structure on the top of one archegonium (Fig. 8C, D).

Fig. 5: Ovule development of Zamia amblyphyllidia D. W. Stev. I. A,C,E – Longitudinal sections of the ovule from young to older stage; B,D,F – same stages of A,C,E with SEM results; B – longitudinal section of a megasporophyll with two young ovules; F – the integument is partly removed and the nucellus and nucellus beak can be seen. I = integument, M = micropyle, N = nucellus, NB = nucellar beak. 19

Fig. 6: Ovule development of Zamia amblyphyllidia D. W. Stev. II. A,C,E – Longitudinal sections of the ovule from young to older stage before pollination; B,D,F – same stages of A,C,E with SEM results; B,D – longitudinal section to show the position of the nucellus, nucellar beak and the integument; F – the integument is removed to show the morphology of the nucellar beak only. I = integument, M = micropyle, N = nucellus, NB = nucellar beak. 20

Fig. 7: Ovule development of Zamia amblyphyllidia D. W. Stev. III. A – longitudinal section of the ovule with the nucellar beak reaching to the micropyle; B – overview of the ovule from the micropyle to show the nucellar beak reach to the micropyle; C – the integument is removed to show the nucellus and nucellar beak with a hole on the top of nucellar beak; E – detail of the nucellar beak in C with a hole on the top of disintegrating cells; D – longitudinal section of the later stage of A with some pollen in the micropylar channel. I = integument, M = micropyle, N = nucellus, NB = nucellar beak. 21

Fig. 8: Seed anatomy of Zamia amblyphyllidia D. W. Stev. A – the upper part of a mature seed; B – detail of the egg part of A; C-D – detailed structure of the archegonium. Red arrows show microprothallium within the nucellus. E = egg, N = nucellus, Sa= sarcotesta, Sc = sclerotesta. 22

3.3 Ovule morphology and anatomy in Cycas revoluta

Cycadaceae, described by Carolus Linnaeus in 1753, comprise the only genus Cycas L. which probably consists of about forty species distributed in South-East Asia, southern China, Malaysia, tropical Australia and various islands of the western Pacific, with gaps in Africa and Madagascar (Jones 1993). Normally, Cycas revoluta Thunb. forms an Fig. 9A: Cycas revoluta Thunb., unbranched monopodial trunk which reaches up to young leaves (following female 2.5 m and 30 cm in diameter. The trunk sporophylls of the previous year). occasionally may produce offsets produced on the trunk or from the base. Those offsets do not develop to mature plants. Young leaves are bright green, bearing grey hairs (Fig. 9A-B). The growing point of the female plant grows through the developing crown of megasporophylls (Fig. 9C). New leaves are erect with circinate leaflets, with emerging hairs getting lost with age (Fig. 9B). The megasporophylls form a loose and open cone on the top of the trunk (Fig. 9C). It is brown and hairy. The megasporophyll (Fig. 10) of Cycas L. is a Fig. 9B: Cycas revoluta Thunb.,young distinct foliar organ, more or less divided into lobes, erect leaves. bearing the ovules on its margins. There are more than two ovules on each megasporophyll. The ovule (Fig. 10) is bilaterally symmetrical with a length of about 1.5 cm and a diameter of 0.8 cm at anthesis. Near the micropyle it is covered with hairs, whereas young ovules are wholly clothed by a dense indumentum except for the micropyle. The Fig. 9C: Cycas revoluta Thunb., ripe ovules are elliptical and can reach a length of female cone. about 3 cm and a diameter of 2.7 cm. The epidermal cells of the ripe ovules are less radically extended than in other Cycads and develop a strong orange-colored 23 epidermis. The integument clearly shows three different layers (Fig. 11), an outer fleshy layer, a middle highly sclerified layer, and an inner fleshy layer.

The outer fleshy layer (Fig. 11A-C) is called sarcotesta and is formed by large, thin- walled undifferentiated cells, filled up with starch. There are many mucilage channels and tannin cells within it. The epidermis is composed of radially elongated cells with thick outer and radial walls, which is very characteristic for the whole genus.

The middle layer (Fig. 11A-C) is called sclerotesta which is thickest at the top and near the chalazal end. In the chalazal region it slightly protrudes, accompanying the vascular bundle entering the seed. The cells are thick-walled and two layers can be distinguished, an inner one with cells elongated in a longitudinal direction and an outer one with cells directed in a more horizontal plane.

Fig. 10: Megasporophyll and ovule of Cycas revolutaThunb. A – megasporophyll; B – one ovule of the megasporophyll with protruding receptive surface.

The inner fleshy layer (Fig. 11A-C) Is called endotesta which is well developed, strongest at the base and composed of parenchymatous cells, great-cellular, thin- 24 walled, delicate tissue, soon collapsing in the ripe seeds. As in the sarcotesta it contains many tannin cells.

At the micropyle region (Fig. 11A) the sarcotesta forms two thick cushions with a groove between them, in which the micropyle stands as a little prominent tube. Eichler (1889) says “speaking about the micropyle of Cycas L, the seeds I have seen had a micropyle that was always round without any indication of slits. But it is quite possible that other species will show those lobes more distinctly”.

At the free tip of the nucellus is a beak (Fig. 11B), in which the pollen-chamber is located. This beak has an epidermis of elongated cells, whose outer walls are slightly cuticle-covered. The nucellus tip is solid, and three regions can be distinguished in the beak. The lower part of the beak is formed by cells which are slightly elongated in the horizontal direction. The cells of the upper part have thick, dark, rich granular content, and in the middle there is a strand of cells, which are more or less vertically orientated, and in which the formation of the pollen chamber begins. The protruding beak of the nucellus reaches at least into the middle of the micropylar tube. In the centre of the protruding beak a longitudinal channel is formed from the tip down to the middle of the nucellar beak. The cells of the beak start to disintegrate when the female cone opens and then form a protruding receptive surface.

The formation of the pollination chamber starts with funnel-like extension of the beak channel leaving a coniform central column somewhat similar to the “tent pole” described in Ginkgo L. (Fig. 11C, cc).

The vascular system consists of three bundles entering the base of the ovule. The median one runs straight on to the endotesta and divides there into several branches. The two lateral bundles enter the sarcotesta, at the base of which they let off one bundle each, breaking through the sclerotesta and running in the endotesta to the micropyle. The main bundles continue their way through the sarcotesta to the top.

Fig. 11: Anatomy of the ovule of Cycas revoluta Thunb.. A-C three longitudinal sections of the same – ovule; A –showing the micropyle; B –showing the nucellar beak; C – showing the central column. CC = central column, I = integument, M = micropyle, N = nucellus, NB = nucellar beak. 26

3.4 Ovule development, seed morphology and anatomy in Magnolia stellata (Siebold & Zucc.) Maxim.

Fig. 12: Magnolia stellata (Siebold & Zucc.) Maxim. in the Botanical Garden Bochum

Samples of Magnolia stellata (Siebold & Zucc.) Maxim. were collected from a single tree about 3-4 m high in the Botanical Garden Bochum (Fig. 12). The flower buds (Fig. 13A), flowers (Fig. 13B), young fruit (Fig. 13C), young seeds (Fig. 13D) and seeds (Fig. 13E-F) were collected from October 2009 to May 2012 continuously once a week during the growing season. The development of the ovule begins at the end of the summer. The entire flower is hidden at this time in an approximately 5 mm long terminal flower bud (Fig. 13A). It opens in the following spring (Fig. 13B).

Fig. 13: Flower buds, flower, gynecium, young fruit, mature fruit and seeds of Magnolia stellata (Siebold & Zucc.) Maxim.

In early September, each carpel has two ovule primordia on the marginal placenta in the basal portion of the carpel which is adnate to the floral receptacle (Fig. 14A). The ovule primordium is conical and dorsiventrally compressed and located in an undifferentiated state along the placenta. Until the end of September the carpels and ovule primordia do not change significantly neither in size nor in the developmental stage. Nevertheless some of the only slightly more developed carpels on the basal part of the floral axis contain ovule primordia with a beginning differentiation.

Comparing the well-developed ovules of mid-December to those of the beginning of February in the following year, the ovules show that Magnolia stellata (Siebold & Zucc.) Maxim. have winter dormancy. At the beginning of March, the differentiation of the inner integument and outer integument of an ovule is easily to be seen (Fig. 14B). In a SEM photo of a two weeks older ovule it can be recognized that the inner integument forms a ring around the nucellus and the outer integument grows more or less around the inner integument and nucellus (Fig. 14C-D). The nucellus extends into the middle. Due to the development of the ring-shaped outer integument, the ovules have their width more than doubled over the course of four to six weeks (Fig°14E-F).

In mid-March, the inner integument is enlarged as an annular rim; the outer integument is semi-annular and slightly interrupted on the concave side by the 28 funiculus (Fig. 15A). Later, peri- and anticlinal divisions occur on the concave side, and an outgrowth is formed.

As the ovule is further incurved, the elongation of the outer integument surrounds the inner integument to an increasing extent. The inner integument becomes irregularly lobed at the distal edge (Fig. 15B).

At a later stage, when the ovule is more incurved, the enlarging outer integument overgrows the inner integument (Fig. 15C). The lateral micropylar lobes of the outer integument are decurrently to the lateral sides of the funiculus, and it becomes obvious that the outer integument is hood-shaped (Fig. 15D). The opening of the outer integument changes in shape from round to deltoid in the top view by a greater elongation of the lateral portions rather than of the middle part (Fig. 15E). When the ovule is further elongating and becoming anatropous, the micropyle is formed by the two, neighboring, distal lobes of the outer integument. At maturity, the exostome is an inverted 人-shaped, transverse slit with a middle notch which is derived from the deltoid opening at an earlier stage (Fig. 15F; Fig. 16A-B).

At the end of September, in the mature seed the outer integument is differentiated into two layers, an outer fleshy one well filled with oil vacuoles, and an inner stony layer (Fig. 16C-D). The inner integument forms only a thin layer in the ripe seed (Fig. 16E-F).

Fig. 14: Ovule development in Magnolia stellataSuc (Siebold & Zucc.) Maxim. A – two ovule primordia per locule; B – young ovule with inner integument and outer integument initiating; C-F the development of inner and outer integument before the covering of the nucellus and ovule incurvation. C = carpel, F = funiculus, II = inner integument, N = nucellus, O = ovule, OI= outer integument, Pl = placenta. 30

Fig. 15: Successive processes of the micropyle development of Magnolia stellata (Siebold & Zucc.) Maxim. A – the outer integument half covering the inner integument; B – the outer integument covering the nucellus, inner integument with some lobes on tip; C – the hood-shaped outer integument; D-F – the 人-shaped micropyle forming. F = funiculus, II = inner integument, M = micropyle, N = nucellus, O = ovule, OI= outer integument. 31

Fig. 16: Seed morphology and anatomy of Magnolia stellata (Siebold & Zucc.) Maxim. A – the mature ovule; B – the micropyle of A; C – longitudinal section of the seed with an embryo; D – hand cutting of the seed in a slightly later stage than C; E – detail of the micropyle portion of C to show the inner integument tissue left on the micropyle and the outer integument developed into sarcotesta and sclerotesta; F – the chalazal portion of C showing the vascular bundle and hypostase. Em = embryo, En = endosperm, H = hypostase, F = funiculus, II = inner integument, M = micropyle, N = nucellus, O = ovule, OI= outer integument, Sa= sarcotesta, Sc = sclerotesta, VB = vascular bundle. 32

3.5 Aril development in Celastraceae

3.5.1 Celastrus orbiculatus Thunb.

Celastrus orbiculatus Thunb. has a pentamerous imbricate perianth. The number of stamens equals the number of petals (Fig. 17A). The gynoecium is trimerous (Fig. 17B) or sometimes tetramerous and starts its formation when the floral bud begins to close. The formation of the two ovules per locule starts prior to the complete closure of the carpel (Fig. 17C). The inner integument is initiated when the incurvation of the ovule starts (Fig. 17D). When the outer integument is formed the nucellus is positioned at right angle to the funiculus (Fig. 17E-F). During further development the ovule gradually turns into an anatropous position (Fig. 17F). When finally the ovule is completely anatropous, the ovule is slightly exostomatic (Fig. 18A). At this stage the exostome becomes thicker, and the thickening meristem at first encircles the micropyle more or less completely. In subsequent steps the micropylar thickening meristem of the outer integument enlarges, and the meristem step by step incorporates the entire funiculus until it finally encircles both the micropyle and the funiculus. The fleshy ring becomes thicker and thicker and develops a distinct rim towards the chalazal part of the ovule. In subsequent steps this ring forms a fleshy duplicature covering the outer integument (Fig. 18B-E). In contrast to a typical aril, this duplicature does not cover the micropyle, but leaves it free laterally to the funiculus. The duplicature leaves the chalazal region entirely free (Fig. 18F).

3.5.2 Euonymus europaeus L.

Euonymus europaeus L. has whorls of 4 perianth segments (Fig. 19A). The number of stamens equals the number of petals and carpels (Fig. 19A-B). The gynoecium is tetramerous (Fig. 19B) and starts its formation when the floral bud begins to close. The formation of the two ovules per locule starts prior to the complete closure of the carpel. The formation of the ovule starts with the differentiation of an obtuse narrow primordium (Fig. 19C). When the incurvation of the ovule starts (Fig. 19D), the inner integument is initiated. When the outer integument is formed, the nucellus is curved approximately 90 degrees and is turned towards the carpel margin (Fig. 19E). The ovule curves into a fully anatropous position when the outer integument finally covers 33 the inner integument and the nucellus (Fig. 19E-F; Fig. 20A). At this stage the micropyle points towards the proximal part of the gynoecium, the ovule is upright anatropous. When the ovule is finally completely anatropous, it is slightly exostomatic (Fig. 20B). At this stage the exostome becomes thicker, and the thickening meristem at first encircles the micropyle more or less completely (Fig. 20C). In subsequent steps the micropylar thickening meristem of the outer integument enlarges, and the meristem step by step incorporates the entire funiculus until it finally encircles both the micropyle and the funiculus (Fig. 20B-D). The fleshy ring becomes thicker and thicker and develops a distinct rim towards the chalazal part of the ovule. In subsequent steps this ring forms a fleshy duplicature covering the outer integument (Fig. 20E-F).

3.5.3 Euonymus planipes Koehne

Euonymus planipes Koehne has a pentamerous imbricate perianth (Fig. 21A). The number of stamens equals the number of petals (Fig. 21A-B). The gynoecium may be pentamerous or sometimes tetramerous, and starts its formation when the floral bud begins to close (Fig. 21B). The formation of the two ovules per locule starts prior to the complete closure of the carpel. Unlike in Euonymus europaeus L. and Celastrus orbiculatus Thunb., the anatropous ovules are pendulous from the top of the locule to the cup-shaped base of the gynoecium. When the incurvation of the ovule starts the inner integument is initiated. When the outer integument is formed the nucellus is positioned at right angle to the funiculus (Fig. 21C-D). During further development, the ovule gradually turns into its final anatropous position (Fig. 21E). When finally the ovule is completely anatropous, the ovule is slightly exostomatic; the micropyle nearly touches the funiculus (Fig. 21E). At this stage the exostome becomes thicker, and the thickening meristem at first encircles the micropyle more or less completely (Fig. 21F). In subsequent steps the micropylar thickening meristem of the outer integument enlarges, and the meristem step by step incorporates the entire funiculus until it finally encircles both the micropyle and the funiculus (Fig. 22A-B). The fleshy ring becomes thicker and thicker and first develops a distinct circular rim towards the chalazal part of the ovule (Fig. 22C). Further growth of the duplicature is asymmetric and leads to an oblique shape of the duplicature (Fig. 22D). 34

In subsequent steps this ring forms a fleshy duplicature covering the outer integument (Fig. 22E-F).

Fig. 17: Celastrus orbiculatus Thunb.: Ovule, successive developmental stages. A – young flower with perianth and stamen primordia; B – trimerous gynoecium initiating; C – initiation of 2 ovules per locule; D – beginning development of the inner integument; E, F – formation of the outer integument and beginning ovule incurvation. c = calyx, f = funiculus, ii = inner integument, n = nucellus, o = ovule, oi = outer integument, p = petal, s = stamen. 36

Fig. 18: Celastrus orbiculatus Thunb.: Caruncula, successive developmental stages. A – caruncula originating from the margin of the outer integument; B-D – enlargement of the caruncula and incorporation of the funiculus; E – beginning formation of the fleshy seed duplicature; F – mature seed. ca = caruncula, f = funiculus, ii = inner integument, m = micropyle, oi = outer integument.

Fig. 19: Euonymus europaeus L.: Ovule, successive developmental stages. A – preanthetic flower bud showing stamina and carpel primordia; B – young tetramerous gynoecium; C – ovule initiation; D – beginning development of the inner integument; E, F – formation of the outer integument and beginning ovule incurvation. c = carpel, f = funiculus, ii = inner integument, n = nucellus, o = ovule, oi = outer integument, s = stamen. 38

Fig. 20: Euonymus europaeus L.: Caruncula, successive developmental stages. A – micropyle formed by 2 integuments prior to caruncula formation; B – caruncula originating from the margin of the outer integument; C – incorporation of the funiculus in the formation of the caruncula; D – beginning formation of the fleshy duplicature; E – slightly older stage in seed development than in D (6 of 8, 2 seeds removed); F – mature seed. ca = caruncula, f = funiculus, ii = inner integument, m = micropyle, oi = outer integument. 39

Fig. 21: Euonymus planipes Koehne: Ovule, successive developmental stages. A – young flower with perianth and stamen primordia; B – pentamerous gynoecium initiating; C-E – formation of the outer integument and ovule incurvation; F – caruncula initiating from the margin of the outer integument. c = calyx, ca = caruncula, f = funiculus, ii = inner integument, n = nucellus, oi = outer integument, p = petal, s = stamen. 40

Fig. 22: Euonymus planipes Koehne: Caruncula, successive developmental stages. A-B – caruncula originating from the margin of the outer integument; C – older circular stage of the caruncula; D – asymmetric growth of the caruncula; E – mature stage of the seed, view from the hilum ; F – entire mature seed, lateral view. ca = caruncula, f = funiculus, m = micropyle. 41

3.6 Ontogeny of the ovule and seed wing in Catha edulis (Vahl) Endl. (Celastraceae)

The flower of Catha edulis has 5 sepals, petals and stamens, and the gynoecium is trimerous (Fig. 23A). The formation of the two ovules per locule starts prior to the complete closure of the carpels. The ovules arise from the basis of the carpel (Fig. 23B). The inner integument is initiated when the incurvation of the ovule starts (Fig. 23C). The outer integument is formed only slightly later peripheral or basal to the inner integument and is initiated as a semi-annular rim (Fig. 23C-D). The outer integument grows faster than the inner integument and because of the asymmetric growth of both during further development, the ovule gradually curves into an anatropous position. After that, the outer integument turns into a hood covering inner integument and nucellus (Fig. 23F). Later, the outer integument prolongates to the basis of the gynoecium leaving a crevice as a micropylar opening. On the funicular side, the micropyle is delimited by the funiculus itself and not by the outer integument. Otherwise the micropyle is exostomatic. While the micropyle at first is more or less transverse truncate (Fig. 24A), it becomes oblique truncate in subsequent steps (Fig. 24B). Further growth of the outer integument leads to a prolongation that positions the micropyle between the prolongation and the funiculus (Fig. 24B-D). The protrusion of the outer integument enlarges. This meristem forms an entirely fleshy rim around micropyle and funiculus (Fig. 24B-D).

The basal part of the gynoecium now undergoes an intercalary growth. The placenta region is thus shifted more to the distal end of the locule. This process generates the space that is filled in a simultaneous growing process by the micropylar part of the outer integument (Fig. 25A-B). At the same time, basal and peripheral parts of the outer integument form a kind of lobed duplicature growing in the opposite direction that covers the seed partly (Fig. 24E-F, arrows). During the seed ripening, the seed- coat becomes hard and dry. The fleshy tissue emerging from the micropylar part to the outer integument will be extruded between seed-coat and placenta and becomes dry and membranous. In this way it becomes a one-side wing attached to the seed (Fig. 24F). 42

Fig. 23: Catha edulis (Vahl) Endl.: Flower and ovule, successive developmental stages. A – Gynoecium with three carpels initiating; B – two ovules initiating in one carpel; C – inner integument starting to develop; D-F – outer integument originating and ovule incurvating. C = calyx, F = funiculus, II = inner integument, M = micropyle, N = nucellus, P = petal, S = stamen, O = ovule, OI = outer integument. Scale bars: A,B = 100 µm; C = 20 µm; D,E = 40 µm; F = 60 µm. 43

Fig. 24: Catha edulis (Vahl) Endl.: Caruncula, successive developmental stages. B Caruncula originating from the edge of the outer integument; C-D – Caruncula development and micropyle opening; E –caruncula and the aril part forming the funiculus and making a rim surrounding the seed; F – mature stage of the seed. Ca = caruncula, II = inner integument, OI = outer integument, M = micropyle, F = funiculus. Scale bars: A = 90 µm; B = 200µm; C,D = 400 µm; E,F = 2mm. 44

Fig. 25: Catha edulis (Vahl) Endl., younger (A) and mature (B) stage of the fruit. ca = caruncula, f = funiculus, p = pericarp, s = seed. © V. Elbrecht & X. Zhang.

3.7 Ovule development in Homalanthus populifolius Graham (Euphorbiaceae) and differentiation of caruncula and aril

3.7.1 Homalanthus populifolius Graham

The inflorescence of Homalanthus populifolius Graham is a thyrsus with the basal cymes reduced to the female primary flower. More distally, the cymes have a female primary flower and two male cincinnate continuations. The most distal cymes are exclusively male. The primordia of the pistillate flowers arise in the axil between the bract and the inflorescence axis. The primordial pistil is surrounded by two hypsophylls or bracteoles. Two staminate cymule primordia emerge at the base of the pedicel of the female flower (Fig. 26A) from the axils of the prophylls. At the top of the growing point, the symmetrical triangular-shaped structures become the carpel primordia on both sides (Fig. 26B). Since the dorsal base of the carpel grows faster than the ventral side and the tip, the two carpels grow closer and closer towards to each other (Fig. 26B-E). At the same time, in the centre of the floral apex, the placenta (Fig. 26B-D) and the primordial ovule are formed (Fig. 26D). However, the central cells of the symmetrical axis grow faster than the marginal ones. The ovule primordia grow outward. At this developmental stage, two emerging primordial ovules are evident. The two hypsophyll lobes are covering the unclosed half of the carpel. The two staminate cymules at the base of the female flower are visible (Fig. 26D). When the two carpels are close together (Fig. 26E), the two ovules grow upward to fill the two carpels. The carpels are not yet closed, but there is an individual space for each ovule. At this stage, there is just a gap between the two carpels (Fig. 26F, arrow).

The nucellus is ellipsoid (Fig. 27A) and forms a beak which extends beyond the inner integument (Fig. 27A). The outer integument rises later and basal next to the inner integument (Fig. 27B). By following the growth of the outer integument, the ovule curves slightly upward. The nucellus grows longitudinally, becomes tongue-like and passes beyond both integuments (Fig. 27C) until it finally even touches the carpel (Fig. 27D). After that the two integuments grow faster than the nucellus. Later they will cover the nucellus. The outer integument grows faster than the inner integument, and the ovule becomes exostomatic. In this way the outer integument forms a hood outside of the nucellus and the inner integument (Fig. 27D-F). Finally, the outer 46 integument surrounds the whole ovule until there is a crevice left which is the micropyle (Fig. 28A). Before the micropyle closes as a crevice, the ovule is slightly exostomatic. The exostome grows thicker and thicker, becoming a caruncula which surrounds the funiculus (Fig. 28B-C) before curving downwards (Fig. 28D). This structure will become fleshy and cover half the seed when the seed is ripe (Fig. 28D-F).

Stomata appear on the surface of the outer integument after the caruncula has been formed (Fig. 28D), they stay open during seed development (Fig 29A, C-F). The stomata are scattered all over the outer integument (Fig. 3D-F). Normally, they appear single (Fig. 29A), but in some cases they are in pairs (Fig. 29C-D). Occasionally, they are in side-by-side clusters (Fig. 29E). When the seed matures, the seed coat becomes dry and the stomata stay permanently open (Fig. 29F). The stomata are of the anomocytic type – the surrounding cells do not differ in shape and size from the rest of the epidermal cells. In this they are similar to the stomata of the leaf (Fig. 29B), and only occur on the abaxial side.

3.7.2 Passiflora citrina J.M. MacDougal

The caruncula of Homalanthus populifolius Graham will compared to the aril structure of Passiflora citrina J.M. MacDougal. The aril-primordium of Passiflora citrina initiates as a rim from the funiculus after fertilization (Fig. 30A-B). The micropyle of Passiflora citrina J.M. MacDougal is formed by the inner integument (Fig. 30B). After that the aril develops as a plate around the funiculus and covers the micropyle (Fig. 30C-D). Later, the aril grows more and more and forms a cup-like envelope around the seed that becomes fleshy at seed ripening (Fig. 30E-F).

Fig. 26: Homalanthus populifolius Graham: Carpel, successive developmental stages. A – Thyrsus unit with one young female flower and two staminate cymules primordia; B-C – dimerous gynoecium initiating; D – initiation of the placenta with ovule primordia and two staminate cymules; E – two carpels closer to each other; F – ovule initiating. b = bract, c = carpel, cp = carpel primordium, hy = hypsophyll, o = ovule, p = placenta, scp = staminate cymule primordium, sc = staminate cymule. Scale bars: 200 µm. 48

Fig. 27: Homalanthus populifolius Graham: Ovule, successive developmental stages. A – Initiation of the inner integument; B – beginning development of the outer integument; C-F – formation of the outer integument and ovule incurvation. c = carpel, f = funiculus, ii = inner integument, n = nucellus, oi = outer integument, p = placenta. Scale bars: 200 µm. 49

Fig. 28: Homalanthus populifolius Graham: Caruncula, successive developmental stages. A – Micro- pyle formation; B – caruncula originating from the margin of the outer integument; C – enlargement of the caruncula and incorporation of the funiculus; D – beginning formation of the fleshy seed duplicature; E-F – mature seed. ca = caruncula, f = funiculus, m = micropyle, oi = outer integument. Scale bars: A-D = 200 µm; E-F = 1mm. 50

Fig. 29: Homalanthus populifolius Graham: Stomata on the outer integument and abaxial side of the leaf. A – Stomata on the outer integument; B – normal stomata on the leaf; C-E – abnormal stomata on the outer integument; F – stomata on the dry seed coat. Scale bars: 10 µm. 51

Fig. 30: Passiflora citrina J. M. MacDougal: Aril, successive developmental stages. A-B – A rim aril primordium developing from the funiculus and pollen tube throwing the micropyle; C-E – developed aril covering the micropyle; D – aril, view from the bottom; F – developed aril torn apart of side showing the micropyle. a = aril, ap = aril primordia, f = funiculus, ii = inner integument, m = micropyle, pt = pollen tube. Scale bars: A, F = 400 µm; B = 100 µm; C = 800 µm; D = 200 µm; E = 1 mm. 52

4 Discussion

Reproduction is the most complex part of life, especially in seed plants. The reproduction process of seed plants starts from the ovule primordium. The structure of the ovule will develop to the different parts of the seed. To understand the evolution of different structures of a seed, the origination of the structure has to be known. The integument of seed plants is a key problem in this respect. The function of the integument is easy to comprehend as protection of the ovule. However, such Fig. 31: Seed envelopes in gymnosperms. © P. Knopf. protective tissues have evolved on different organizational levels and in different ways from the outermost tissues. It may be that the integument became protective just because it represents the peripheral tissue, but that its evolution was driven by completely different needs.

4.1 Integument and aril or aril-like structures in gymnosperms

In gymnosperms, there is only one integument covering the nucellus leaving a small opening at the apex called micropyle. This opening enables the pollen to get into the interior of the ovule for fertilization (Fig. 31). In gymnosperms, except for Cycads, the ovule is located on a short shoot. In different taxa, however, there are different conditions of integument development. In Cycads (Fig. 31) and Ginkgo L. (Fig. 32), there is only one integument covering the nucellus, but this integument differentiates into two layers. The outer layer becomes a fleshy sarcotesta, the inner layer becomes a hard and woody sclerotesta. However, in Cycads the ovules are on the margin of the megasporophyll and in Ginkgo L. the ovules are positioned on a stalk in the leaf axils. There is a circular structure outside the only integument called collar in Ginkgo L. One can find a similar structure in Cycas L., but not in Zamia L. This structure doesn’t grow around the seed and abscises with the seed as such, and it is 53 a special structure of Cycadaceae and Ginkgoaceae. The ovules have one integument which becomes hard and woody to protect the embryo. However, in families like Taxaceae (Fig. 31), and Podocarpaceae (Fig. 34) the ovule is on the axil between the bracts. There is one fleshy structures arising from the outside of integument. In Taxus and Cephalotaxus the homology of aril and sarcotesta has been shown (Stützel and Röwekamp 1999). In Podocarpaceae (Fig. 34) this structure is called epimatium. In Gnetales, all the families have a second structure outside the integument called chlamys. If fleshy, the chlamys may be called aril. In Welwitschiaceae this structure becomes membranous and forms lateral wings. As pointed out by Endress (1996), the outer envelope (chlamys) in Gnetales develops prior to the inner one, the latter one being homologous to the integuments of the other gymnosperms. In contrast, the aril in Taxaceae as well as the epimatium in Podocarpaceae develops after the integument (Tomlison and Takaso 1989; Rieger 2002).

Fig. 32 Ginkgo biloba L.: Ovule. Fig. 33: Pinus parviflora Siebold Fig. 34: Podocarpus macro- & Zucc.: Ovule. phyllus (Thunb.) Sweet var. macrophyllus: Ovule. 4.2 Evolution of the (only) integument in gymnosperms

The only integument of the gymnosperms appears together with the ovule in land plants. There are obviously no fossils showing an ovule without integument. This hinders an understanding of its evolution. The only integument is homologous with the first integument of angiosperms. However, this general assumption should be reevaluated carefully as discussed below.

There are four factors that can endanger the growing process from the apical meristem to a seed. These are fungi, animals, low temperature and dry weather conditions. The nucellus must be isolated against fungal infection e.g. by an integument. Taken up by animals a seed can be destroyed by chewing and by the digestion process. Nevertheless, a sclerotesta is the best way to solve this problem 54 because it is hard and woody. The third factor is unfavorable weather conditions. Low temperature may destroy the cell structure, draught may cause irreversible damage due to siccation of cells.

Another function of the only integument is linked to fertilization, here: to carry the pollination drop. The function of the pollination drop is to catch the pollen from the air and to transport the pollen inside the ovule and attach them to the nucellus.

The seed function is to protect and disperse the embryo. Dispersal may be autonomous; the fruit of the Balsaminaceae, for example, can shoot off the seeds. In gymnosperms, however, the seeds could transported by environmental means. There are one or two wings on the seed which can carry the Fig. 35: Structures of angiosperm ovules. © P. Knopf. seed some distance away from the mother plant. The seeds can also be dispersed by animals, especially birds. Since birds are not sensitive to smell, the color of the seed plays an important role for being recognized. Another thing which can attract birds is the taste. So the fleshy structure stimulates birds to feed on the seed. However, this kind of the seed has a sclerotesta, and as a result it will not be destroyed by the digestion system of the animals.

4.3 The structure of the ovule in angiosperms

The structure of the ovule (Fig. 35) in angiosperms is more complex compared to the gymnosperms because in gymnosperms the ovule is normally ‘naked’. In angiosperms, the ovule is covered by the carpel and has two integuments which arise from the placenta that is normally located on the margin of the carpel. The funiculus connects ovule and placenta. A vascular bundle is running through it and supplies the ovule with nutrients as it develops into a seed. The nucellus Fig. 36: Liriodendron chinense (Hemsl.) Sarg.: Two develops on the top of the funiculus and integuments. is generally covered by two integuments arising from the base of the nucellus. The integuments leave a small opening called micropyle at the apex of the ovule which allows the pollen tube to grow into the ovule for fertilization. Located opposite to the micropyle is the chalaza where the nucellus, funiculus and integuments merge (Eames 1961).

4.4 The envelope of the ovule within angiosperms

The envelope structure (Fig. 35) of ovules within angiosperms is more variable and may differentiate into different layers. The envelopes can be formed by integuments with one to about four layers. If the ovule has only one integument it is termed unitegmic ovule (Fig. 37) while that one with two integuments is known as bitegmic ovule (Fig. 36). In addition, a third envelope arising from the funiculus is called aril (Fig. 35). However, there is a fleshy structure arising from the exostome called caruncula which looks like an aril in Celastraceae (Fig. 38). Corner (1953) reported the post-fertilization formation of a fourth integument in some members of the Annonaceae. 56

The bitegmic ovule is considered as the original type in angiosperms because most of the basal taxa have two integuments (Endress and Igersheim 2000b). Davis (1966) counted the unitegmic and bitegmic families: Among the 319 families of which data were available, the bitegmic condition characterizes 208 families, of which 155 are dicotyledonous, as well as the 90 families with exclusively unitegmic ovules. From the bitegmic ovule to the unitegmic ovule there is a Fig. 37: Actinidia arguta (Siebold & Zucc.) Planch. ex Miq.: Unitegmic ovule. reduction. The process happened several times on different taxonomic levels. (Stebbins 1974; Bouman 1984; McAbee et al. 2005). McAbee et al. (2005) reinforce that this reduction happened early in the history of the Asterids.

The angiosperms have originally two integuments, and in different taxa they develop into different structures when they become the seed coat (Fig. 35). In some families (like Magnoliaceae) the outer integument differentiates into two layers, the outer layer becoming a red and fleshy structure called sarcotesta and the inner layer becoming a hard and woody structure called sclerotesta (Fig. 35, Magnolia). In most unitegmic taxa like Juncus L., two integuments fuse and become dry and hard forming a sclerified seed coat(Fig. 36), while some taxa have a third envelope which arises from the funiculus, becomes a fleshy structure and covers the seed coat. This third envelope is termed aril and occurs e.g. in Passiflora L. and Nymphaea L. (Fig. 36). In this case the two integuments inside of the aril become dry and hard. Fig. 38: Typical aril and caruncula in Celastraceae. © P. Knopf.

4.5 The function of the second envelope of seed plants

The seed coat is intended as a stable and closed container to allow the embryo to survive rather long phases of dormancy. The seed development results in two different sets of problems, each of them giving rise to new evolutionary constraints. First the ovule is not completely enclosed by integument. There is the micropyle as an open hole. Different means have evolved to close this functional gap. The most simple way is by mere shrinking. However, this works properly only under dry conditions. Another mode to close them is by cell divisions inside (e.g. Taxus L.) or on top of the micropyle. Second, in the same way protrusions from neighboring cells can close vessels passing the hilum. This may explain why fleshy seed appendages originate generally either from the micropyle or from the funiculus. The more sophisticated opening mechanisms are needed to allow germination, the more perfectly the seed coat encloses the embryo. The latter aspect will not be touched here, but left to further studies.

The second (outer) integument develops always later than the first (inner) integument in angiosperms. It is problematic to regard the second integument as a protective structure, too. If more protection is needed, the layer will evolve thicker. Therefore, the evolutionary relevance of the outer integument was most likely different from that of the inner integument. Initial stages of the second integument would not be protective, and a stepwise evolution of this feature seems to be problematic if only the final structure is functional. It is more likely that the outer integument is equivalent to the aril of Taxaceae and the epimatium of Podocarpaceae in gymnosperms. This aril also develops later than the first (inner) integument. Some rather basal angiosperms like some Magnoliaceae show a sarcoesta and sclerotesta formed by the outer integument only.

The function of the integument(s) has been seen basically in mature seeds. At that time the function is clearly protection. In early stages it is however involved in holding and exposing the pollination drop in gymnosperms. This droplet is still present as a micropylar exudate in many angiosperms (dicots as well as monocots). Therefore, it may be that the primary function is related to fecundation. Protection might have been taken over secondarily as protection is best achieved by the outermost layers. 58

Micropyles may be open, forming an open channel, or closed. In the former the micropyle may be sealed by secretion (e.g. Annona, Igersheim and Endress 1997; Ornithogalum, Tilton1981, Tilton and Lersten 1981; Beta, Olesen and Bruun 1990). In the latter, a closed pollen tube transmitting tract is formed by post-genital fusion (Helianthus, Yan et al. 1991). This diversity is thus analogous to that of the carpels sealed by secretion or by post-genital fusion (Endress and Igersheim 2000a). However, details of the histology of mature micropyles have only rarely been studied (Endress 2011).

One of the best known and widespread fleshy seed appendages is the aril. An aril is defined as any specialized outgrowth from the funiculus that covers or is attached to the seed. Arils occur in gymnosperms as well as in angiosperms. Typical gymnosperm arils start to develop markedly later than the only integument. The developmental sequence of integument and aril in gymnosperms is therefore the same as the developmental sequence of inner and outer integument in angiosperms. Usually both integuments in angiosperms are regarded as protective structures. However, it seems much easier in evolution to make one integument stronger than to add a second one starting its development from the ovular base. Therefore it makes sense to assume that the functional meaning of the outer integument in angiosperms was originally not protection of the seeds. In need of a fleshy seed coat angiosperms would have to regain such a structure according to Dollo’s principle in a different way or as different structure. As arils are much more frequent in angiosperms than in gymnosperms, we tried to learn about modes of formation and evolutionary pathways towards intercalated seed envelopes from studies in angiosperms. One of the puzzling questions is why the epimatium in Podocarpaceae follows the recurvature of the seeds similar to the second integument in anatropous angiosperm seeds while the aril in angiosperms obviously does not, so that the opening of the arils is opposite to the position of micropyle in anatropous angiosperm seeds. 59

4.6 Origin of ovule and integument

The widely accepted theory regards the integument as the result of the fusion of sheathing telomes (Fig. 39). For this concept usually Andrews (1961) is cited, his ideas are, however, probably based on earlier precursors. In the 19th century ovules were termed “seed buds” in accordance with this, the integument is regarded as a protective structure. However, there are two problems.

1: At the time of pollination there is no nutritive tissue to be protected yet; the function of the micropyle is needed.

2: The development of the integument in gymnosperms does not coincide with the theory in angiosperms. Fig. 39: The telome hypothesis after Andrews (1961), from Stewart (1983).

4.7 Ovules in Lyginopteridatae

Fig. 40: Structures of the ovules in Lyginopteridatae (after Camp & Hubbard (left), modified by Stützel (right)). 60

The ovule structure of Lyginopteridatae (Fig. 40) is a model for the origin of ovules. However, how do palaeobotanists distinguish between the two possible interpretations between the ovule of extinct and extant plants? There is a gap in understanding the evolution of the seed plants, especially the evolution of the ovules in seed plants

Here based on different taxa of basic gymnosperms (Cycas L.and Zamia L.) and angiosperms, Magnolia L. and Celastraceae, it is tried to get some clues of the development and function of the envelopes of the ovule and to interpret the evolution of the ovule in seed plants. Details will be given in the follow discussion.

4.8 The pollination drop

The pollination drop exists in most gymnosperms, except for a few members of the Pinaceae (Abies Mill., Cedrus Trew, Larix Mill., Tsuga (Endl.) Carriére, Pseudo- tsuga Carriére), which show a stigmatic micropyle, and, moreover, some genera (Araucaria Juss., Agathis Salisb., Tsuga (Endl.) Carriére) in which the pollen grains do not land on the micropyle at the time of pollination (Singh 1978).

The presence of the pollen drop Fig. 41: Pollination drops of Zamia amblyphyllidia D.W. has been mentioned even in fossil Stev.and Cupressus arizonica Greene var. glabra (Sudw.) Little. A.- the pollination drop of Z. amblyphyllidia; B. - the gymnosperms (Andrews 1963). longitude section of Zamia ovule in A; C. - the pollination The fluid serves as a receptor of drops of C. arizonica; D. - the longitude section of the ovule of C. arizonica. the wind-borne pollen, and also as a vehicle for transporting it to the nucellus. It can also be found that the droplet is a very suitable medium for pollen germination (Anhaeusser 1953; Ziegler 1959). There are two kinds of pollination drops, depending on their formation (Fig. 41A-D). Chamberlain (1935) and Dogra (1964) and our own results (Fig. 41A-B) show for 61

Cycads and van der Pijl (1953) for Gnetum L. that the pollination drop is formed by a collapse of the cells of the nucellar tip. Ziegler (1959) concluded that the osmotically active substances released by the degenerating cells absorb water from the nucellus or the atmosphere, fill the micropylar channel and are seen as a droplet at the micropyle. The other kind of pollination drop in those plants (e.g. Cupressus L.) in which the nucellus cells do not degenerate before pollination. Doyle and O’Leary (1935) and McWilliam (1958) working on Pinus L., and Ziegler (1959) working on Taxus L., stated that the pollination drop is secreted by the cells forming the apex of the nucellus. This is similar to the results found here for Cupressus L. (Fig. 41C-D). The different composition and the connection between those two different types of pollination drops are not clear until now.

4.9 Pollination in Cycads

Chamberlain (1919) considered that Cycas L. is a wind-pollinated taxon. However, there are also reports that Encephalartos Lehm. and Macrozamia Miq. are insect- pollinated. Oberprieler (1993) recorded the life cycle of 63 weevil species (belonging to 12 genera) associated with Cycads, and even regarded that ants can be the pollinators of Cycas L.

There are two steps in the pollination of Cycads. First, the pollen needs to be transported from the micro- to the macrosporophylls either by wind or by insects. Then there is also the need of a medium to transport the pollen from the macro- sporopylls to the pollination drop. Norstog and Stevenson (1980), after his simulated experiment of the wind pollination in Dioon Lindl., Zamia L. and Cycas L. considered that insects probably are the pollinators of Dioon Lindl. and Zamia L., and that Cycas needs the fluid to transport the pollen from the megasporophylls to the ovule.

According to our results, before pollination the cells of nucellar beak of Zamia L. will start to disintegrate (Fig. 7A) which leads to a protruding receptive surface on the micropyle. This disintegrating proceeds from the top to the base of the nucellar beak, there will remain just a circular wall of cells around the micropylar channel (Fig. 7A, C, E). At this moment, if any pollen gets from the opening into the strobilus, it will stick to this protruding receptive surface, and the nucellus will absorb the liquid into the micropylar channel (Fig. 7C-D). Later the pollen will be in the pollen chamber. 62

At this time the multinucleate macroprothallium develops into a multicellular tissue (Fig. 7D).

On the other hand, artificial pollination under greenhouse conditions is performed in Zamia L. by using water with dispersed pollen that is rinsed through the strobilus. Freely exposed pollination drops would be washed away applying this method. Thus, the pollination drop of Cycads is formed differently from the pollination drop of conifers which is secreted by the cells forming the apex of the nucellus. The pollination drop of Cycads is formed by a collapse of the cells of the nucellar tip. This process is the same as mentioned by Andrews(1963) in fossil gymnosperms.

4.10 Aril development in Celastraceae

The term aril is historically used in a rather broad sense. This may lead to an ambiguous understanding of structures described as arils. In diagnostics this might be acceptable as this character is generally used together with others. However, if the focus is on character evolution, processes starting in a different way and leading to different final stages should be distinguished carefully. Gaertner (1788) was the first to describe the aril as an accessory integument. Botanists commonly use the term aril for fleshy structures arising from the funiculus that enclose the ovule more or less totally. Nevertheless, different terminologies were used by different workers in different research contexts (Planchon 1845; Baillon 1876; Corner 1949, 1953, 1976; Van Der Pijl 1972; Kapil et al. 1980). One of the defining properties of an aril is that it is generally fleshy, but the most important property of an aril is how it is initiated and how it develops. A true aril originates from the funiculus, and it can be described as the third seed envelope of the ovule (Endress 2011). This organ can be fleshy or hairy or can form wings; it may have its own vasculature, and sometimes it produces a mucilaginous pulp filling the locule.

Planchon (1845) described the putative arils of Celastrus scandens L. and Euonymus latifolius (L.) Mill. as arillodes – false arils – and cited these false arils as derived from the exostome of the outer integument rather than from the funiculus. This is very similar to our results, but not according to Planchon’s drawing. Here the arillode originates from two distinct primordia, one coming from the funiculus and the 63 other from the exostome. Despite this unusual development Planchon regarded it as an aril. In contrast to that, Miers (1856) described the same structure as originating exclusively from the funiculus and treats it as a normal aril.

Our results on Euonymus europaeus L., however, are completely different from those of Miers and from Planchon as well. One reason is that Miers did not complete developmental study, but analyzed only seeds and rather late stages of seed development in Euonymus europaeus L. (Miers 1856). Pfeiffer (1891) described the arils of Celastrus P.Browne, Euonymus L. and Celastrus cassinoides L’Her. as derived from the exostome and the hilum. Corner (1976) described the aril of Euonymus glandulosus (merr.) Ding Hou as entirely derived from the funiculus, and the aril of other species (Catha edulis (Vahl) Endl. , Celastrus paniculatus Willd., Sarawakodendron filamentosum Ding Hou) as derived from the exostome and the funiculus. This can be interpreted as diversity within the Celastraceae which also gives us a way to understand the phylogeny of Celastraceae and the order of Celastrales. Van der Pijl (1972) concluded that Euonymus L. has an arillode, not an aril. Van der Pijl used the term arillode rather as an ecological term than as a morphological one. In his interpretation, the fleshy structures which cover the seed and do not fulfil his definition of an aril are all summarized as arillodes. This, however, is not really a solution as it shifts ambiguities to another term instead of solving them in a meaningful way. Together with the aril there are four different types of fleshy seed appendages, including the sclerotesta in Magnolia L. and the strophiole and the caruncula in Euphorbiaceae.

It is of some interest that all these secondary fleshy appendages are either associated with the micropyle or with the funiculus or the hilum. Both the micropyle and the hilum are weak points in the protective structure covering endosperm and embryo. They can get closed by mere shrinking of the surrounding tissues, but they can also close by active growth of adjacent cells. This could be a preadaptation leading to the fleshy appendages by stepwise enlargement. In orthotropous seeds, fleshy structures originating from micropyle or funiculus are clearly separate from each other, and usually only one option is present. In anatropous seeds, the two zones are close together. Only minor meristem incorporation is necessary to build up mixed forms in which micropylar appendages and funicular appendages merge. In the same way the fleshy parts originating only from the micropyle or only from the 64 funiculus may incorporate the complementary structure easily. It can be thus expected that structures that are very similar at maturity display rather different developmental patterns. For diagnostic purpose, it is not meaningful to distinguish these structures in detail. Used in an evolutionary context, fleshy seed appendages might supply even more information than actually used.

Lophopyxis Hook. was regarded as a genus with an arillate structure belonging to Celastraceae, but according to Simmons and Hedin (1999), Lophopyxis Hook. is better placed in Euphorbiaceae. This conclusion is also supported by our results saying that the aril in Celastraceae should be called caruncula. In the system of APG III (Angiosperm Phylogeny Group 2009), Celastrales is only sister group of the orders of Malpighiales and Oxalidales. So the relationship between the Celastraceae and Euphobiaceae is closer than generally suggested. In the papers by Simmons and Hedin (1999), Simmons (2004), and Simmons et al. (2008, 2011) the character aril is used without any clear description, they regard this structure in Celastraceae as a true aril. In 2012, Simmons et al. concluded that "current intrageneric classifications of Euonymus L. are not completely natural and require revision". But prior to all tests, the development of the morphological characters of this family should be clearly understood.

4.11 Seed wing development in Catha edulis

In contrast to most other Celastraceae, the gynoecium of Catha edulis (Vahl) Endl. enlarges and prolongates during the development. The seed insertion is shifted at the same time from the base to the middle of the carpel (Fig. 3). This generates a space for the caruncula rising from the exostome and growing towards the base of the gynoecium. In Euonymus L. and Celastrus L. (Zhang et al. 2012), the gynoecium enlarges more or less equally without pronounced prolongation. As a consequence, the growth of the micropylar appendages is directed more or less exclusively to the chalazal end. In Catha Endl. this growth towards the chalazal end is also present, but it occurs late and leads to a membranous duplicator.

Simmons (2004) mentioned that the seed wings in Catha edulis (Vahl) Endl. have been described as an aril derived from the funiculus and the side of the exostome 65 next to the funiculus. His descriptions are based on Loesener (1942) and Corner (1976). However, Loesener described the Catha edulis (Vahl) Endl. seed wing as an arillus and did not mention an origin from the exostome next to the funiculus or from the funiculus. According to his illustrations, this arillus does not originate from the funiculus at all, but from the micropyle. Corner (1976) studied the seed wing of Catha edulis (Vahl) Endl. and concluded that the aril originates from the funiculus and the side of the exostome next to the funiculus. Our results are different from those by Corner and show that the seed wing arises only from the exostome parts on the other side of the funiculus. The funiculus may be covered by this structure, but it is not incorporated in it. Finally, the two lobes of the micropylar part of the outer integument overlap both the micropyle and the funiculus. From the illustration by Corner we know that only mature seeds have been sectioned. In this final stage it is very difficult to detect the correct position of the wing. It is not possible to understand the rather complex process leading to the mature structure. According to our results, the micropyle is freely accessible during the entire development in Celastraceae as already described for Celastrus L. and Euonymus L. (Zhang et al. 2012). The term aril is often used for a variety of ovular appendages or seed appendages. Kapil et al. (1980) discussed the concept of arils in the historical context and explained classification and terminology. While “aril” is mostly applied to fleshy structures it is sometimes also used for dry appendages that are supposed to be homologous to fleshy ones. For diagnostics, the more superficial approach may be sufficient and very helpful as plant identifications have to be possible without developmental studies. Within an evolutionary framework a more precise approach is essential that includes formation, position and structure of the appendages under consideration. Otherwise, functional analogies might get mixed up with synapomorphies (Endress 1973).

It is interesting to see the duplicature that develops from the micropylar part of the outer integument in direction to the chalazal end of the micropyle. When becoming fleshy, these duplicature is the attractor for animals involved in seed dispersal. In this way the dispersing animal cannot separate the edible part from the part to be transported. In Catha Endl. the duplicature becomes dry and does not longer serve for dispersal. The dispersal in Catha Endl. is promoted only by the wing-like micropylar prolongation directed parallel to the funiculus. This indicates that the 66 fleshy caruncula might be the primitive condition while dry wings are derived from the later. Furthermore, it is interesting to note that the wing is a conduplicate structure. To fulfil the needs of a wing, a simple structure would be sufficient. In earlier studies, the conduplicate nature of the wing has not been described.

A long period of dry-condition environmental pressure can be interpreted as one reason for fleshy structures modified to membranous wings. In Celastraceae, Catha edulis (Vahl) Endl. with its basal wings is similar to Canotia holacantha Torr. (Simmons and Hedin 1999) which lives at the higher elevations of Baja California and Sonora desert in North America (Little 1976). Under natural conditions, Catha edulis (Vahl) Endl. grows in and on the margins of dry evergreen forests and mist forests at elevations of 1200-2500 m at the Horn of Africa and on the Arabian Peninsula (Corkery et al. 2011). Those two species, Catha edulis (Vahl) Endl. and Canotia holacantha Torr., live at higher elevations near to the sea. The one-sided wing provides the means for dynamic propulsion to disperse the seed in the dry environment (van der Pijl 1972). Moreover, Corner (1976) and Simmons and Hedin (1999) interpreted the seed wing as modified arils throughout the Celastraceae, even the large apical and circular wings of Kokoona Thw., Lophopetalum Benth. & Hook. f. and Peripterygia Baill. The large basal wings of Hippocratea s.l. are also interpreted as modified arils. The fleshy tissue transformation to a seed wing is also apparent in other arillate groups with incidental wings (Meliaceae and Bombacaceae), and in some African leguminous high trees belonging to the Piptadenieae (van der Pijl, 1972). However, within Celastraceae s.l., a basal wing with vasculature of the funiculus along the wing occurs only in Hippocratea s.l. An aril modified to be a wing surrounding the seed with median (or basal in Kokoona Thw.) attachment of the vasculature of the funiculus occurs only in Kokoona Thw., Lophopetalum Benth. & Hook. f. and Peripterygia Baill.. More developmental studies of the seed wing within Celastraceae are needed to interpret the evolution of this structure.

The seed wing of Catha edulis (Vahl) Endl. rises from the exostome and then extends the entire funiculus until it finally encircles both the micropyle and the funiculus. This proves again that the aril in Celastraceae is better referred to as a caruncula. This one-side wing provides the means for dynamic propulsion to disperse the seed in the dry environment. 67

4.12 Caruncula and aril

Gaertner (1788) is considered the first to use the term aril to describe accessory seed appendages (Kapil et al. 1980, Rodriguez-Rian et al. 2006). Until Kapil et al. (1980) classified the various types of seed appendages based on growth mode and attachment, different workers used different terms to name the same structure or the same term for different structures (Planchon 1845, Baillon 1867, Pfeiffer 1891, Corner 1949, 1953, 1976, van der Pijl 1972, Kapil et al. 1980). This led to much confusion, especially for species in which the existing descriptions are incomplete or even contradictory.

The seed appendage of Homalanthus Juss. has been documented as a pulpy testa (Corner 1976), an arillode (Esser 1997), a well-developed aril (Gardner 1999) and a fleshy aril (caruncula) (Tokuoka and Tobe 2002). However, none of these workers studied the ontogeny of this structure. From our developmental study it is clear that the seed appendage of Homalanthus populifolius Graham develops from the exostome and therefore must be termed a caruncula. This partially supports the terminology of Tokuoka and Tobe (2002), but on the other hand they did not distinguish clearly between aril and caruncula. Fig. 6 presents the interpretation of the differences between aril and caruncula: A typical aril is a fleshy structure arising from the funiculus and later covering the anatropous ovule including the micropyle, as exemplified by Passiflora citrina J.M. MacDougal(Fig. 5). A caruncula is only a small and disc-like appendage, with its attachment and growth limited to the exostome rim; the micropyle may be discernible in its centre (Baillon 1858). Size and shape are not diagnostic, instead we characterize a caruncula as a fleshy structure arising from the exostome and always leaving the micropyle open. The difference between aril and caruncula was discussed in Celastraceae (Zhang et al. 2012), and we would interpret the so-called aril in Celastraceae as a caruncula. There is also an important functional difference between caruncula and aril. A Passiflora-like aril can be formed only after fertilization (Fig. 5B) as it would otherwise block the access of the pollen tube to the micropyle. The caruncula does not cover the micropyle and may start to develop even prior to fecundation fertilization.

4.13 Stomata on the outer integument

The presence of stomata on the outer integument is rare among angiosperms (Netolitzky 1926, Jernstedt and Clark 1979, Paiva et al. 2006, Werker 1997). Stomata on the outer integument were first discovered in 1839 in Canna L. by Schleiden (Werker 1997, and reference therein). Until now, they have been observed almost exclusively in the exotesta of 28 families of angiosperms, four of them monocotyledons and 24 dicotyledons. Among these, stomata are relatively common in Malvales, while rare in Fabaceae (Werker 1997).

The occurrence of stomata on the outer epidermis of the testa was suggested by Corner (1976) to indicate a primitive or unspecialized state of the testa, which according to the classical theory of the carpel represents the underside of a pinna. However, in the classical theory, a megaspore is borne on the back of the megasporophyll, and the carpel exhibits centripetal doubling or centripetal folding. If the ovule was borne on the carpel edge, it should also lie in the radial plane. Dorsiventral gene expression in the outer integument was interpreted as potentially reflecting a leaf-derived origin (Meister et al. 2002, Yamada et al. 2003). So the stomata may be the relic of the outer integument as a leaf structure.

The difference between stomata on the outer integument and in the leaf is their function. The stomata on the leaf surface can open and close under different conditions. However, the stomata on the outer integument stay open for the whole development of the seed, even in the ripe seed (Fig. 4F). Maheshwari (1950) reported that stomata on the outer integument of Gossypium L. are concerned with respiration rather than with transpiration or photosynthesis. However, the ovule and immature seeds which are covered by carpels, the nucellus or the embryo do not need to use the stomata on the outer integument for respiration. The stomata stay open during ovule development what indicates that the stomata on the outer integument have no function for respiration. When the seed is mature, the stomata stay permanently open. The stomata in the seed coat are large static pores, the function of which is related to imbibitions (Paiva et al. 2006).

We marked the 28 families reported by Werker (1997) and counted in APG III (Angiosperm Phylogeny Group 2009). There are four families of monocots and 24 families of dicotyledons which have stomata on the outer integument (Angiosperm 69

Phylogeny Group 2009). There is 1 family (Nymphaeaceae) of ANITA-group, 2 families (Magnoliaceae and Myristicaceae) of Magnoliids, 21 families of Eudicots, 17 families of the Rosids and 4 families of Monocots. There is only one family (Myrsinaceae) of Asterids. No family within Euasterids I and II. Myrsinaceae is the only member within Asterids for which stomata on the outer integument are reported suggesting a more close relationship to Rosids. So this result is not like the normal conclusion that the presence of stomata on outer integument are scattered among unrelated families.

5 Summary

The integument is the morphologic unit that has received the most attention in discussions on the evolution of the seed plants. There are several theories and hypotheses about the origin of the integument were presented in the history. However, the development and function of the ovule envelopes are not so clear until now. We selected the basal gymnosperms, Cycas L. and Zamia L., and basal angiosperms, Magnolia L., and relatively derived Celastraceae to investigate the development of the ovule, especially of the integument to complement the existing knowledge in seed plants. The development of ovules of seed plant is documented with morphological and anatomical techniques using LM and SEM.

The nucellar beak found in Zamia is a structure that has not been recorded previously. It protrudes from the micropyle at pollination and may be the primary acceptor for pollen. There are striking similarities to the lagenostom or salpinx in Lyginopteridatae. There may be an evolutionary way to interpret the pollination drop existing in the Lyginopteridatae. Probably the nucellar beak of Cycads, even Ginkgoales, have the same function as the lagenostom or salpinx of the Lyginopteridatea. Unfortunately, pollen and transport inside the pollination chambers have not been observed. Further analysis of this unusual structure seems to be very important.

The development of Magnolia L. shows that the outer integument differentiates into two layers, an outer fleshy one well filled with oil receptacles, and an inner stony layer of bony hardness. The inner integument forms only a thin layer. This supports the results of Boer and Bouman (1972).

The developmental studies of three species of Celastraceae, however, turned out that the structure termed aril in this family does not originate from the funiculus or the hilum, but from the exostomatic micropyle. As a consequence, the micropyle is not inside the aril, but at the base of the fleshy structure which is thus better referred to as a caruncula. The fleshy part of seeds in Celastraceae differs thus markedly from those seed appendages usually referred to as an aril. The seed wing of Catha edulis (Vahl) Endl. (Celastraceae) has been described as an aril derived from the funiculus, while the aril of other Celastraceae is considered arising from the micropyle instead of the funiculus. The wing is a modified caruncula derived from the micropylar region. 71

Thus, seed wings are not only interesting in respect to function, but may carry also interesting and important evolutionary signals. The evolution of different micropylar appendages within Celastraceae is discussed.

Seed appendages are structures of rich diversity which have different biological functions and can show different evolutionary pathways. However, a description of these structures without a developmental study would be incomplete or even contradictory. Homalanthus populifolius Graham (Euphorbiaceae) and Passiflora citrina J. M. MacDougal were selected to study the processes of carpel, ovule and seed appendage development in order to compare different models of caruncula and aril. A caruncula is confirmed to exist in Homalanthus populifolius Graham. The presence of the stomata on the outer integument is a new discovery in this genus. The stomata on the outer integument are compared with those on the leaf, and the evolutionary implications are discussed.

6 Zusammenfassung

Das Integument ist die morphologische Einheit, der in Diskussionen über die Evolution der Samenpflanzen die meiste Aufmerksamkeit gewidmet wurde. In der Vergangenheit wurden mehrere Theorien und Hypothesen über den Ursprung des Integuments aufgestellt, dennoch sind Entwicklung und Funktion des Integuments bis jetzt weitgehend unklar. Wir haben die basalen Gymnospermen Cycas und Zamia, die basale Angiosperme Magnolia und die relativ abgeleiteten Celastraceae ausgesucht, um die Entwicklung der Samenanlage und insbesondere des Integuments zu untersuchen und den aktuellen Wissensstand zu erweitern. Die Entwicklung der Samenanlagen der Samenpflanzen wird mit morphologischen und anatomischen Methoden (LM und SEM) dokumentiert.

Der Nucellus-Fortsatz, der in Zamia gefunden wurde, ist eine zuvor noch nicht beschriebene Struktur. Er ragt zum Zeitpunkt der Bestäubung aus der Mikropyle heraus und könnte der primäre Pollenempfänger sein. Es gibt auffällige Ähnlichkeiten mit dem Lagenostom oder der Salpinx der Lyginopteridatae. Es könnte einen evolutionären Weg geben, um den Bestäubungstropfen der Lyginopteridatae neu zu interpretieren. Wahrscheinlich hat der Nucellus-Fortsatz der Cycadeen und Ginkgoales dieselbe Funktion wie das Lagenostom oder die Salpinx der Lyginopteridatea. Leider konnten Pollen und deren Transport in die Bestäubungskammern nicht beobachtet werden. Eine weitere Analyse dieser ungewöhnlichen Struktur scheint sehr wichtig zu sein.

Die Entwicklung von Magnolia zeigt, dass sich das äußere Integument in zwei Schichten aufteilt, eine äußere fleischige, die mit Ölkörperchen gefüllt ist, und eine innere, steinige von knöcherner Härte. Das innere Integument bildet nur eine dünne Schicht. Das unterstützt die Ergebnisse von Boer und Bouman (1972).

Die Entwicklungsstudien an drei Arten der Celastraceae zeigen jedoch, dass die Struktur, die in dieser Familie Arillus genannt wird, nicht aus Funiculus oder Hilum entsteht, sondern aus der exostomatären Mikropyle. Deshalb befindet sich die Mikropyle nicht innerhalb des Arillus, sondern an der Basis der fleischigen Struktur, die daher besser Caruncula genannt werden sollte. Der fleischige Teil der Samen der Celastraceae unterscheidet sich somit deutlich von denjenigen Samenanhängseln, die normalerweise als Arillus bezeichnet werden. Der 73

Samenflügel von Catha edulis (Vahl) Endl. (Celastraceae) wurde als vom Funiculus abgeleiteter Arillus beschrieben, während der Arillus anderer Celastraceae als von der Mikropyle abgeleitet aufgefasst wird. Der Flügel ist eine modifizierte Caruncula, die sich von der Mikropylenregion ableitet. Samenflügel sind daher nicht nur wegen ihrer Funktion interessant, sondern können auch interessante und wichtige evolutionäre Signale aussenden. Die Evolution verschiedener Mikropylenanhängsel innerhalb der Celastraceae wird diskutiert.

Samenanhängsel sind Strukturen großer Diversität mit verschiedenen biologischen Funktionen und evolutionären Pfaden. Eine Beschreibung dieser Strukturen ohne eine Entwicklungsstudie wäre allerdings unvollständig oder sogar widersprüchlich. Homalanthus populifolius Graham (Euphorbiaceae) und Passiflora citrina J. M. MacDougal wurden gewählt, um die Abläufe der Entwicklung von Karpell, Samenanlage und Samenanhängsel zu untersuchen, um verschiedene Modelle von Caruncula und Arillus zu vergleichen. Das Vorhandensein einer Caruncula bei Homalanthus populifolius wird bestätigt. Das Vorhandensein von Stomata am äußeren Integument ist eine neue Entdeckung in dieser Gattung. Die Stomata des äußeren Integuments werden mit denen der Blätter verglichen und die evolutionären Schlussfolgerungen werden diskutiert.

7 Bibliography

Andrews, H. N. 1961. Studies in palaeobotany. New York.

Andrews, H. N. 1963. Early seed plants. Science 142: 925-931.

Angiosperm Phylogeny Group 2009: An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG III. – Bot. J. Linn. Soc. 161(2): 105-121.

Anhaeusser, H. 1953. Keimung und Schlauchwachstum des Gymnospermenpollens unter besonderer Berücksichtigung des Wuchsstoffproblems. Beitr. Biol. Pfl. 29: 297-338.

Baillon, H. 1858. Etude générale du groupe des Euphorbiacées. Paris.

Baillon, H. E. 1867. Sur I'origine de I'arille des Carumbium. Adansonia 6: 348-351.

Baillon, H. E. 1876. Sur l'origine du macis de la Muscade et des arilles en general. Adansonia 11, 329-340.

Beck, C. B. 1974. Origin and early evolution of angiosperms: A perspective Origin and Early Evolution of Angiosperms. Columbia University Press, New York.

Benson, M. 1904. Telangium scottii, a new species of Telangium (Calymmatotheca) showing structure. Ann. Bot. 18: 161-176.

Bhandari, N. N. 1967. Magnoliaceae, in: Seminar on comparative embryology of Angiosperms. New Delhi.

de Boer, R. and Bouman, F. 1972. Integumentary studies in the Polycarpicae. II. Magnolia stellata and Magnolia virginiana. Acta Bot. Neerl. 21: 617-629.

Boesewinkel, E. D. and Bouman, F. 1984. The seed: Structure. In: Johri B. M. (ed.), Embryology of angiosperms. Berlin: Springer Verlag 567-610.

Bouman, F. 1984. The ovule. In: Embryology of angiosperms. (Ed. BM Johri) 123- 157. (Springer- Verlag: Berlin)

Camp, W. H. and Hubbard, M. M. 1963. On the origins of the ovules and cupule in Lyginopterid pteridosperms. Am. J. Bot. 50: 174-178. 75

Chamberlain, C. J. 1919. The living Cycads. The University of Chicago Press, Chicago, Illinois.

Chamberlain, C. J. 1935. (reprint 1966). Gymnosperms: structure and evolution. Chicago Univ. Press, Chicago.

Corkery, John M. et al. 2011. Overview of literature and information on "khat-related" mortality: a call for recognition of the issue and further research. Ann. Ist. Super. 47, 445-464.

Corner, E. J. H. 1949. The Durian Theory or the Origin of the Modern Tree. Ann. Bot. 13(4): 367-414.

Corner, E. J. H. 1951. The leguminous seed. Phytomorphology 1: 117-150.

Corner, E. J. H. 1953. The Durian Theory extended I. Phytomorphology 3: 465-476.

Corner, E. J. H. 1976. The Seeds of Dicotyledons, Vols. 1 and 2. Cambridge, Cambridge University.

Davis, G. L. 1966. Systematic Embryology of the Angiosperms. John Wiley & Sons, INC. New York. London. Sydney.Davis. de Haan, H. R. M. 1920. Contribution to the knowledge of the morphological value and phylogeny of the ovule and its integuments. Rec. Trav. Bot. Néerl. 17 219- 324.

Dogra, P. D. 1964. Pollination mechanisms in gymnosperms, - In: P. K. K. Nair (Ed.): Recent Advances in Palynology. 142-175. National Botanic Gardens, Lucknow.

Dörken V., Zhang, Z. X., Mundry, I. and Stützel, Th. 2011. Morphology and anatomy of male reproductive structures in Pseudotaxus chienii (W.C. Cheng) W.C. Cheng (Taxaceae). Flora 206(5): 444-450.

Doyle, J. and O’Leary, M. 1935. Pollination in Pinus. Sci. Proc. Roy. Dublin Soc. 21: 181-190.

Eames, A. J. 1961. Morphology of angiosperms. New York.

Earle, T. T. 1938. Origin of the seed-coats in Magnolia. Am. J. Bot. 25, 211. 76

Eichler, 1889. in: Engler u. Prantl, Natuerl. Pflanzenfamilien II. Leipzig.

Endress, P. K. 1973. Aril and aril-like structures in woody Ranales. New Phytol. 72, 1159–1171.

Endress, P. K. 1996. Structure and function of female and bisexual organ complexes in Gnetales. Int. J. Plant Sci. 157: S113-S125.

Endress P. K. and Igersheim A. 2000a: The reproductive structures of the basal angiosperm Amborella trichopoda (Amborellaceae). Int. J. Plant Sci. 161, Suppl.: 237-248.

Endress P. K. and Igersheim A. 2000b: Gynoecium structure and evolution in basal angiosperms. Int. J. Plant Sci. 161, Suppl.: 211-223.

Endress, P. K. 2001. Origins of flower morphology. In: The character concept in evolutionary biology. Ed. GP Wagner San Diego, CA: Academic Press 493-510

Endress, P. K. 2006. Angiosperm Floral Evolution: Morphological developmental framework. Adv.Bot. Res. 44: 1-61.

Endress, P. K. 2011. Angiosperm ovules: diversity, development, evolution. Ann. Bot. 107:1465-1489.

Esser, H. J. 1997. A revision of Homalanthus (Euphorbiaceae). In: Malesia. Blumea 42: 421-466.

Floyd, A. G. 2008. Rainforest Trees of Mainland South-Eastern Australia 2nd ed. Terania Rainforest Publishing, b&w illustrations.

Friedman, W. E. 2009. The meaning of Darwin’s “abominable mystery.” Am. J. Bot. 96: 5-21.

Gaertner, J. 1788. De fructibus et seminibus plantarum. Stuttgart.

Gardner, Rh. O. 1999. Notes on the fruit and seed of Homalanthus (Euphorbiaceae). Adansonia 21(2): 301-305

Gensel, P. G. and Andrews, H. N. 1984. Plant life in the Devonian. Praegler, New York. 77

Gerlach, D. 1984. Botanische Mikrotechnik. 3. Aufl. Thieme, Stuttgart.

Gerstberger, P. and Leins, P. 1978. Rasterelektronenmikroskopische Unter- suchungen an Blütenknospen von Physalis philadelphica (Solanaceae). Ber. Dt. Bot. Ges. 91:318-387.

Gifford, E. M. and Foster, A. S. 1996. Morphology and Evolution of Vascular Plants. W. H. Freeman and Company.

Gray, A. 1848. Genera Florae Americae Boreali-Orientalis Illustrat I. The Genera of the Plants of the United States illustrated by figures and analyses from nature. James Munroe and Company, Boston.

Gray, A. 1857. A short exposition of the structure of the ovule and seed-coats of Magnolia. J. Linn. Soc. Bot. 2, 106-110. von Guttenberg, H. 1971. Bewegungsgewebe und Perzeptionsorgane. In: Zimmermann W, Carlquist S, Ozenda P, Wulff HD (Hrsg) Handbuch der pflanzenanatomie, 2. Neubearb. Aufl. Band V/5. Borntraeger, Berlin Stuttgart. de Haan, H. R. M. 1920. Contribution to the knowledge of the morphological value and phylogeny of the ovule and its integuments. Rec. Trav.Bot.Neerl. 17:219-324.

Herr, J. M., Jr. 1995. The origin of the ovule. Am. J. Bot. 82(4) 547-564.

Heywood, V. H. 1993: Flowering Plants of the World. – Oxf. Univ. Press., New York.

Hou, D. 1962. Celastraceae I. In: C. G. G. J. VAN STEENIS (ed.): Flora Malesiana 6(2): 227-291.

Igersheim A. and Endress P. K. 1997: Gynoecium diversity and systematics of the Magnoliales and winteroids. – Bot. J. Linn. Soc. 124: 213-271.

Jernstedt, J. A. and Clark, C. 1979. Stomata on the fruits and seeds of Eschscholtzia (Papaveraceae). Am. J. Bot. 65: 586-590.

Jones, D. L. 1993. Cycads of the world. Reed Books, Chatswood.

Kapil, R. N. and Bhandari, N. N. 1964. Morphology and embryology of Magnolia. Proc. Nat. Inst. Sci. India 30, 245-262. 78

Kapil, R. N., Bor, J., Bouman, F. 1980. Seed appendages in angiosperms. Botanische Jahrbücher für Systematik, Pflanzengeschichte und Pflanzengeo- graphie 101: 555-573.

Krikorian, A. D. 1985. Growth mode and leaf arrangement in Catha edulis (kat). – Econ. Bot. 39: 514-521.

Little, E. L. 1976. Atlas of United States Trees: Minor western hardwoods, Volume 3. Government Printing Office, American.

Loesener, L. E. T. 1942. Celastraceae. in: Engler, A., Harms, H. & Mattfeld, J. (eds.), Die natürlichen Pflanzenfamilien.

Long, A. G. 1966. Some lower carboniferous fructications from Berwickshire, and carpels. T. Roy. Sci. Proc. Roy. Dublin. Soc. 23:35-54.

Maheshwari, P. 1950. An introduction to the embryology of angiosperms. McGraw- Hill Book Co., Inc., New York, Toronto, London.

Maneval, W. E. 1914. The development of Magnolia and Liriodendron. Bot. Gaz. 57:1-31.

Matsui, M., Imaichi, R., and Kato, M. 1993. Ovular development and morphology of some Magnoliaceae species. J. Plant Res. 106:297-304.

Matte, M. 1904. L’appareil libéro-ligneux des Cycadacées. Caen.

McAbee, J. M., Kuzoff, R. K. and Gasser, C. S. 2005. Mechanisms of derived unitegmy among Impatiens species. Plant Cell 17 1674–1684.

McWilliam, J. R 1958. The role of the micropyle in the pollination in Pinus. Bot. Gae. 120: 109-117.

Meeuse, A. D. J. 1964. The bitegmic spermatophytic ovule and the cupule a re- appraisal of the so-called pseudomonomerous ovary. Acta Bot. Neerl. 13:97 - 112.

Meeuse, A. D. J. and Bouman, F. 1974. The inner integument – Its probable origin and homology. Acta. Bot. Neerl. 1974 23: 237-249. 79

Meister, R. J., Kotow, L. H. and Gasser, C. S. 2002. SUPERMAN attenuates positive positive INNER NO OUTER autoregulation to maintain polar development of Arabidopsis ovule outer integuments. Development 129: 4281-4289.

Miers, J. 1856. Remarks on the nature of the outer fleshy covering of the seed in the Clusiaceae, Magnoliaceae, and on the development of the raphe in general, under its various circumstances. Trans. Linn. Soc. London 22: 81-96.

Netolitzky, F. 1926. Anatomie der Angiospermen-Samen. In: K. Linsbauer [ed.], Handbuch der Pflanyenanatomie. Band X, Teil 2.

Nordal, A. 1980. Khat: pharmacognostical aspects. Bull. Narc. 32, 51 –64.

Norstog, K and Stevenson, D.W. (1980). Wind? Or insects? The pollination of Cycads. Fairchild Trop. Gard. Bull. 35(1): 28-30.

Oberprieler RG.1993. Cycads and their weevils: Can one survive without the other? Plant Pro. News. 4: 5-6.

Olesen, P. and Bruun, L. 1990. A structural investigation of the ovule in sugar beet, Beta vulgaris: integuments and micropyle. Nord. J. Bot. 9:499-50

Oliver, F. W. and Scott, D. H. 1904. On the structure of the Palaeozoic seed Lagenostoma lomaxi, with a statement of the evidence upon which it is referred to Lyginodendron. Philosophical transactions of the Royal Society, London 197B: 193-247.

Paiva, A. S., Lemos-filho, and J. P., Oliveirad, M. T. 2006. Imbibition of Swietenia macrophylla (Meliaceae) seeds: the role of stomata. Ann. Bot. 98(1): 213-217.

Paris, R. and Moyse, M. H. 1958. Abyssinian tea (Catha edulis Forsk, Celastraceae). Bull. Narc. 10, 29 – 34.

Pfeiffer, A. 1891. Die Arillargebilde der Pflanzensamen. – Bot. Jahrb. Syst. 13: 492- 540. Pflanzenfamilien, (ed. 2), Engelmann, Leipzig, 87–197.

Pijl, L. van der 1953. On the flower biology of some plants from Java. – Ann. Bogor. 1: 77-99. 80

Pijl, L. van der 1972. Principles of dispersal in higher plants. – Springer, Berlin, Heidelberg, New York.

Planchon, M. J. E. 1845. Développements et caractères des vrais et des faux arilles. – Ann. Sci. Nat. Bot., sér. 3, 3: 275-312.

Revri, R. 1983. Catha edulis Forsk: Geographical dispersal, botanical, ecological and agronomical aspects with special reference to Yemen Arab Republic. Göttingen: Göttinger Berträge zur Land – und Forstwirtschaft in den Tropen und Subtropen.

Rieger, M. 2002. Morphologischer und histologischer Vergleich der Zapfenentwicklung bei Podocarpaceae und Phyllocladaceae. (diploma thesis).

Rodriguez-Rian, O., Francisco, J., Valtuen, A., and Ortega-Olivencia, A. 2006. Mega-sporogenesis, megagametogenesis and ontogeny of the aril in Cytisus striatus and C. multiflorus (Leguminosae: Papilionoideae). Ann. Bot. 98: 777-791.

Schulz, Ch. and Stützel, Th. 2007. Evolution of taxodiaceous Cupressaceae. Org. Divers. Evol. 7: 124-135.

Shisode, S. B. and Patil, D. A. 2011. Taxonomic and phylogenetic census of the Celastrales: A synthetic review. Curr. Bot. 2(4): 36-43.

Simmons, M. P. and Hedin, J. P. 1999. Relationships and morphological character change among genera of Celastraceae sensu lato (including Hippocrateaceae). Ann. Missouri Bot. Gard. 86, 723–757.

Simmons, M. P. 2004. Celastraceae. In: Kubitzki, K. (Ed.), The families and genera of flowering plants VI. Springer, Berlin, 29-64.

Simmons, M. P., Cappa, J. J., Archer, R. H., Ford , A. J., Eichstedt, D. and Clevinger, C. C. 2008. Phylogeny of the Celastreae (Celastraceae) and the relationships of Catha edulis (qat) inferred from morphological characters and nuclear and plastid genes. Mol. Phyl. and Evo. 48, 745-757.

Simmons, M. P., Mckenna, M. J., Bacon, C. D., Yakobson, K., Cappa, J. J., Archer, R. H. and Andrew J. 2012. Phylogeny of Celastraceae tribe Euonymeae inferred from morphological characters and nuclear and plastid genes. – Mol. Phylogenet. Evol. 62(2): 9-20. 81

Singh, H. 1978. Embryology of Gymnosperms Encyclopedia of Plant Anatomy. Gebrüder Borntraeger, Berlin, Stuttgart.

Smith, D. L. 1964. The evolution of the ovule. Bio. Rev. 39: 137-159.

Stebbins, G. L. 1974. Flowering plants. Evolution above the species level. Cambridge, MA: Belknap Press of Harvard University Press.

Stevenson, D. W. 1987. Comments on character distribution, taxonomy, and nomenclature with respect to the genus Zamia L. in the West Indies and Mexico. Encephalartos 9: 3.

Stewart, W. N. (1983). Paleobotany and the Evolution of Plants, 1st ed. New York: Cambridge University Press.

Stopes, M. C. 1904. Beiträge zur Kenntnis der Fortplanzungsorgane der Cycadeen. Flora 93, 435-482.

Stopes, M. C. 1905. On the double nature of the Cycadean integument. Ann. Bot. Vol. 19. No. 76.

Stützel, Th. and Röwekamp, I. 1999. Female reproductive structures in Taxales. Flora 194: 145-157.

Taylor, T. and Taylor, E. 1993. The biology and evolution of fossil plants, Prentice Hall, Englewood Cliffs, NJ.

Taylor, T., Taylor, E., Krings, M. 2009. Paleobotany – the biology and evolution of fossil plants. Second edition. Elsevier.

Tilton, V. R. 1981. Ovule development in Ornithogalum caudatum (Liliaceae) with a review of selected papers on angiosperm reproduction. IV. Egg apparatus structure and function. New Phytologist 88:505-531.

Tilton, V. R. and Lersten, N. R. 1981. Ovule development in Ornithoglum caudatum (Liliaceae) with a review of selected papers on angiosperm reproduction. III. Nucellus and megagametophyte. New Phytologist 88:477-504.

Tokuoka, T. and Tobe, H. 2002. Ovule and seed in Euphorbiodeae (Euphorbiaceae): structure and systematic implications. J. Plant Res. 115:361-374. 82

Tomlinson, P. B. and Takaso, T. 1989. Cone and ovule ontogeny in Phyllocladus (Podocarpaceae). Bot. J. Linn. Soc. 99, 209-221.

Walton, J. 1953. The evolution of the ovule in the pteridosperms. Adv. Sci. 38:223- 230.

Werker, E. 1997. Seed anatomy. Gebrüder Borntraeger, Berlin, Stuttgart.

Whitelock, L. M. 2002. The Cycads. Timber Press, Portland.

William, H. L. 1900. Studies in the development and morphology of Cycadean Sporangia: II the ovule of Stangeria paradoxa, Ann. Bot. Vol 14

Worsdell, W. C. 1904. The structure and morphology of the ovule: an historical sketch. Ann. Bot. 18: 57-86.

Yamada, T, Ito, M. and Kato, M. 2003. The outer integument and funicular outgrowth complex in the ovule of Magnolia grandiflora (Magnoliaceae). J. Plant. Res. 116:189-98

Yan H, Yang H.-Y. and Jensen WA. Ultrastructure of the micropyle and its relationship to pollen tube growth and synergid degeneration in sunflower. Sexual Plant Reproduction 1991;4:166-175.

Zhang, X., Zhang, Z. and Stützel, T. 2012. Aril development in Celastraceae. Feddes Repert.122, 445-455.

Ziegler, H. 1959. Über die Zusammensetzung des “Bestäubungstropfens” und den Mechanismus seiner Sekretion. Planta (Berlin) 52:587-599.

Zimmermann, W. 1938. Die Telomtheorie. Biologe: Monatsschrift zur Wahrung der Belange der Deutschen Biologen 7: 385-391.

Zimmermann, W. 1952. Main results of the telome theory. Palaeobotanist, 1, 456- 470.

Curriculum Vitae

Name: Xin Zhang Date of Birth: 25-01-1984

Birth Place: Urad Qiangi (Inner Mongolia, China)

Tel.: +49.176.99927185

E-mail: [email protected]

Address: Ruhr-Universität Bochum LS Evolution und Biodiversität der Pflanzen Universitätsstraße 150 D-44801 Bochum

Education:

2001-2005 B.Sc. in Environmental Science, Inner Mongolia University, Hohhot, China. 2005-2008 Participation in M.Sc. degree program for Wildlife Conservation and Utilization, Beijing Forestry University, Beijing, China. 2008-2009 PhD student in Wildlife Conservation and Utilization Beijing Forestry University, Beijing, China. 2009-2013 PhD thesis in Plant Morphology and Evolution, Ruhr-Universität Bochum, Bochum, Germany.

Selected Articles:

(1) Zhang, Xin; Zhao, Yizhi and Zhang, Zhixiang, 2007. Classification and distribution of Comastoma (Gentianaceae) in Helan Mountains in China by floristic, ecological and geographical approaches. Forestry Study in China, 9(2): 147-151. (2) Zhang, Xin; Zhang, Binwei; Feng, Lei and Zhang, Zhixiang 2010. A research of rare medicinal plant resources in Beijing Songshan Nature Reserve, Advances in Biodiversity Conservation and Research in China VIII. (3) Zhang, Xin; Zhang, Zhixiang and Stützel, Thomas 2011. Aril development in Celastraceae, Feddes Repertorium, 122: 445-455. (4) Zhang, Xin; Zhang, Zhixiang and Stützel, Thomas, 2013. Ontogeny of the ovule and seed wing in Catha edulis (Vahl) Endl. (Celastraceae) (under revision in: Flora) 84

(5) Zhang, Xin; Zhang, Zhixiang and Stützel, Thomas 2013. Ovule development of Homalanthus populifolius Graham (Euphorbiaceae) and differentiation of caruncula and aril (manuscript finished) (6) Zhang, Xin; Zhang, Zhixiang and Stützel, Thomas 2013. Floral ontogeny of Illicium lanceolatum A.C. Sm. (Illiciaceae), with emphasis on carpel and ovule development. (manuscript finished) (7) Zhang, Xin and Stützel, Thomas 2013. Ovule development in Zamiaceae – nucellus morphology, anatomy and its implication for the pollination syndrom (in prep.)

Attended Academic Conferences:

2008/05: China's Second National Biodiversity Conservation and Nature Reserve Management building workshops in Kunming. 2012/09: “Biodiversity and Evolutionary Biology" of the German Botanical Society (DBG) 21st International Symposium in Mainz.

Books:

‘Plants in Beijing’ (in Chinese and English) author and photographer (published in 2008) ‘Trees in China’ (in Chinese) as a co-editor (in prep.).

Awards:

Poster Awards 2012 in September 16th-19th 2012, 21st International Symposium “Biodiversity and Evolutionary Biology” of the German Botanical Society (DBG) in Mainz.

Acknowledgments

It was a pleasure for me to work with all the wonderful people in our lab here in Bochum. First of all, I would like to thank Prof. Dr. Thomas Stützel, who gave me the chance to be his student and supported me in all aspects. Without his help, I would not have been able to finish my PhD thesis. I learned a lot during this time and I am convinced that this will help me in the future of my life.

I would like to thank Prof. Dr. Ralph Tollrian for reviewing my thesis. I am happy to have such a supportive co- supervisor.

I would like to thank Prof. Dr. Dominik Begerow and Prof. Dr. Wilfried Bennert for their support during courses and research.

I would like to thank Prof. Dr. Zhixiang Zhang for supporting me to get the chance to study in Germany

My thanks to my colleagues and friends for the four years I had in our department. I enjoyed their friendship, their support and the atmosphere. It was a pleasure to work with all these people and to benefit from their knowledge. My thanks to Sabine Adler; she did not only give technical support in the lab, but is also my closest friend in Germany. All my papers and the thesis were read by her for language support. She was so kind and invited me to her birthday parties and her lovely garden every year. Especially, I would like to thank Dr. Veit Dörken for the help at the very beginning of my research, for making schedules and teaching me experimental skills. Furthermore, thanks to Dr. Christian Schulz, Dipl. Biol. Nadja Balnis and Dr. Iris Mundry for the interesting and fruitful discussions and Dr. Patrick Knopf for excellent help with preparing the schematic drawings using CorelDRAW®.

My thanks to Dennise Stefan Bauer, Kristina Klaus, Friederike Grimmer and Jovani Pereira for their nice talk while writing their theses in our group.

My thanks to Hanno Boeddinghaus for computer support and Ilse Weßel, Sabine Kühle, Annika Fink and Pet Lerch for their kind management service.

Special thanks to all the gardeners in the Botanical Garden Bochum for their kind support. Furthermore, I would especially like to thank Frau Annette Höggemeier for her friendship and kind support in any situation you can think of.

My thanks to Jan Sieverding, Vasco Elbrecht and Abraham van Veen and his family for their friendship and making me feeling warm in the cold winter.

My thanks to Prof. Yanlu Ma, Prof. Yunliang Wang, Dr. Wei Xia, PhD. Zhiwei Chen, PhD. Qin Tang, PhD. Yue Yu, PhD. Binjia He and Ph.D Yellow River for your surrounding cured my homesick.

Last but not least, I wish to thank my families who have always supported me, my father Jianrong Zhang, my mother GaiGai Zhang and my sister Tao Zhang as well as little Hongbo Zhao. I thank my best friend Xiongfeng Yuan who is always at my side to support me.

This thesis has been supported by the Chinese Scholarship Council (CSC) and DAAD. Their support is gratefully acknowledged.

Feddes Repertorium 122 (2011) 7–8, 445–455

Research Paper

Aril development in Celastraceae

*, 1, 2 2 1 XIN ZHANG ; ZHIXIANG ZHANG & THOMAS STÜTZEL

1 Faculty of Biology and Biotechnology, Ruhr-Universität Bochum, Germany 2 Faculty of Biology and Biotechnology, Beijing Forestry University Beijing, China

Keywords: aril, Celastraceae, outer integument, development, Euonymus, Celastrus, caruncula

* Corresponding author: Universitätsstraße 150, Gebäude NDEF 05/770, D-44780 Bochum, Germany, E-mail: [email protected], [email protected]

Accepted for publication: June 6th, 2012.

DOI 10.1002/fedr.201200007

Abstract north-eastern Africa, the Arabian Peninsula, To learn more about the evolution of secondarily and Madagascar (KRIKORIAN 1985); Euony- intercalated seed envelopes, a series of developmen- mus, Celastrus, and Paxistima are widely culti- tal studies of arillate seeds in gymnosperms and vated as ornamentals; Kokoona zeylandica is angiosperms was undertaken. The goal was to test used as a source of oil, from Salazia the pulp of whether the second (outer) integument could be the fruit is eaten, and various species of Euo- derived from an aril of gymnospermous ancestors. In nymus are used for latex, medicine, and dyes our developmental studies of three species of Celas- (HOU 1962; HEYWOOD 1993). traceae, however, it turned out that the structure The aril is a very conspicuous and taxo- termed aril in this family does not originate from the funiculus or the hilum but from the exostomatic nomically important character within Celas- micropyle. As a consequence, the micropyle is not traceae. PLANCHON (1845) described the puta- inside the aril but at the base of the fleshy structure tive arils of Celastrus scandens and Euonymus which is thus better referred to as a caruncula. The latifolius as arillodes – false arils. PLANCHON fleshy part in seeds of Celastraceae differs thus cited these false arils as derived from the markedly from those seed appendages usually re- exostome of the outer integument rather than ferred to as an aril. from the funiculus. However, MIERS (1856) disputed PLANCHON’s conclusion and, based on Introduction his own investigation of Euonymus europaeus, he concluded that the aril is derived from the Celastraceae are a subcosmopolitan family of funiculus and is therefore a true aril. PFEIFFER 98 genera and about 1211 species with the (1891) described the arils of Celastrus, Euony- highest diversity in the tropics and subtropics mus, and Gymnosporia cassinoides as derived and with few temperate species (SIMMONS from the exostome and the hilum. CORNER 2004). Many species of Celastraceae are eco- (1976) described the aril of Euonymus glandu- nomically important for both traditional medi- losus as derived entirely from the funiculus, cine and horticulture. For example, “khat” and the aril of other species (Catha edulis, (Catha edulis), is used socially as a stimulant in Celastrus paniculatus, Sarawakodendron fila-

© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 0014-8962/11/7–810-0445 446 Feddes Repert., Weinheim 122 (2011) 7–8 mentosum) as derived from the exostome and the ovule starts (Fig. 1D). When the outer in- the funiculus. VAN DER PIJL (1972) concluded tegument is formed the nucellus is positioned at that Euonymus has an arillode, not an aril, but a right angle to the funiculus (Fig. 1E, F). During detailed study on the seed development is still further development the ovule gradually turns wanting. This study focuses on the different into an anatropous position (Fig. 1F). When stages of development and intends to show the finally the ovule is completely anatropous, the whole process of aril development in Celas- ovule is slightly exostomatic (Fig. 2A). At this traceae, and thus may help to understand the stage the exostome becomes thicker, and the evolution of this family and the order Celas- thickening meristem at first encircles the mi- trales sensu lato which is a loose assemblage of cropyle more or less completely. In subsequent probably not closely related taxa (SHISODE steps the micropylar thickening meristem of the 2011). It may furthermore be helpful and serve outer integument enlarges, and the meristem towards an evolutionary understanding of the step by step incorporates the entire funiculus formation of different types of fleshy seed until it finally encircles both the micropyle and appendages. the funiculus. The fleshy ring becomes thicker and thicker and develops a distinct rim towards the chalazal part of the ovule. In subsequent Material and methods steps this ring forms a fleshy duplicature covering the outer integument (Fig. 2B–E). In con- Three species, Celastrus orbiculatus THUNB., Euo- trast to a typical aril, this duplicature does not nymus europaeus L., and Euonymus planipes (KOEH- cover the micropyle, but leaves it free laterally NE) KOEHNE, were studied. The material was col- to the funiculus. The duplicature leaves the lected from one individual per species each, all chalazal region entirely free (Fig. 2F). samples were collected from the Botanical Garden Euonymus europaeus L. has whorls of of the Ruhr-Universität Bochum. Sampling was done nearly every week from April to July 2010. 4 perianth segments (Fig. 3A). The number of The fresh floral buds, flowers or young fruits were stamens equals the number of petals and car- fixed using FAA (formalin:acetic acid:ethyl alcohol pels (Fig. 3A, B). The gynoecium is tetramer- 70% = 5:5:90). In order to optimize infiltration of ous (Fig. 3B) and starts its formation when the the fixative, the material was kept in the fixative floral bud begins to close. The formation of the under moderate vacuum for at least 30 min. For two ovules per locule starts prior to the com- further storage, the FAA was replaced by 70% ethyl plete closure of the carpel. The formation of the alcohol after 2 days. ovule starts with the differentiation of an ob- Dissections were performed in ethyl alcohol tuse narrow primordium (Fig. 3C). When the under a ZEISS Stemi SV II dissection microscope. Dehydration for SEM studies was performed using incurvation of the ovule starts (Fig. 3D), the FDA (Dimethoxymethane) for 24 hours and sub- inner integument is initiated. When the outer sequent critical point drying with a BALZERS CPD integument is formed, the nucellus is curved 030. Dried parts were mounted on aluminium stubs approximately 90 degrees and is turned towards and sputter coated (BAL-TEC SCD 050, gold, the carpel margin (Fig. 3E). When the outer 200 sec, 42 mA). integument finally covers the inner integument and the nucellus, the ovule curves into a fully anatropous position (Figs. 3E, F; 4A). At this Results stage the micropyle points towards the proximal part of the gynoecium, the ovule is up- Celastrus orbiculatus THUNB. has a pentamer- right anatropous. When finally the ovule is ous imbricate perianth. The number of stamens completely anatropous, the ovule is slightly equals the number of petals (Fig. 1A). The exostomatic (Fig. 4B). At this stage the exo- gynoecium is trimerous (Fig. 1B) or sometimes stome becomes thicker, and the thickening tetramerous and starts its formation when the meristem at first encircles the micropyle more floral bud begins to close. The formation of the or less completely (Fig. 4C). In subsequent two ovules per locule starts prior to the com- steps the micropylar thickening meristem of the plete closure of the carpel (Fig. 1C). The inner outer integument enlarges, and the meristem integument is initiated when the incurvation of step by step incorporates the entire funiculus

Fig. 1 Celastrus orbiculatus I: Ovule, successive developmental stages. A — young flower with perianth and stamen primordia; B — trimerous gynoecium initiating; C — initiation of 2 ovules per locule; D — beginning development of the inner integument; E, F — formation of the outer integument and beginning ovule incurvation. c = calyx, f = funiculus, ii = inner integument, n = nucellus, o = ovule, oi = outer integument, p = petal, s = stamen

Fig. 2 Celastrus orbiculatus II: Caruncula, successive developmental stages. A — caruncula originating from the margin of the outer integument; B–D — enlargement of the caruncula and incorporation of the funiculus; E — beginning formation of the fleshy seed duplicature; F — mature seed. ca = caruncula, f = funiculus, ii = inner integument, m = micropyle, oi = outer integument

Fig. 3 Euonymus europaeus I: Ovule, successive developmental stages. A — preanthetic flower bud showing stamina and carpel primordia; B — young tetramerous gynoecium; C — ovule initiation; D — beginning development of the inner integument; E, F — formation of the outer integument and beginning ovule incurvation. c = carpel, f = funiculus, ii = inner integument, n = nucellus, o = ovule, oi = outer integument, s = stamen

Fig. 4 Euonymus europaeus II: Caruncula, successive developmental stages. A — micropyle formed by 2 integuments prior to caruncula formation; B — caruncula originating from the margin of the outer integument; C — incorporation of the funiculus in the formation of the caruncula; D — beginning formation of the fleshy duplicature; E — slightly older stage in seed development than in D (6 of 8, 2 seeds removed); F — mature seed. ca = caruncula, f = funiculus, ii = inner integument, m = micropyle, oi = outer integument

© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.feddes-journal.com X. ZHANG et al.: Aril development in Celastraceae 451 until it finally encircles both the micropyle and character is generally used together with others. the funiculus (Fig. 4B–D). The fleshy ring However, if the focus is on character evolution, becomes thicker and thicker and develops a processes starting in a different way and lead- distinct rim towards the chalazal part of the ing to different final stages should be distin- ovule. In subsequent steps this ring forms a guished carefully. GAERTNER (1788) was the fleshy duplicature covering the outer integu- first to describe the aril as an accessory in- ment (Fig. 4E, F). tegument. Botanists commonly use the term Euonymus planipes (KOEHNE) KOEHNE has aril for fleshy structures arising from the fu- a pentamerous imbricate perianth (Fig. 5A). niculus that enclose the ovule more or less The number of stamens equals the number of totally. Nevertheless, different terminologies petals (Fig. 5A, B). The gynoecium may be were used by different workers in different pentamerous or sometimes tetramerous, and research contexts (PLANCHON 1845; BAILLON starts its formation when the floral bud begins 1876; CORNER 1949, 1953, 1976; VAN DER PIJL to close (Fig. 5B). The formation of the two 1972; KAPIL et al. 1980). One of the defining ovules per locule starts prior to the complete properties of an aril is that it is generally fleshy, closure of the carpel. Unlike in Euonymus eu- but the most important property of an aril is ropaeus L. and Celastrus orbiculatus THUNB., how it is initiated and how it develops. A true the anatropous ovules are pendulous from the aril originates from the funiculus, and it can be top of the locule to the cup-shaped base of the described as a the third seed envelope of the gynoecium. When the incurvation of the ovule ovule (ENDRESS 2011). This organ can be starts the inner integument is initiated. When fleshy or hairy or can form wings, it may have the outer integument is formed the nucellus is its own vasculature, and sometimes produces a positioned at right angle to the funiculus mucilaginous pulp filling the locule. (Fig. 5C, D). During further development, the PLANCHON (1845) described the putative ovule gradually turns into its final anatropous arils of Celastrus scandens and Euonymus position (Fig. 5E). When finally the ovule is latifolius as arillodes – false arils – and cited completely anatropous, the ovule is slightly these false arils as derived from the exostome exostomatic, the micropyle nearly touches the of the outer integument rather than from the funiculus (Fig. 5E). At this stage the exostome funiculus. This is very similar to our results, becomes thicker, and the thickening meristem but not according to PLANCHON’s drawing. at first encircles the micropyle more or less Here the arillode originates from two distinct completely (Fig. 5F). In subsequent steps the primordia, one coming from the funiculus and micropylar thickening meristem of the outer the other from the exostome. Despite this un- integument enlarges, and the meristem step by usual development PLANCHON regarded it as an step incorporates the entire funiculus until it aril. In contrast to that, MIERS (1856) described finally encircles both the micropyle and the the same structure as originating exclusively funiculus (Fig. 6A, B). The fleshy ring be- from the funiculus and treats it as a normal aril. comes thicker and thicker and first develops a Our results on Euonymus europaeus, how- distinct circular rim towards the chalazal part of ever, are completely different from those of the ovule (Fig. 6C). Further growth of the du- MIERS, and from PLANCHON as well. One rea- plicature is asymmetric and leads to an oblique son is that MIERS did no complete developmen- shape of the duplicature (Fig. 6D). In subse- tal study, but analyzed only seeds and rather quent steps this ring forms a fleshy duplicature late stages of seed development in Euonymus covering the outer integument (Fig. 6E, F). europaeus (MIERS 1856). PFEIFFER (1891) described the arils of Celastrus, Euonymus, and Gymnosporia cassinoides as derived from the Discussion exostome and the hilum. CORNER (1976) described the aril of Euonymus glandulosus as The term aril is historically used in a rather entirely derived from the funiculus, and the aril broad sense. This may lead to an ambiguous of other species (Catha edulis, Celastrus pani- understanding of structures described as arils. culatus, Sarawakodendron filamentosum) as In diagnostics this might be acceptable as this derived from the exostome and the funiculus.

Fig. 5 Euonymus planipes I: Ovule, successive developmental stages. A — young flower with perianth and stamen primordia; B — pentamerous gynoecium initiating; C–E — formation of the outer integument and ovule incurvation; F — caruncula initiating from the margin of the outer integument. c = calyx, ca = caruncula, f = funiculus, ii = inner integument, n = nucellus, oi = outer integument, p = petal, s = stamen

Fig. 6 Euonymus planipes II: Caruncula, successive developmental stages. A, B — caruncula originating from the margin of the outer integument; C — older circular stage of the caruncula; D — asymmetric growth of the caruncula; E — mature stage of the seed, view from the hilum; F — entire mature seed, lateral view. ca = caruncula, f = funiculus; m = micropyle

This can be interpreted as diversity within the cula. In the system of APG III (2009), Celas- Celastraceae which also gives us a way to un- trales is only sister group of the orders of Mal- derstand the phylogeny of Celastraceae and the pighiales and Oxalidales. So the relationship order of Celastrales. VAN DER PIJL (1972) con- between the Celastraceae and Euphobiaceae is cluded that Euonymus has an arillode, not an closer than generally suggested. In the papers aril. VAN DER PIJL used the term arillode rather by SIMMONS & HEDIN (1999), SIMMONS as an ecological term than as a morphological (2004), and SIMMONS et al. (2008, 2011) the one. In his interpretation, the fleshy structures character aril is used without any clear descrip- which cover the seed and do not fulfil his defi- tion, they regard this structure in Celastraceae nition of an aril are all summarized as arillodes. as a true aril. In 2012, SIMMONS et al. con- This, however, is not really a solution as it cluded that “current intrageneric classifications shifts ambiguities to another term instead of of Euonymus are not completely natural and solving them in a meaningful way. Together require revision”. But prior to all tests, the with the aril there are four different types of development of the morphological characters of fleshy seed appendages including the sclero- this family should be clearly understood. testa in Magnolia and the strophiole and the caruncula in Euphorbiaceae. It is of some interest that all these secondary Conclusion fleshy appendages are either associated with the micropyle or with the funiculus or hilum. Celastrus orbiculatus has a pentamerous peri- Both the micropyle and the hilum are weak anth and five anthers, but a trimerous gy- points in the protective structure covering en- noecium. Euonymus europaeus is tetramerous dosperm and embryo. They can get closed by throughout the genus. However, the flower of mere shrinking of the surrounding tissues, but Euonymus planipes has five sepals, five petals, they can also close by active growth of adjacent and five carpels (seldom four carpels). The cells. This could be a preadaptation leading to developmental process in Celastrus orbicula- the fleshy appendages by stepwise enlarge- tus, Euonymus europaeus, and Euonymus ment. In orthotropous seeds, fleshy structures planipes is very similar. The visible difference originating from micropyle or funiculus are is the position of the ovule primordium. The clearly separate from each other, and usually ovule primordium of Euonymus planipes is on only one option is present. In anatropous seeds, the top of the placenta, so the micropyle is the two zones are close together. Only minor directed towards the style. meristem incorporation is necessary to build up The structure in Celastraceae called aril mixed forms in which micropylar appendages originates from the margin of the micropyle and funicular appendages merge. In the same which is here formed by the outer integument. way the fleshy parts originating only from the The funiculus is incorporated in the formation micropyle or only from the funiculus may in- of the fleshy part of the seed rather late in the corporate the complementary structure easily. primary morphogenesis. The relevant steps are It can be thus expected that structures that are so early that the development differs markedly very similar at maturity display rather differ- from what has to be expected for a true aril. ent developmental patterns. For diagnostic The pattern of development described here was purpose, it is not meaningful to distinguish hitherto unknown. According to this pattern, these structures in detail. Used in an evolution- the micropyle is not inside the aril, but at the ary context, fleshy seed appendages might base of the fleshy structure which is thus better supply even more information than actually referred to as a caruncula. used. Lophopyxis was regarded as a genus with an Acknowledgements arillate structure belonging to Celastraceae, but The Botanical Garden of the Ruhr-Universität Bo- according to SIMMONS (1999) Lophopyxis is chum has generously allowed us to collect the mate- better placed in Euphorbiaceae. This conclu- rial for the present and several other studies from its sion is also supported by our results saying that collections. The first author is especially grateful to the aril in Celastraceae should be called carun- Sabine Adler for her efforts in improving the manu-

© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.feddes-journal.com X. ZHANG et al.: Aril development in Celastraceae 455 script linguistically and structurally. He also grateful MIERS, J. 1856: Remarks on the nature of the outer acknowledges the financial support by the Chinese fleshy covering of the seed in the Clusiaceae, Scholarship Council. Magnoliaceae, and on the development of the raphe in general, under its various circumstances. – Trans. Linn. Soc. London 22: 81–96. PFEIFFER, A. 1891: Die Arillargebilde der Pflanzen- References samen. – Bot. Jahrb. Syst. 13: 492–540. VAN DER PIJL, L. 1972: Principles of dispersal in Angiosperm Phylogeny Group 2009: An update of higher plants. – Springer, Berlin, Heidelberg, the Angiosperm Phylogeny Group classification New York. for the orders and families of flowering plants: PLANCHON, M. J. E. 1845: Développements et APG III. – Bot. J. Linn. Soc. 161(2): 105–121. caractères des vrais et des faux arilles. – Ann. BAILLON, H. E. 1876: Sur l’origine du macis de la Sci. Nat. Bot., sér. 3, 3: 275–312. Muscade et des arilles en general. – Adansonia SHISODE, S. B. & D. A. PATIL 2011: Taxonomic and 11: 329–340. phylogenetic census of the Celastrales: A syn- CORNER, E. J. H. 1949: The Durian Theory or the thetic review. – Curr. Bot. 2(4): 36–43. Origin of the Modern Tree. – Ann. Bot. 13(4): SIMMONS, M. P. & J. P. HEDIN 1999: Relationships 367–414. and morphological character change among gen- CORNER, E. J. H. 1951: The leguminous seed. – era of Celastraceae sensu lato (including Hippo- Phytomorphology 1: 117–150. crateaceae). – Ann. Miss. Bot. Gard. 86: 723– CORNER, E. J. H. 1953: The Durian Theory extend- 757. ed I. – Phytomorphology 3: 465–476. SIMMONS, M. P. 2004: Celastraceae. In: KUBITZKI, CORNER, E. J. H. 1976: The seeds of dicotyledons K. (ed.): The families and genera of vascular Vol. 1 + 2. – Cambridge University Press. plants 6: 29–64. – Springer, Berlin. ENDRESS, P. K. 2011: Angiosperm ovules: diversity, SIMMONS, M. P., CAPPA, J. J., ARCHER, R. H., FORD, development, evolution. – Ann. Bot. 107: 1465– A. J., EICHSTEDT, D. & C. C. CLEVINGER 2008: 1489. Phylogeny of the Celastreae (Celastraceae) and GAERTNER, J. 1788: De fructibus et seminibus the relationships of Catha edulis (qat) inferred plantarum – Stuttgart. from morphological characters and nuclear and HOU, D. 1962: Celastraceae I. In: C. G. G. J. plastid genes. – Mol. Phylogenet. Evol. 48: 745– VAN STEENIS (ed.): Flora Malesiana 6(2): 227– 757. 291. SIMMONS, M. P., MCKENNA, M. J., BACON, C. D., HEYWOOD, V. H. 1993: Flowering Plants of the YAKOBSON, K., CAPPA, J. J., ARCHER, R. H. & J. World. – Oxford Univ. Press., New York. ANDREW 2012: Phylogeny of Celastraceae tribe KAPIL, R. N., BOR, J. & F. BOUMAN 1980: Seed Euonymeae inferred from morphological charac- appendages in Angiosperms I. – Bot. Jahrb. Syst. ters and nuclear and plastid genes. – Mol. Phylo- 101: 555–573. genet. Evol. 62(2): 9–20. KRIKORIAN, A. D. 1985: Growth mode and leaf VAN DER PIJL, L. 1972: Principles of dispersal in arrangement in Catha edulis (kat). – Econ. Bot. higher plants. – Springer, Berlin, Heidelberg, 39: 514–521. New York.