THE VO~L.66

BOTANICAL

REVIEW

APRIL-JUNE 2000

NO. 2

Clarification of the Carpel Number in Papaverales, Capparales, and Berberidaceae ClaudiaBriickner Humboldt University Institute of Biology Division of Systematic Botany and Arboretum Spi~thstrasse 80/81 D-12437 Berlin, Germany

I. Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III. Systematics of the Families . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Ordinal Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Intrafamiliar Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Papaveraceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Fumafiaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Capparaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Brassicaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Berberidaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Fruit Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Papaveraeeae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Fumariaeeae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Capparaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Brassicaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Berberidaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. What Is a Carpel? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. Theories to Explain Bivalvate Gynoecia in Papaverales and Capparales (Also Considering the Monovalvate Pistils of Berberidaceae) . . . . . . . . . . . . . . A. The nI Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. The nII Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. The 2n Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. The 3n Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Saunders's Theory of Carpel Polyrnorphism . . . . . . . . . . . . . . . . . . . . . . . . . . . . F. Other Theories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII. Morphological Support of Carpel Number Theories . . . . . . . . . . . . . . . . . . . . . . . . .

Copies of this issue [66(2)] may be purchased from the NYBG Press, The New York Botanical Garden, Bronx, NY 10458-5125, U.S.A. Please inquire as to prices.

TheBotanicalReview66(2): 155-307, April-June2000 O 2000 The New York Botanical Garden

155

157 157 160 160 162 162 163 163 164 164 165 165 167 169 171 174 176 180 180 182 183 186 186 187 188

156

VIII. IX. X. XI. XII. XIII.

THE B O T A N I C A L R E V I E W

A. Stigma Shape . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189 1. Papaveracear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189 2. Fumariacear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189 3. Capparaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192 4. Brassieaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192 5. Berberidaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .......... 194 6. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194 B. Zones o f Dehiscence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197 1. Papaveraceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200 2. Fumariaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200 3. Capparaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200 4. Bmssieaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 5. Berberidaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 6. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 C. Structure o f Placental Regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205 1. Papaveraceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205 2. Fumariaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207 3. Capparaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207 4. Brassieaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209 5. Berberidaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211 6. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211 D. Vascular Pattern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213 1. Papaveraceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213 2. Fumariaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218 3. Capparaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219 4. Brassicaeeae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220 5. Berberidaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222 6. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 E. O n t o g e n y o f the G y n o e c i u m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 1. Papaveraceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 2. Fumariaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237 3. Capparaeeae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239 4. B m s s i c a c e a e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241 5. Berberidaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242 6. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242 F. Teratology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246 1. Homoeotic Phenomena . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247 a. O y n o e c i a in Virescent Flowers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247 b. Petaloid Carpels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250 c. Carpelloid Stamens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250 i. Papaveraceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251 ii. Brassieaceae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253 d. Carpelloid Sepals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257 e. Carpelloid Ovules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257 2. Increase in Gynoeeial Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258 a. Additional Carpel Whorls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258 b. Increase in M e m b e r N u m b e r within the Regular Carpel Whorl . . . . . . . . 259 S u m m a r i z i n g Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269 Acknowledgments .................................................... 273 Literature Cited . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274 A p p e n d i x 1: Chronological List of Essential Literature Supporting the nI Theory o f Carpel N u m b e r . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300 A p p e n d i x 2: Chronological List of Literature Dealing with Oynoecial V i r e s c e n c e . . 302 A p p e n d i x 3: Chronological List o f Literature Dealing with Carpelloid Stamens . . . . 303

CARPEL NUMBER XlV. Appendix 4: Chronological List of Literature and Unpublished Observations Concerning Additional Encaptic Carpel Whorls . . . . . . . . . . . . . . . . . . . . . . . . . XV. Appendix 5: Chronological List of Literature and Unpublished Observations Dealing with Increased Carpel Numbers in the Gynoeeial Whorl . . . . . . . . . . . . Note: The Russian abstract appears at the end of this article.

157

304 305

I. Abstract For more than 170 years there has been a controversy about the organization of the siliqua, a fruit typical for the Brassicaceae and, in modified forms, also for members of Capparaceae, Papaveraceae, and Fumariaceae. Because in the Berberidaceae fruit forms resembling a "semi-siliqua" are produced, they are also controversial. A siliqua is typically furnished with two placental regions joined by a septum and dehiscing through detachment of two sterile valves. Modified forms lack a septum and have only one or more than two valves, or are indehiscent. The controversial issue is the number of carpels composing a siliqua, typical or modified. Aside from the fact that the nature and phylogeny of the angiosperm organ "carpel" are still insufficiently known and therefore speculative, carpel numbers of two, four, and six have been proposed for a bivalvate siliqua; moreover, an "acarpellate" state as an axis-derived structure has been postulated. Within the framework of these theories there are additional theories concerning the position, shape, and fertility or sterility of what are believed to be carpels. Each of these concepts is reviewed here, and its morphological basis is checked. Gynoecial features used as evidence of the manifold hypotheses are shape of the stigma, zones of dehiscence, structure of the placental regions, vascular pattern, ontogeny, and teratologieal transformations. They are discussed for each family and compared in the context of the conclusions derived from them. The result is that Robert Brown's (1817) classical theory, explaining the siliqua as a product of fusion of two transverse carpels with the valves being opercular structures and the septum formed of placental outgrowths, cannot be invalidated by any of the later theories. Stigmatic lobes should not a priori be equated with carpel tips, and their number is not a definite indication of carpel number. The zones of dehiscence are not carpel borders but secondary separation tissues within the carpel blade. Massive placental regions with complex venation need not be solid carpels. Number and course of vascular bundles may be interpreted in ontogenetic and functional terms, and the concept of vascular conservatism is unsound. Gynoecial growth centers must not uncritically be equated with carpel primordia. Terata, such as tetravalvate siliquae, are not atavisms. Thus, carpel numbers higher than those of placentae in the given gynoecium cannot be ascertained. The gynoeeium of Berberidaeeae is truly monomerous. The identical organization of the gynoecia in the families concerned demands their explanation by a single theory. Many textbooks, floras, and monographs should be revised from this point of view.

H. Introduction The Papaveraceae Juss., Fumariaceae DC., Brassicaceae Bumett (= Cruciferae Juss.), and the majority ofCapparaceae Juss. have superior paracarpous gynoecia with parietal placentation. These gynoecia produce various dehiscent or indehiscent fruit forms that are, however, predominantly a siliqua-like type. A siliqua s.str., the typical crueiferous fruit (Fig. 1), is characterized by a peculiar mode of dehiscence with two sterile sections of the pericarp (valves) separating from the persistent placental frame (replum). The latter bears the loosely attached seeds and is topped by the style and stigma remnants. The locule is two chambered, with a dissepimentum (septum) that covers the replum frame and, at fruit maturity, obtains a membra-

158

THE BOTANICAL REVIEW

Fig. 1. Schematic transection ofa Brassicaceae siliqua (in vascular bundles, xylem black and phloem white). neous consistence. Modified siliquae include absence of the septum, valve numbers greater than two, and a differing degree of valve detachment in incompletely dehiscent fruit forms. The organization (in the sense of Froebe and Classen-Bockhoff, 1994) of all these modified siliquae is essentially the same in the four families concerned. The dehiscent fruit forms of the Berberidaceae Juss. should also be considered, because their most important deviation from the bivalvate siliqua s.I. is a meristic one, for there is only one placental region and one valve formed, thus recalling "half a siliqua" (of. Endress, 1995). The composition of the fruits and especially of the siliquae (in a broad sense) has been debated for more than 170 years. The number and structure of the "congenitally fused" gynoecium units, the carpels, have been the subject of much discussion, as the many different concepts presented in the literature attest. The oldest and most widespread theory to explain the nature of the siliqua was first presented by Robert Brown in 1817. He and his successors proposed that bivalvate fruits with their placental regions in the median plane comprise two united carpels in transverse position. Other ideas--two median carpels, four carpels, six carpels, etc.--have been offered since then, sometimes defended in a long series of publications with concentrations around 1830, 1880, 1930, and 1980, or approximately every fifty years (Fig. 2). In the intervals, the supporters of the classical theory always hoped to give ultimate proof for the unsoundness of the deviating concepts, but, as Michel Gu6d~s (1966b: 59) somewhat sadly states, they appeared to be of"palingenetic" nature. At that time he announced a

CARPEL NUMBER

159

Fig, 2. Chronology of publications differing from Brown's (! 817) theory of carpel number in Papaveraceae, Fumariaceae, Capparaceae, Brassicaceae, and Berberidaceae. The length of the horizontal bars is relative to the number of publications by each author.

160

THE BOTANICAL REVIEW

historical review of carpel theories concerning Brassicaceae, but the manuscript was never published (A. Le Thomas, pers. comm.). Due to temporal separation by about a half-century the creators of"new" concepts could hardly contact and exchange ideas with their mental predecessors. They referred to the literature but often did not know the whole array of papers relevant to the subject. Moreover, most of the concepts applied to only one of the families and often only to selected members, so theoreticians frequently did not know that the same phenomenon had already been discussed in detail for representatives of another family. The regular reappearance of the same arguments for "new" concepts is explained by these facts. Number and position of stigmatic lobes, course of the zones of dehiscence, structure of the placental regions, vascular pattern, ontogeny of the gynoecium, and teratological phenomena have been repeatedly stressed to found "new" hypotheses that, in reality, had already been established before. These character sets have been examined in increasing detail and with rapidly improving equipment. Nevertheless, no essentially new aspects have been revealed. So the relevant data are sufficiently known. The aims of this article are to analyze and discuss the data in broad comparison to elucidate whether the siliqua needs an interpretation different from the classical one and thus to terminate cyclical revival of deviating concepts. It will also show that the gynoecia of the five families can be explained by the same theory. Most results of the author's practical studies preceding this widely theoretical paper have already been published (e.g., Briickner, 1982, 1992a), so there is no need to discuss material and methods. The present paper is, for the most part, a translation of a postdoctoral thesis (Brtickner, 1996b).

III. Systematics of the Families A. O R D I N A L CLASSIFICATION

In the past, Papaveraceae (including Fumariaceae), Capparaceae, Brassicaceae, and some small families (Resedaceae Gray, Tovariaceae Pax, Moringaceae Dumort.) were combined to form the order Rhoeadales sensu Wettstein (1935). This grouping was mainly based on floral and fruit structures. The Rhoeadales were directly derived from the Polycarpicae and were thought to be very closely related to the Berberidaceae (see Harms [1936] for the history of the classification). More recently, the name "Rhoeadales" was replaced by "Papaveralcs." The large order has often been subdivided into Papaverineae, Capparineae, and other monotypic suborders. It is part of the classifications by, for example, Melchior (1964), Tamura (1974), Benson (1979), and Tutin et al. (1993). Wettstein's opinion of a close relationship between the rhoeadalean families was, however, challenged by biochemical and serological data. The difference in chemical compounds of Papaveraceae and Brassicaceae was mentioned by Bernhardi as early as 1833 (p. 456). Whereas Papavcraccae and Fumariaceae are characterized by typical benzylisoquinoline alkaloids, there are exclusively glueosinolates in the Brassicaceae, Capparaeeae, and their satellite families (Blagowestschenski, 1955; Hegnauer, 1961, 1964, 1969, 1973; Gershenzon & Mabry, 1983). Furthermore, serological investigations revealed clear relationships between Papaveraceae and Fumariaceae on one hand and Brassicaceae, Capparaceae, Resedaceae, and Tovariaceae on the other (Frohne, 1962; Jensen, 1967; Kolbe, 1978). The four latter taxa also have in common the peculiar mircomorphological character of dilated cistemae in the endoplasmic reticulum (Behnke, 1977; Behnke & Barthlott, 1983). Differences in floral morphology were stressed by Merxmtiller and Leins (1967). Remarkable divergences in seed structure

CARPEL NUMBER

161

were mentioned by Lindley (1847), who assumed a close relationship among Papaveraceae, Fumariaceae, Berberidaceae, and Ranunculaceae but grouped Brassicaceae, Capparaceae, and Resedaceae with Cistaceae. In recent years macromolecular sequence data have been accumulated that show large incongruities in the papaveralean=capparaleanalliance (protein sequences: Martin & Dowd, 1991; rbcL sequences: Chase et al., 1993; 18S rDNA sequences: Soltis et al., 1997). All these facts support splitting Papaverales s.1. (= Rhoeadales) into Papaverales s.str. (Papaveraceae, Fumariaceae) and Capparales (Capparaceae, Brassicaceae, etc.). The first to recognize that separation was Takhtajan (1959), who placed the Papaverales s.str, in the subclass Ranunculidae. More recently, the order has been merged to the Ranunculales (Kubitzki et al., 1993). The Ranunculidae are the closest relatives of the Magnoliidae and are treated as a subgroup of the latter by many authors (Cronquist, 1981; Dahlgren, 1987; Frohne & Jensen, 1992). A serological comparison of seed proteins points to a very strong affinity between Papaverales and Magnoliaceae (Jensen & Greven, 1984). Takhtajan and other authors place Berberidaceae within Ranunculales. The family shares several morphological as well as chemical characteristics with Papaveraceae (cf. Himmel= baur, 1914). Bemhardi (1833) and Baillon (1861-1862) mentioned the remarkable similarity in habit that is displayed by certain rhizome geophytes of both families. The position of the Berberidaceae within the Ranunculidae (or Magnoliidae, respectively) is undisputed (cf. Drinnan et al., 1994; and several papers in Jensen & Kadereit, 1995). Thorne (1974, 1983, 1992a, 1992b) prefers the ordinal name Berberidales instead of Ranunculales and divides the order into suborders Berberidineae (with Ranunculaceae, Berberidaceae, and related families) and Papaverineae (Papaveraceae including Fumariaceae). The Berberidales of Huber (1991) are of nearly identical composition. Takhtajan's placement of the Capparales in the Dilleniidae (1959, 1987) is well supported by such characters as centrifugal development of multistaminate androecia, parietal placentation, biochemical constitution, and molecular data (Martin & Dowd, 1991; Chase et al., 1993). Probable ancestors are seen in primitive members of the Violales. Today, the separation of Papaverales and Capparales is widely accepted (e.g., see Dahlgren, 1977, 1983; Cronquist, 1981, 1988; Frohne & Jensen, 1992; Thorne, 1992b). Nevertheless, some authors still emphasize that both orders, though distinctly separate, are certainly related (e.g., Rohweder & Endress, 1983). It is supposed that the innovation of glucosinolates might have been accompanied by a compensational loss of the alkaloids. In this context it is worth mentioning that another family, Euphorbiaceae, although it is large and heterogeneous, contains benzylisoquinoline alkaloid-bearingtaxa, the Crotonoideae, as well as the glucosinolate= producing taxon Drypetes Vahi (see Frohne & Jensen, 1992). Comparativenodal anatomy sup= ports a close relationship between Papaveraceae (including Fumariaceae), Brassicaceae, Capparaceae, Resedaceae, and Berberidaceae, with the Ranunculaceae closely allied (Dormer, 1954; Ezelarab & Dormer, 1963, 1966). Johri et ai. (1992) state that Papaverales s.l. in the old sense form a rather homogeneous order with respect to embryological characters. The gap in floral morphology between Papaverales and Capparales is also not as large as previously thought (Endress, 1987; Karrer, 1991; Ronse Decraene & Smets, 1992, 1993a, 1993b, 1997a). Karrer suggests a possible alliance of Magnoliidae and Dilleniidae through Capparales. Other ideas of systematic positioning of the taxa appear less convincing. For example: Papaverales are close to Campanulales and Asterales (Nenukov, 1939); Papaverales, including Capparales, form a natural group with Batales and Fabales (Goldberg, 1986); Papaveraceae and Begoniaceae are allied (Ronse Decraene & Smets, 1990). Huber (1982, 1991) puts Berberidales, including Papaverales, in his group of"three-furrowed pollen taxa" with the Cen= trospermae and, probably, the monotypic Polygonales and Plumbaginales.A close relationship

162

THE BOTANICAL REVIEW

among these taxa is also indicated by the results of biochemical investigations (cf. Kubitzki et al., 1993). Huber, in his system, created a "First Principal Group" containing the "threefurrowed pollen taxa," the "oil-cell taxa," Illiciales, Monocotyledoneae, Piperales, and Nymphaeales, which corresponds to Magnoliidae s.l., Centrospermae, and Monocotyledoneae. The Capparales and their allies were placed in his "Second Principal Group," which consists of the remaining dicotyledonous taxa. It is not the aim of this article to supply any proof for the relationship between Papaverales and Capparales. However, a comparison of gynoecial organization Coauplan) is meaningful notwithstanding the objection of Carlquist, who wrote (1969:358), "In many of the crucifer gynoecium papers, we find repeated mentions of Papaveraceae, and comparisons of poppies with crucifers. In fact, Brassicaceae, Capparaceae, and Moringaceae are probably not at all related to Papaveraceae .... If this is true, comparisons between these two groups were largely a waste." B. INTRAFAMILIARCLASSIFICATION

1. Papaveraceae The Papaveraceae s.I. consist of the following taxa (the number of genera/species according to Kadereit, 1993): Subfamily Papaveroideae Tribus Papavereae (8/ca. 170) Tribus Chelidonieae (9/ca. 50) Tribus Eschscholzieae (3/ca. 18) Subfamily Platystemonoideae (3/5) Subfamily Hypecoideae (1/ca. 18) Subfamily Pteridophylloideae (1 / 1) This arrangement is based on the systems of Fedde (1909), Ernst (1962b), and Giinther (1975). However, the latter two authors exclude the monotypic Hypecoideae and Pteridophylloideae. These taxa have often been allied with the Fumariaceae or treated as families 0ftheir own (Brtickner, 1985; Lid6n, 1993; Hoot et al., 1997). Carpology as well as other characters, however, allow them to be regarded as papaveraceous subfamilies. Chelidonieae and Eschscholzieae are sometimes given subfamily rank (Ernst, 1962b; Kadereit, 1993; Kadereit et al., 1994, 1995; Blattner & Kadereit, 1995). The exclusion of the two genera Glaucium Miller and Dicranostigma Hook.f. & Thomson from Chelidonieae and their inclusion in Papavereae (Kadereit et al., 1994; Loconte et al., 1995, Shneyer et al., 1995) or separation as Glancioideae are not considered imperative by the present author (cf. Hoot et al., 1997). Neither a union of Platystemonoideae and Papaveroideae (Kadereit et al., 1994; Schwarzbach & Kadereit, 1995) nor a close alliance of Platystemonoideae and Canbya Parry ex A. Gray is supported by carpology. The Platystemonoideae is the only papaveraceous taxon that lacks the valvate fruit dehiscence so characteristic of the family and has simple sutural dehiscence instead. Regarding carpology, there would rather be support for family rank that was proposed by Smith (1972). This is, however, not in congruence with molecular data (cf. several papers in Jensen & Kadereit, 1995; Hoot et al., 1997). For practical reasons Smith (1972) raised Eschscholzieae to family rank, which is obviously unsound. Within the Papavereae, there are close relationships between Roemeria Medik. and Papaver sect. Argemonidium and between Stylomecon G. Taylor and Papaver sect. Californicum; and the large genus Papaver may well be a polyphyletic group (Kadereit et al., 1997).

CARPEL NUMBER

163

2. Fumariaceae In the past, the taxon was often treated as a subfamily of Papaveraceae. However, its representatives possess a set of unique floral (Ernst, 1961; Layka, 1976; Heslop-Harrison & Shivanna, 1977; Lid6n, 1993) and biochemical (Alston & Turner, 1963; Gershenzon & Mabry, 1983; Jensen, 1995) characters that justify a separate family. Lid6n (1993) points out that the Fumariaceae is a very natural family and provides the following classification (Lid6n, 1986 [as Fumarioideae]; 1993): Tribus Corydaleae (5/ca. 425) Tribus Fumarieae Subtribus Sarcocapninae (4/12) Subtribus Fumariinae (4/59) Subtribus Discocapninae (2/2) In a recent paper based on plastid DNA sequences as well as morphology, Lid~n et al. (1997) state that Dicentra spectabilis (L.) Lem. is a sister group to the rest of the family and exclude it from Dicentra Bemh., reestablishing the monotypic genus Lamprocapnos Endl. with L. spectabilis (L.) Fukuhara. (For currency, the old species name is applied in this paper.) Moreover, two species ofDicentra subgen. Chrysocapnos Engelm. are given generic rank as Ehrendorferia Fukuhara et Lid6n (E. "chrysantha (Hook. et Am.) Rylander: Dicentra chrysantha Hook. et Am., and E. ochroleuca (Engelm.) Fukuhara: Dicentra ochroleuca Engelm.). Another rather isolated species, D. macrantha Oliv., is treated as the only member of the newly coined genus Ichtyoselmis Lid6n et Fukuhara as/. macrantha (Oliv.) Lid~n. Cysticapnos Miller (five species), of questionable position (Lid6n, 1986), most probably belongs to the tribe Fumarieae.

3. Capparaceae The Capparaceae (-- Capparidaceae) consist of 42-46 genera and approximately 800 species. Thome (1983) expected 930 species. Takhtajan (1987) gives the following division (supplements after Melchior, 1964; and Hutchinson, 1967): Subfamily Capparoideae Tribus Cappareae (l 6/ca. 330) Tribus Maerueae (l/90) Tribus Cadabeae (6/ca. 72) Tribus Stixeae (l/ca. 15) Tribus Apophylleae (2/10) Subfamily Pentadiplandroideae (l/l) Subfamily Koeberlinioideae (1/ l) Subfamily Cleomoideae Tribus Cleomeae (11/ca. 260) Tribus Podandrogyneae (l/ca. 20) Tribus Oxystylideae (2/ca. 1l) The position of some small taxa is uncertain. The monotypic genus Pentadiplandra Baill. is sometimes treated as a family (e.g., Hutchinson, 1967; Goldberg, 1986), possibly related to Celastraceae, whereas Farr et al. (1979) allied it with Tiliaceae. The monotypic Koeberlinia Zucc. is closer to Capparaceae than to Celastraceae; family rank is, however, recommended

164

THE BOTANICAL REVIEW

by Mehta and Moseley (1981). Thorne (1983, 1992a,b) relates the taxon with the Tovariaceae, which he treats as a capparaceous subfamily, Tovarioideae, following D'Arcy (1979). Takhtajan (1987) removes the monotypic Emblingia F.Muell from the Capparaceae. There are palynological congruences with Polygalaceae (Erdtman, 1952), and a position near or within Sapindaceae is also considered possible (Keighery, 1981; Thorne, 1992a,b). The transfer of the monotypic Dipterygium Decne. to the Brassicaceae has not been generally accepted (Takhtajan, 1987; Thorne, 1992a,b). The monotypie Stefaninia Chiov. is included in Reseda L. (Resedaceae) by Hutchinson (1967). According to Hutchinson (1967), Calyptrotheca Gilg (two species) belongs to Portulacaceae, whereas Thorne (1983) favored a position within Cleomoideae. Hutchinson (1967) transfers Physena Noronha ex Thouars (two species) to Passifloraceae, whereas Takhtajan (1987) and Cronquist (1988) plead for family rank and suggest ties to either Sapindaceae or Urticales; Farr et al. (1979) question the capparalean alliance as well. Hutchinson (1967) proposed the exclusion of Cleomoideae from Capparaceae and its transfer to Tovariaceae. Aleykutty and Inamdar (1978) favored family rank for Cleomoideae, but this concept has not been generally accepted (see Anuradha et al., 1988). Polanisia Raf. is often combined with Cleome L., although several characters of floral morphology of the genera are clearly distinct. Gynandropsis DC. and Isomeris Nutt. ex Torr. et A. Gray have also been included in Cleome (see Bailey & Bailey, 1976; Everett, 1981).

4. Brassicaceae The crucifers are a voluminous family that comprises, according to AI-Shehbaz (1984), about 340 genera and more than 3,350 species. There is probably a common ancestor with Capparaeeae-Cleomoideae. (For monophyly, the Angiosperm Phylogeny Group [1998] proposed combining Capparaceae and Brassicaceae under the name of the latter, so that Capparales would become Brassicales.) The most extensive family classification is Schulz's (1936), listing 19 tribes, which are reduced to 15 by Janchen (1942). A1-Shehbaz (1984) accepts the following 13 tribes: Thelypodieae (11/ca. 110); Pringleeae (1/1); Cremolobeae (2/36); Sisymbrieae (ca. 70/ca. 400); Hesperideae (ca. 40/500); Arabideae (36/570); Alysseae (41/ca. 650); Lepidieae (60/600); Brassiceae (ca. 52/ca. 230); Chamireae (1/1); Schizopetaleae (3/13); Stenopetaleae (1/8); and Heliophileae (4/77). Owing to the multitude of species, this system will probably contain some unnatural taxa, but it is nevertheless a useful basis for further studies.

5. Berberidaceae The family consists of 16 genera with more than 600 species and is subdivided by Loconte (1993) as follows: Subfamily Nandinoideae (1/ l) Subfamily Berberidoideae Tribus Leonticeae (3/ca. 17) Subtribus Berberidinae (3/600) Subtribus Epimediinae (8-9/ca. 45)

Nandina domestica Thunb., the only species of Nandinoideae, differs considerably from the remaining taxa in several reproductive characters as well as in chromosome number (Kumazawa, 1938a, 1938b; Comer, 1976; Nowicke and Skvarla, 1981). Several authors (Hutch-

CARPEL NUMBER

165

inson, 1959, 1973; Tamura, 1974; Thorne, 1983; Goldberg, 1986; Takhtajan, 1987; Huber, 1991) treat it as a monotypic family. However, the separation from Berberidaeeae is not supported by comparative serology (Jensen, 1974), morphology of inflorescence and flower (Terabayashi, 1983c, 1985b; Nickol, 1995), or molecular data (Adachi et al., 1995; Kim & Jansen, 1995). For similar reasons, the rank of a monotypie family for Ranzaniajaponica (T.Itr) TAt6, established by Takhtajan (1994), seems still less founded (of. papers on Berberidaceae systematies in Jensen & Kadereit, 1995). The monotypic genus shares some essential morphological and molecular features with Berberis L. and Mahonia Nutt. Hutchinson (1959, 1973) circumscribed the Berberidaceae s.str. (Berberidales) as restricted to the woody genera Berberis and Mahonia. The herbaceous taxa are combined as Podophyllaceae and put into Ranales. Takhtajan (1969) reduces Podophyllaeeae to Podophyllum L. and Diphylleia Michx. Goldberg (1986) adopts this concept. However, splitting of the Berberidaceae s.I. is not sufficiently founded (see Meacham, 1980; Terabayashi, 1985b; Loeonte & Estes, 1989b; and several papers in Jensen & Kadereit, 1995). The monotypic genera deffersonia Barton and Plagiorhegma Maxim. are considered congeneric by several authors. Their combination under the name of Jeffersonia seems to be justified. The relationship between the two voluminous genera Berberis and Mahonia is also very close. A combination (made, for example, in the "Flora of North America" [Flora of North America Editorial Committee, 1997]) would, however, create a giant genus with approximately 500 species. IV. Fruit Forms A multitude of terms, several of which date back to Linnaeus and his contemporaries, have been used to describe the vast number of fruit forms. Spjut (1994) gives a comprehensive compilation and redefinition of them. However, these terms serve descriptive purposes only and, at least for the families of concern here, do not reflect the same gynoecial organization that underlies the manifold functional forms of fruits. For the sake of completeness they are included in this section, even though, as will be shown in the next section, discriminating between, for example, "siliqua" (bivalvate pod with septum and stigmatic lobes standing over the placental regions) and "ceratium" (bi- to polyvalvate pod without septum and stigmatic lobes alternating with the placental regions) does not make much sense. A. PAPAVERACEAE The typical bivaivate siliquae s,l. (Ceratium, Figs. 3.1-3.4 & 3.12) are produced in the Chelidonieae (exception: Stylophorum diphyllum (Michx.) Nutt. with 3-5 valves), Esehscholzieae, Pteridophyllum Siebold et Zucc., and Hypecoum erectum L. They are many seeded except for the one-seeded pods of Bocconia L. In Chelidonium L., Hylomecon Maxim., and Sanguinaria L. the xerochastic dehiscence starts unspecifically along the sides and continues toward the base and top of the fruit. Acropetally dehiscing fruits are formed in Pteridophyllum, Bocconia, Glauciump. p., Eschscholzia Cham., Hunnemannia Sweet, and Dendromecon Benth. Basipetal dehiscence characterizes the fruits ofDicranostigma, Stylophorum Nutt., Glauciump. p., and Macleaya cordata (Willd.) R.Br. The three Esehscholzieae genera dehisce explosively; for the mechanism see Eichholz (1886) and Berg (1966). In completely dehiscent fruits the valves detach from the replum (Chelidonium, Hylomecon, Sanguinaria, Bocconia); otherwise, the valve tips or bases remain connected with the fruit tip or base (Pteridophyllum, Glaucium, Dicranostigma, Stylophorum, Eschscholzieae).

166

THE BOTANICAL REVIEW

Fig. 3. Fruit forms in Papaveraceae (redrawn; relative size not to scale). 1. Sanguinaria canadensis L., dehisced siliqua. 2. Chelidonium majus L., siliqua dehiscing acropetally (A) and basipetally (B). 3. Glaucium corniculatum (L.) Rudolph, siliqua with basipetal incomplete dehiscence. 4. Dendromecon rigida Benth., siliqua with explosive acropetal dehiscence. 5. Hypecoumprocumbens L., indehiscent, transversely splitting Iomentum. 6. Meconopsis cambrica (L.) Vig., polyvalvate capsule dehiscent in the upper third. 7. Canbyacandida Parry, trivalvate capsule. 8. Stylomecon heterophylla(Benth.) (3. Taylor, polyvalvato capsule with style-bearing cartilaginous roof. 9. Papaver rhoeas L., porate capsule with stigmatic disc. 10. Hesperomeconlinearis (Benth.) Greene, suturally dehiscing capsule without replum. 11. Platystemon californicus Benth., choricarpoid fruit, fruitlets disintegrating into single-seededjoints. 12. Schematic transection of bivalvate ovary with parietal placentae in the median plane and valves in the transverse plane. Sources: 1,5, and 9 redrawn from Wettstein, 1935; 2 and i 2, from Rohweder & Endress, 1983; 3 and 6, from Castroviejo et al., 1986; 4, from Fedde, 1936; 7 and 10, from (3unn & Seldin, 1976; 8, from (3onnermann, 1979; I l, from Thorne, 1974.

CARPEL NUMBER

167

The one-seeded fruits ofMacleaya microcarpa (Maxim.) Fedde are indehiscent and should be designated "utriculus," according to Spjut (1994). The style and stigma persist on the more or less broad replum. In Glaucium pods, there is a prominent false septum formed by outgrowths of the two placentae. It occupies the locule nearly completely, thereby enclosing the seeds. The Papavereae have polyvalvate capsules (ceratium type). However, only the valve tips separate hasipetally (extending deeper than the middle of the fruit in Arctomecon Torr. et Fr6m., Canbya as in Fig. 3.7, Roemeria, andMeconopsis villosa [Hook.f.] G.Taylor). In some species ofMeconopsis Vig., as well as in Stylomecon and Papaver L. (Fig. 3.9), the seeds are released by small pores. The cultivated Papaver somniferum L. var. album DC. was selected for indehiscent capsules to harvest the seeds. Stopp (1950) considered the porate capsule to be the most primitive fruit type in the family which, by continuous enlargement of the valves, gave rise to the fenestrate capsule (Fensterkapsel) with completely separating valves. In most of the Papavereae members the styles are obscure or absent; well-developed styles are observed only in some Meconopsis species and Stylomecon. Special capsule structures are the cartilaginous disclike projections over the valve tips of some Asiatic species of Meconopsis (Ernst, 1962a), the cartilaginous disclike roof that bears the style in Stylomecon (Fig. 3.8), and the lignified, conical to fiat stigmatic disc characteristic of Papaver (Fig. 3.9). The placentae consist of loose, spongy parenchyma and, except in Argemone L., are intruded deeply into the Iocule. In Romneya Harv. the placentae are connected with a central column of tissue, thus producing a completely septate gynoecium and fruit. Some species of Papaver have distinctly stipitate fruits. The members of the Hypecoideae, except Hypecoum ereetum and H. lactiflorum (Kar. et Kir.) A. E. Dahl, have articulated indehiscent fruits with two placental regions (Fig. 3.5, bilomentum type) that, by transverse abscission, disintegrate into one-seeded fragments. In H. parviflorum Kar. et Kir. the segments slip from the epidermis that remains intact as a membraneous sleeve. Cases of synaptospermy have also been described in the genus (cf. Dahl, 1989, 1990). As mentioned above, the carpological characters of the Platystemonoideae differ remarkably from those in the rest of the family. Their three- to polycarpellate capsules (Figs. 3.10 & 3.11 ) dehisce basipetally along the sutures without any replum formation. Hoot et al. (1997) regard this type as derived within the family. The fruits ofPlatystemon Benth. appear to he subchoricarpous, as the torulose individual carpels have a postgenitally closed ventral slit and separate at maturity. Gynoecium inception is, however, syncarpous. Several ovules develop into naked seeds in the central cavity surrounded by the carpels, whereas the majority of ovules, by fusion of the carpellary margins, are enclosed in the carpel Iocules. The naked seeds are released by longitudinal separation of the carpels. Moreover, each carpel breaks transversely into single-seeded joints (Fig. 3.11, Iomentarium type). Hannan (1980) referred to the enclosed seeds of Platystemon germinating conspicuously later than the naked ones as "heteromericarpy" According to his study, only a few natural populations produce naked seeds beside enclosed seeds. The present author, however, found the phenomenon to be rather frequent among cultivated plants of garden origin. B. FUMARIACEAE All fumariaceous fruits have two parietal placentae. In the Corydaleae, siliquae (ceratium type) with a persistent style and stigma are produced (Figs. 4.1-4.4). At maturity the pericarp is still somewhat fleshy, and the dehisced fruit quickly withers. In Dicentra Berne andAdlu-

168

THE BOTANICAL REVIEW

Fig. 4. Fruit forms in Fumariaceae (redrawn; relative size not to scale). 1. Corydalispallidiflora N. Busch, capsule seen from placental plane. 2. C. intermedia (L.) M~at, dehisced capsule. 3. C. caseana Gray, explosively dehisced capsule. 4. Dicentra ochroleuca Engelm., valve view of dehisced capsule. 5. Discocapnos mundtii Chain. et Schltdl., flattened nut. 6. Cysticapnos vesicaria (L.) Fedde, bladdery fruits. 7. Cryptocapnos chasmophyticus Reich.ill., nut (or bivalvate capsule?) in transverse (A) and median view (B). 8. Platycapnos tenuiloba Pomel, nut: A. Gynoecium with the style not yet dropped; B. Transverse view; C. Median view from placental plane; D. Opened fruit, seed enclosed by detached inner epidermis. 9. Ceratocapnos heterocarpa Durieu, heterocarpy: A. Gynoeciumwith the style not yet dropped; B. Dimorphic fruits, upper bivalvate two-seeded capsules and lower one-seeded nuts; C. Longisection of dehiscent fruit morph; D. Longisection of indehiscent fruit morph. 10. Fumaria mirabilis Pugsley, nut: A. Transverse view; B. Median view; C. Top view. 11. Rupicapnos muricaria Pomel, beaked nut. 12. Schematic transection ofbivalvate ovary with parietal placentae in the median plane and valves in the transverse plane. Sources: 2 redrawn from Hegi, 1986; 3, from Ownbey, 1947; 4, from Berg, 1969; 5 and 6, from Fedde, 1936; 7, 10, and 11, from Lid~n, 1986; 8 and 9, from Castroviejo et al., 1986; 12, from Rohweder & Endress, 1983.

CARPEL NUMBER

169

mia DC. the capsules remain more or less enclosed by the marcescent corolla. The fruits of Dactylicapnos Wall. sect. Minicalcara (KMnh) Lid6n are torulose. The valves separate completely or detach laterally in the upper part of the fruits. The tendency of the valve tips to remain attached to the replum is strong (Figs. 4.2-4.4). In several sections of the large genus Corydalis DC. explosive dehiscence is observed (Fig. 4.3; for an explanation of the relatively unspecialized turgor mechanism, see Schneider, 1935). In sect. Oocapnos Popov ex Wendelbo indehiscent globose bladder fruits (utricuhs type) are produced. The fruits of Fumarieae are characterized by deciduous styles. Capsules are rare. In Sarcocapninae, Pseudofumaria Medik. and Ceratocapnos claviculata (L.) Lid6n have few-seeded siliquae (ceratium type). Two other species of Ceratocapnos Durieu display heterocarpy (Ruiz de Clavijo, 1994). In the infrutescence, the lower fruits are one-seeded achenes with a strongly ribbed pericarp, whereas in the upper part there are two-seeded fruits that have long beaks and, after shedding, may tardily dehisce with two valves (Figs. 4.9A-4.9D). Sarcocapnos DC. has one- or two-seeded indehiscent fruits with conspicuous ribs and an apical appendage with two germination pores on each side. The indehiscent fruits of Platycapnos (DC.) Berne are ovate and flattened with thickened margins; they contain a single seed (Figs. 4.8A-4.8D). Within Fumariinae, one-seeded indehiscent fruits are produced in Fumariola Korsh. (subcylindrical with a flattened top and four small mucros), FumariaL. (roundish with two apical pits and a smooth to rugose surface; Figs. 4.10A-4.10C), Cryptocapnos Rech.fi/. (elliptical with excavate beak, perhaps dehiscent?; Figs. 4.7A & 4.7B; cf. Rechinger, 1967), Rupicapnos Pomel (roundish with short acute beak and rugose surface; Fig. 4.11), Discocapnos Chain. et Schltdl. (discoid with the style often persisting; Fig. 4.5), and Trigonocapnos Schltr. (asymmetrical, triangular in cross-section). All these fruit forms belong to the achena type. Cysticapnos (including Phacocapnos Bernh.) has many-seeded fruits that are vesicular in C. vesicaria (L.) Fedde (utriculus type, Fig. 4.6). These vesicular fruits consist of spongy parenchyma and have broad placental regions. The small seeds are enclosed in an "inner seed container" (Medicus, 1789; Borckhausen, 1797; Heinricher, 1925) that is formed by the detached inner fruit epidermis and the placental tissue. The tendency of the inner fruit epidermis (and sometimes also the adjoining parenchyma layer) to produce loose folds or to separate more or less completely is seen in several taxa of the family (e.g., in Corydalis [Brtickner, 1993] and Sarcocapninae [Fig. 4.8D; Lid6n, 1986; Briickner, 1992a]). Myrmecochorous dispersal is described for the fruits of some Ceratocapnos, Sarcocapnos, Platycapnos, and Fumaria taxa (Fukuhara & Lid~n, 1995). The pedicels of the chasmophytic genera Rupicapnos, Sarcocapnos, and Cryptocapnos display negative phototropism and bury the fruits at the base of the mother plant or in rock crevices (geocarpy; Wendelbo, 1974; Lid6n, 1986). C. CAPPARACEAE The Cleomoideae--Cleomeae have siliquae of the ceratium type (Fig. 5.14). Two transversal valves detach from the median seed-bearing replum (Figs. 5.1 & 5.2). Explosive dehiscence is reported in Cleome torticarpa H. Iltis et T. Ruiz Zapata (Iltis & Ruiz Zapata, 1997). The style and stigma may persist but are often dropped, giving rise to an apically open, bifurcated replum (Fig. 5.2B). Haptocarpum bahiense Ule is able to climb with the help of the hooked replum remainders (Pax & Hoffmann, 1936). IRis (1957) described a series of fruit reductions in the western North American Cleomoideae. The basic type, the many-seeded siliqua, is produced only in Cleome. Cleomella DC. has siliculae, and in some species the valves are convex to conical and more or less torulose

170

THE BOTANICAL REVIEW

Fig. 5. Fruit forms in Capparaceae (redrawn, relative size not to scale). 1. Cleome spinosa Jacq., dehisced siliqua. 2. C. ornithopodioides L.: A. Siliqua; B. Replum after abscission of valves and seeds. 3. C. isomeris, dehiscent bladdery fruit with two valves and a replum. 4. Buhsia trinervia (DC.) Stapf, bladder/fruit without rcplum. 5. Cadaba farinosa Forsk., berries. 6. Thylachium paradoxum Giig, berry. 7. Buchholzia coriacea Engl., berry. 8. Maerua angolensis DC., torulose berries. 9. Stixisfasciculata (King) Gagnep.: A. Gynoecium; B. Longisection of ovary. 10. Capparis spinosa L., berry. 11. Cladostemon kirkii (Oily.) Pax et Giig.: A. Gynoecium; B. Transection of ovary. 12. Cleomella obtusifolia Torr. et Fr6m., longisection of silicula with conical valves. 13. Oxystylis lutea Torr. et Fr6m., iongisection of schizocarpium. 14. Schematic transection ofbivalvate ovary with parietal placentae in the median plane and valves in the transverse plane. Sources: I and 10 redrawn from Baillon, 1872; 2, from Hutchinson, 1969; 3 and 4, from Briquet, 1914; 5-9 and 1!, from Pax & Hoffmann, 1936, 12 and 13, from Iltis, 1957; 14, from Rohweder & Endress, 1983.

CARPEL NUMBER

171

(Fig. 5.12), thus often enclosing a seed in the tip when falling off. In Wislizenia Engelm. the obovoid to obconical valves enclose the one or two seeds permanently, for they are too large to pass through the narrow valve opening. This fruit type may be termed a schizocarp. It is more perfectly elaborated in Oxystylis Torr. et Fr6m. (Fig. 5.13), in which the modified replum forms an "inverted V-shaped" tube with down-facing open ends. Attached to them are the obovoid valves that tightly enclose the single seed and act like a second seed coat. Cleome isomeris Greene (= Isomeris arborea Nutt.) has pear-shaped bladder fruits (utriculus type, Fig. 5.3) that dehisce tardily but produce a replum. The rather similar fruits of Buhsia Bunge (Fig. 5.4) dehisce without forming a replum (Briquet, 1914). Some species of Podandrogyne Ducke are also said to have no replum (Pax & Hoffmann, 1936). In the Capparoideae, two- and multichambered fruits are formed that are mainly berries (bacca or amphisarcum types, Figs. 5.5-5.8 & 5.10). They may attain the size of a mandarin or pear; some are edible (e.g., Crateva religiosa G. Forst. and Capparis subg. Calyptrocalyx Eichler [Iltis et al., 1996]). In Oceanopapaver Guillaumin the two intruding placentae fuse postgenitally, thus dividing the locule completely. The ripe fruit breaks transversely into oneseeded segments (bilomentum type; Schmid et al., 1984). Postgenital fusion of two to several placentae also occurs in the gynoecia ofCapparis L., giving rise to septate indehiscent fruits (bacca and utriculus types; of. Leins & Metzenauer, 1979). Similar secondary septation is seen in the fruits of Steriphoma Spreng. (2 placentae), Morisonia L. (4 placentae), Maerua Forssk. (2--4 placentae, with torulose fruits that articulate transversely into one-seeded pans; a bilomentum), and Courbonia Brongn. (2 placentae, partially connate). Primary septa are formed in the three- or four-chambered gynoecia ofStixis Lour.; the placentation is axile (Fig. 5.9), and the fruits become drupes. Other genera with divided locules and seeds in axile positions are Koeberlinia (2 placentae), Forchhammeria Liebm. (2 placentae), Titania Pierre (4 placentae), Neothorelia Gagnep. (3 placentae), and Pentadiplandra (4 or 5 placentae). The gynoecia and fruits of many species are still insufficiently known. The majority of the capparaceous gynoecia have long gynophores that elongate further during fruit development and may become lignified. D. BRASSICACEAE The remarkable diversity of fruit forms, being an important tool for systematic treatment of the family, can be derived from the basic siliqua (Fig. 6. l) and silicula (Fig. 6.12) types. Discrimination between these types is given only by fruit dimension (the siliqua is more than three times longer than wide, whereas the silicula is three times longer than wide at most). The fruits are two chambered with a papery septum (Figs. 6.9 & 6.19). As for the siliculae, latisept and angustisept forms are discerned. Latisept forms appear to be compressed from the dorsal sides of the valves (parallel to the septum) and thus have a wide septum (Fig. 6.9). Angustisept forms are flattened from the valve edges (rectangular to the septum) and have a narrow septum (Fig. 6.11). Fruits appear to be one chambered when seeds are formed in only one of the chambers and, by enlargement, press the septum to the fruit wall or eventually tear it. The development of the septum may also be more or less inhibited (Harming, 1901). According to Zohary (1948b), the most primitive fruit type in the family is the shortstyled, many- to few-seeded, dehiscent siliqua (including silicula) with valves extending over the entire length of the fruit (e.g., Sisymbrium L. as in Fig. 6.1 and Erysimura L.). The direction of dehiscence is mainly acropetal; less often basipetal; the valves detach completely from the replum. This basic type, termed "valvoid" (Zohary, 1948b), gave rise to several specialized forms that probably developed independently several times:

172

THE BOTANICAL REVIEW

Fig. 6. Fruit forms in Brassicaceae (redrawn; relative size not to scale). 1. Sisymbrium officinale (L.) Scop., dehisced siliqua. 2, Cardamine impatiens L., explosively dehiscing siliqua. 3. Stanleya pinnata (Pursh) Britton, siliqua borne on gynophore. 4. Sinapis alba L., heteromericarpous siliqua (valvonucamentoid) with one- or two-seeded stylar segment. 5. Fezia pterocarpa Pit., heteromericarpous and heterospermous siliqua (valvo-nucamentoid), stylar segment one seeded. 6. Erucaria cakiloidea, heteromericarpous siliqua (valvo-lomentoid), stylar segment disintegrates into one-seeded joints. 7. Raphanus raphanistrum, heteromericarpous fruit (lomentoid) with reduced valvar segment. 8. $inapis aucheri (Boiss.) O. E. Schulz, heterocarpy: A. Lower upright siliqua; B. Upper pendent siliqua; both heteromericarpous. 9. Lunaria annua L., replum oflatisept silicula with seeds attached to it. 10. Coluteocarpus vesicarla (L.) Hoimboe, latisept bladdery silicula. 11. Capsella bursa-pastoris (L.) Medik., dehisced angustisept silicula. 12. Thlaspi arvense L., angustisept silicula with winged valves. 13. Boreava orientalis Jaub. et Spach, winged nucamentoid. 14. Loxoptera stenophylla (Muschl.) O.E.Schulz: A. Schizocarpoid silicula splitting into two meriearps; B. Longisection of a meriearp. 15. Pugionium cornutum (L.) Gaertn., angustisept nucamentoid. 16. Psychine stylosa Desf., heteromericarpous silicula with sterile stylar segment. 17. Rapistrum rugosum (L.) All., heteromericarpous silicula (valvo-nucamentoid). 18. Rorippa barbareifolia (= Tetrapoma barbareaefolium): A. Dehiscing tetravalvate silicula; B. Transection of incompletely septate ovary. 19. Schematic transection of bivalvate septate ovary with parietal placentae in the median plane and valves in the transverse plane. Sources: 1, 2, and 9 redrawn from Hegi, 1986; 3, 5-8, and 14-17, from Schulz, 1936; 4 and 12, from Wettstein, 1935; 10 and 13, from Davis, 1965; 11, from Baillon, 1872; 18, from Le Maout & Decaisne, 1868; 19, from Rohweder & Endress, ! 983.

CARPEL NUMBER

173

Folliculoid: A rare variety of the valvoid type with unilateral dehiscence (Leptaleumfilifolium (Willd.) DC.). Nucamentoid: One- or few-seeded indehiscent fruits act as a dispersal unit (siliculaderived nut, very heterogeneous in origin; e.g., Boreava Jaub. et Spach as in Fig. 6.13; Pugionium Gaertn. as in Fig. 6.15). Schizocarpoid: The fruit splits along the placental regions into one- or two-seeded closed mericarps (e.g., Loxoptera O. E. Schulz in Fig. 6.14). The following fruit types deviate from the former four in being heteromericarpous in the sense of Delpino (1894) or heteroarthrocarpous in the sense of Voytenko (1968), respectively. True heteroarthrocarpy is restricted to the tribe Brassiceae. The fruits consist of a lower seed-bearing part opening by two valves, the valvar segment, and an upper indehiscent part, the stylar segment (or beak), that may also contain one or few seeds. The stylar segment is not identical to the style (Brassai, 1838), for the ovary and style are not sharply delimited (Zohary, 1948b). Thus, the upper part of the gynoecium, though stylelike in appearance, may well have ovules within it and be part of the ovary. The term "seed-bearing style" that is sometimes found in the literature (e.g., Thellung, 1919) is a misnomer. A discussion of suitable versus misleading terminology concerning heteroarthrocarpous fruits is given by Appel (1999). Valvo-nucamentoid: valvar segment well developed; many seeded, indehiscent stylar segment with one to few seeds (e.g., Sinapis L. as in Fig. 6.4; Fezia Pit. as in Fig. 6.5). A special form of this type is a fruit with one or two seeds in each part and the valvar segment opening tardily or not at all (e.g., Rapistrum Crantz as in Fig. 6.7). Valvo-lomentoid: valvar segment well developed; stylar segment constricted between the seeds and breaking transversely (e.g., Erucaria cakiloidea (DC.) O. E. Schulz as in Fig. 6.6). Lomentoid: valvar segment extremely reduced and mostly indehiscent, containing few or no seeds; stylar segment well developed, many seeded, and constricted between the seeds (e.g., Raphanus raphanistrum L. as in Fig. 6.7). Spjut (1994) designates the indehiscent fruit forms "achena," "utriculus," "carcerculus," and "bilomentum.". As noted above, these historical terms are unspecific and do not explain the peculiar structures and the evolutionary connections among the types. Several taxa display heterocarpy (dimorphic fruits on the same plant; cf. Voytenko, 1968; Fig. 6.8). This phenomenon may concern shape only, such as winged and wingless fruits (Isatis boissieriana Rechb. f.), or the dehiscence pattern within an infrutescence. In the upper infrutescence dehiscent fruits may be formed, whereas in the lower part fruits tardily and untypically dehisce (e.g., in a folliculoid manner) or remain closed. Differences in seed number are also seen in fruits ofheterocarpous plants. Amphicarpous taxa produce aerial dehiscent siliquae and subterranean one-seeded nuts (e.g., Cardamine chenopodiifolia Pers.). Exclusively geocarpous fruits are formed by Geococcus pusillus Drumm. Special adaptations to dispersal are explosive dehiscence (autochory: species of Cardamine L. [Overbeck, 1925; Schneider, 1935; Fig. 6.2]), a corky pericarp (hydrochory: species of Cakile Miller and Crambe L.), flattened fruits with broad wings or bladdery fruits (anemochory: according to Al-Shehbaz [1984], this arose several times in different groups as in Figs. 6.10, 6.14, & 6.16), and hooked hairs or glochidia (epizoochory: Tauscheria Fisch. ex DC. and Clypeola L.). Some genera of Thelypodieae produce comparatively long gynophores (e.g., Stanleya Nutt.; Fig. 6.3; Warea Nutt.), and are proposed to relate Brassicaceae to CapparaceaeCleomoideae.

174

THE BOTANICAL REVIEW

Tetravalvate fruits are occasionally found and, in several taxa, appear to be a quite constant character, causing some authors to create new genera for tetravalvate forms. Draba kusnetsovii (Turcz.) Hayek has tri- or tetramerous ovaries and was separated as genus Holargidium Turcz. (Solms-Laubach, 1900). Taxa ofRorippa Scop. often show tetravalvate fruits (R. hispida [DC.] Britton, R. globosa [Turcz.] Vassilcz., R. palustris [L.] Besser etc.; cf. Borb~, 1879; Fotoni([, 1892; Gerber, 1899c, 1899d; Solms-Laubach, 1900). The genus Tetrapoma Turcz. was based on this character (Fig. 6.18) and contained T. barbareaefolium (DC.) Fisch. et Mey. (presently treated as either Rorippa hispida f. tetrapoma N.Busch [Vasil'chenko in Komarov, 1939] or as the species R. barbareifolia [DC.] Kitag.). The genus Tetrapoma cannot be maintained, because bi-, tri-, and tetravalvate fruits occur on the same plant. However, a geographical variation in gynoecial structure ofR. hispida has been described. On Kamchatka there are populations that produce consistently bivalvate fruits, whereas in the Amur region exclusively tetravaivate forms occur (Vasil'chenko in Komarov, 1939). Rorippa plants with tetravalvate fruits that were found in Middle Europe were also often thought to belong to Tetrapoma. A similar case of tetravalvate populations beside normal bivalvate populations is given in the Californian Tropidocarpum capparideum Greene (see Rollins, 1993). A taxon with poly- (mainly tetra-) valvate fruits designated Capsella viguieri Blar. was separated from C bursa-pastoris (L.) Medik. (Blaringhem, 1910; Blaringhem & Viguier, 1910). It was based on a single mutant plant, the descendants of which displayed an unaltered deviating phenotype (Shull, 1929). Nevertheless, the taxon was not accepted as a new species (see also Becquerel, 1911; and Buchet, 1911, for critical remarks). A taxon with the number of fruit valves permanently higher than two is Brassica rapa L. ssp. trilocularis (Roxb.) Hanelt (Hanelt, 1986; Gladis & Hammar, 1992). E. BERBERIDACEAE The fruits of the Berberidaceae develop from ascidiate gynoecia ("ascidate" of Taylor, 1991) and have, in several cases, massive placental regions; the seeds are inserted parietally or more or less basally (Fig. 7). The gynoecium as well as the fruit ofGymnospermium albertii (Regel) Takht. are pronouncedly stipitate. The indehiscent fruit forms (e.g., Berberis, Podophyllum) are berries (bacca type, Spjut, 1994). Dehiscent capsules are produced in the Epimediinae. In Vancouveria C.Morr. et Decne., Epimedium L., Jeffersonia, and Plagiorhegma, dehiscence runs basipetally, delimiting a valve that extends nearly to the fruit base in the two former genera (Figs. 7.1B & 7.2; see also Berg, 1972). In Plagiorhegma, the valve ends at the lower third of the fruit (Fig. 7.3). The remaining part of the fruit is nearly as wide as the valve and bears the broad placenta along its middle; it is topped by the style. The fruit ofJeffersonia dehisces with a transverse slit in the upper third (Fig. 7.4B) and the valve is thus reduced to its upper edge. In the literature, this fruit type is often termed "follicle-like with horizontal dehiscence" (Rachenbalg; cf. Winkler, 1940). Spjut (1994) applies the term "pyxidium"; however, the fruit top is not dropped. The one-seeded fruit of Achlys DC. shows demarcation of a valve, but because it is functionless there is no dehiscence. Loconte (1993) calls this fruit an achene; however, it is berrylike with an appendage that indicates a myrmecochorous dispersal (Endress, 1996). Basal placentation is seen in,4chlys, Bongardia C. A. Mey. (with tendency to a free central placenta; Terabayashi, 1983a), Leontice L. (Fig. 7.5A), Gymnospermium Spach, and Caulophyllum Michx. The capsules of Bongardia and Leontice are thin-walled bladders. The Bongardia fruits tear irregularly at the top, whereas those of Leontice remain indehiscent and are adapted to dispersal as tumbleweeds eroding in the upper half(Fig. 7.5A; Townsend & Guest,

CARPEL NUMBER

175

Fig. 7. Fruit forms in Berberidaceae (redrawn except I & 3; relative size not to scale). 1. Vancouveria hexandra (Hook.) Morren et Decne., capsule: A. Valve view; B. Lateral view. 2. V. chrysantha Greene, dehisced capsule. 3. Plagiorhegrna dubia Maxim., lateral view of capsule. 4. Jeffersonia diphylla, capsule: A. Lateral view of gynoecium; B. Dehisced fruit. 5. Leontice leontopetalum L.: A. Longisection of apically dehisced fruit; B. Immature capsule. 6. Podophyllumpeltatum L., berry. 7. Berberis integerrima Bunge, berry. 8. Schematic transection of berheridaceous ovary with one placenta. Sources: 2 and 8 redrawn from Rohweder & Endress, 1983; 4A and 5A, from Baillon, 1872; 4B, from Wettstein, 1935; 5B and 7, from Townsend & Guest, 1980; 6, from Mitchell, 1983.

176

THE BOTANICAL REVIEW

1980). The term "pyxidium" used by Hilger and Hoppe (1995) for that type is not very suitable. The fruits of Gymnospermium and Caulophy!lum (glandispermidium type after Spjut, 1994) are destroyed by the accrescent seeds. Portions of the papery pericarp partly enclose the ripe Gymnospermium seeds. Caulophyllum is secondarily fruitless, as the gynoecium displays a transverse opening pattern immediately after anthesis, and the ripe seeds are freely exposed (cf. Brown, 1818; Mitchell, 1983; Loconte & Estes, 1989a; Endress, 1996). The remaining seven genera possess indehiscent berries (Figs. 7.6 & 7.7). Placentation is predominantly parietal; (sub)basal placentation is seen in Berberis and Mahonia (Fig. 21.8).

V. What Is a Carpel? Knowledge of the phylogeny, morphology, and ontogeny of the gynoecial unit, the 'carpel,' is needed before making statements and coclusions concerning carpel numbers in syncarpous gynoecia. However, information about the origin of the female reproductive organs of the angiosperms is still inadequate. Extensive literature is available on carpel theories, and the most important will be reviewed briefly; for a more detailed discussion see Lain (1959), Moeliono (1970), Friis and Endress (1990), and Briickner (1991a). The term"carpel" was coined by Dunal, a student ofA. P. de Candolle, in his Annonaceae monograph of 1817. The diminutive of the Greek carpon (fruit), it was used to designate the individual fruitlets of the annonaceous multiple fruit and was thus a descriptive term for a well-defined organ. Its meaning changed when the difficulties of explaining the organization of syncarpous and inferior gynoecia became obvious. Thereafter, "carpel" was also applied to the theoretical units of these complex organs that had to be mentally dismembered, as there were no histological borderlines between the "congenitally fused" components. The first botanists to use the term in this transformed sense were A. P. de Candolle, R. Brown, and J. G. C. Batsch. For a historical review of the term "carpel," see Lorch (1963). The main interpretations of the nature of the gynoecial units are: 1. Carpels are organs sui generis. Gr~goire (1931, 1938) discounted homologies between vegetative and reproductive organs of angiosperms. He supposed fundamental differences between a vegetative apex and a floral apex. This concept has not been confirmed, for the transition from the vegetative apex to the floral apex is continuous (cf. Goyal & Pillai, 1985). 2. Carpels are modified leaves. The foliar theory, supposing a primitive vegetative leaf to be the forerunner of the ovule-beating sporophyll, dates back to the classical essay "Versuch die Metamorphose der Pflanzen zu erkl~en," by J. W. v. Goethe (1790). It is based mainly on teratological findings (e.g., proliferated flowers, the generative apex of which returns to the production of vegetative phyllomes above the carpels, or virescent flowers with organs resembling green leaves; Fig. 8.1). The German term Fruchtblatt for the carpel reflects the acceptance of its phyliomic nature. 3. Carpels are phyllomes that have never been vegetative leaves but always ovule-beating sporophylls. This concept was developed after W. Hofmeister discovered the homologies in the alternation of generations of seed plants and cryptogams (1851). Supported by palaeohotanical evidence of the closed conduplicate nature of the carpels of the fossil Archaefructus Sun, Dilcher, Zheng et Zhou (Sun et al., 1998), it has been widely accepted. In light of this idea, attribution of the most primitive state to carpels that look most leaflike cannot be justified (see criticism by Carlquist, 1969). 4. Carpels are ofphyllomic nature and envelop the ovules, which are, however, axis borne. In the nineteenth century, ovules were thought to be bud homologues and thus could never be leaf borne (St.-Hilaire, 1841; Schleiden, 1843; Payer, 1857; Eichler, 1875; and others). The

CARPEL NUMBER

177

Fig. 8. Illustrations of carpel theories (redrawn). 1. Virescent monocarpellate gynoecium of Prunus serrulata Lindl., prepared from a doubled flower (A) and unfolded (B). 2. Pseudanthium theory by Wettstein: Angiosperm gynoecium (B) originates from reduced female partial inflorescence of Ephedra type (A) that becomes enveloped by its subtending bracts. 3. Gonophyll theory by Melville: A. Primitive gynophyll (leaf with epiphyllous ovuliferous branch system); B, C. Reduction and enclosure of the fertile branch leads to a monocarpellate gynoecium; D. Sterile gonophyll (tegophyll); E. Ebracteate fertile branch; F. Tegophylls alternate with fertile branch systems and envelop branches belonging to neighboring forks: origin of choricarpous and syncarpous gynoecia. 4. Croizat's concept: A. Placenta ("maerosporophyll") with subtending scale (neither "telome" nor "phyllome" but "emergence," having potencies of both); B, C. Merging of placenta and scale into an apparently homogeneous leaf organ with (sub)marginal ovules. Sources: 1 and 2 redrawn from Weberling, 1981; 3, from Melville, 1960; 4, from Croizat, 1964.

178

THE BOTANICAL REVIEW

placentae were interpreted as parts of the undivided or forked floral axis that had been phylogenetically joined with their subtending leaf, the carpel. Although, in more recent times, the majority of botanists have been convinced that ovules are primarily carpel borne, several authors have pointed out that ovules may arise directly from the floral apex and that, at least in some angiospermous taxa, true stachyospory is seen (Sachs, 1870; Hagerup, 1934; Fagerlind, 1946, 1958; Lam, 1948, 1950, 1959; Moeliono, 1970; Pauz6 & Sattler, 1979; Sattler & Lacroix, 1988). Various pseudanthium hypotheses have deduced the angiosperm flower from a pluriaxial reproductive region of gymnospermous or pteridospermous ancestors. Wettstein (1907, 1935), Neumayer (1924), and Janchen (1950) postulated an inflorescence resembling that of Ephedra L. as the starting point, which implies homology of angiosperm ovules with female Ephedra flowers and homology of carpels with the subtending bracts of these flowers (Fig. 8.2). Melville (1960, 1962, 1963, 1983) suggested that the basic component of a gynoecium is a leaf with an epiphyllous fertile branch, based on glossopteridalean ancestors of the angiosperms (Fig. 8.3). Fusion of leaf and bifurcated ovuliferous branch leads to an organ that, to avoid confusion, is termed not "carpel" but "gynophyll." (The latter term was coined by Neumayer (1924), who used it in a somewhat different sense: for only the leaf member of the leafbranch system. However, Melville does not refer to this paper.) Syncarpous gynoecia are said to consist of bifurcated fertile axes and secondarily sterile leaves ("tegophylls") that alternate with and marginally fuse with one of the fertile branches (Fig. 8.3F). 5. Carpels are emergences of the sporogenous axis. This concept resembles the preceding one, but the flattened envelopes of the fertile axes are not thought to have any leaf character. Carpels are seen as protective scales that may be either be of axial origin (Thompson, 1934, 1937; Plantefol, 1949) or neither phyllome nor telome (Croizat, 1960, 1964; Fig. 8.4); they may fuse with the axial placentae. The term "carpel" in the sense of"Fruchtblatt" is refuted. 6. Carpels are ovuliferous cupulae or organs associated with cupular structures. Thomas (1931, 1934, 1938) referred to the mesozoic Caytoniales that produced cupuliferous axes (Fig. 9.3) and derived an angiospermous ovary from a phyletic fusion of two many-seeded eupules. Meeuse (e.g. 1979, 1990) theorizes an ovuliferous cupule of the pteridosperms transformed to an angiosperm female reproductive organ that he terms "monogynon," not "carpel." He points out that a monogynon is not of phyllomic nature and has nothing in common with a sporophyll. A condensed female branch (gynocladium) with several monogyna would give rise to a female flower with a choricarpous gynoecium (Fig. 9.2). The derivation of the carpel via cupuliferous pteridosperms ofCaytoniaH. H. Thomas type is also favored by Stebbins (1974), Doyle (1978), Crane (1985, 1986), and Doyle and Donoghue (1986, 1987). However, here the cupule wall is thought to transform into the second integument of the ovule, whereas the cupuliferous rhachis should have flattened and enfolded the ovules. Retallack and Dilcher (1981) base the derivation of the carpel on the glossopterid taxon Denkania Surange et Subh. Chandra, which had bitegmic ovules inserted on a leaflike bract (Fig. 9.1). According to Long (1977), the pteridospermous cupule gave rise to the carpel, whereas the second integument developed from a proliferation of the eupule. Doweld (1996) proposes a dimerous cupule ofLeptostrobus Heer type (Czekanowskiales) as the forerunner of the true angiosperm carpel. 7. Carpels are derived from fertile fronds of ophioglossoid ancestors. Kato (1990, 1991) proposes a close relationship among Ophioglossaeeae, glossopterids, and angiosperms. The tridimensional structure of the fertile Ophioglossaceae fronds, with their epiphylious sporophores (Fig. 9.4), is thought to be homologous to an angiosperm carpel with adaxially inserted ovules.

CARPEL NUMBER

179

Fig. 9. Illustrations of carpel theories, continued (redrawn). 1. Denkania, bitegmic ovules borne on a leaflike bract. 2. Anthocorm theory by Meeuse: A. Primitive gynoeladium with ovuliferous cupules (monogyna) still open; B. Condensed gynocladium (female reproductive unit) with closed monogyna. 3. Caytonia: A. Cupuliferous axis; B. Cupule bent inward; C. Longisection ofeupule showing enclosed ovules. 4. Kato's concept: A. Botrychium Sw., three-dimensional frond with sporophore inserted on the adaxial side of the trophophore; B. OphioglossumL., two rows ofsporophores near the leafmargin. 5. Sattler's concept (gynoeeial primordia dotted, ovules black): A. Carpellary organization (ovules carpel borne); B. Oehna atropurpurea DC. (ovules initiated at the flank of the floral apex); C. Myrica gale L. (floral apex transformed into the single ovule); in B and C "carpellary" organization is inferred by imaginary carpel borders (dashed lines). Sources: I redrawn from Retallack & Dilcher, 1981; 2, from Meeuse, 1979; 3, from Thomas & Spicer, 1987; 4, from Kato, 1991; 5 from Sattler, 1974.

8. Carpels are transformed microsporophylls. Meyen (1988) considers the Bennettitales the closest relatives of the angiosperms. As their female reproductive organs differ strongly from those of the angiosperms, he postulates the occurrence of"gamoheterotopy." According to this theory, an expanded microsporophyll would have experienced a change in sex, producing ovules instead ofmicrosporangia. Microsporophylls resembling carpels in their shape are known in the fossil taxon.

180

THE BOTANICAL REVIEW

9. Carpels are gynoecial appendages that enclose the ovules; they do not necessarily bear them. This is a neutral carpel definition from an organogenetic background given by Sattler and Perlin (1982). The discussion of"phyllome versus caulome" is considered useless, as the meristematic floral apex produces a continuum. The observable growth centers are not sharply delimitated, and the receptacle should at least be considered a caulome-phyllome transition zone (of. Sattler, 1974; Endress, 1977; Leins & Erbar, 1985a; general discussion in Rutishauser & Sattler, 1985). A strictly empirical morphology is demanded with its descriptive terms free of theoretical burdens. The historical term "carpel" is recommended to either be fully avoided or, at most, be used for unambiguously ovule-bearing appendages (Fig. 9.5A); and gynoecia with ovules arising directly from the floral apex (Figs. 9.5B & 9.5C) would be termed "acarpellate," unless the redefinition of"carpel" given above is accepted. 10. The term "carpel" is used in an abstract manner to refer to a character-complex comprising abstract entities (character states). The conceptual framework by Cresens and Smets (1989, 1992) approaches the carpel as a character set (in this context, a "character" is required to have systematic relevance and predictive value sensu Dahlgren, 1980). The carpel is one particular form of expression of the category "ovuliferous phyllome." "Phyllome" is used here to designate all flattened appendages irrespective of their morphological nature. The earliest stages of carpel ontogeny are emphasized. A carpel primordium may consist of two "phyllomes" (growth centers) developing on a basal zone. Cresens and Smets stress the similarity of this developmental mode to the development of the outer envelope ("third integument") of the ovule in the gymnosperm Gnetum gnemon L. Endress (1994b) also considered possible homologies between the female structures of Gnetum and uniovulate gynoecia of several paleoherbs. Thus, the origin of the carpel is not resolved. In the analysis presented here, phylogenetic hypotheses and the problem of phyllospory versus stachyospory play only a minor role. Regardless of homology, carpels are considered lateral appendages of the receptacle, from which they lack a sharp boundary.

VI. Theories to Explain Bivalvate Gynoecia in Papaverales and Capparales (Also Considering the Monovalvate Pistils of Berberidaceae) The various theories concerning carpel number in valvate gynoecia are only partly based on pistil morphology. Especially for Brassicaceae, the phyllotaxis and organization of the whole flower, be it dimerous, tetramerous, mixed di- and tetramerous, (ancestrally) pentamerous, or polyaxial, influenced the interpretation of gynoecial composition. This practice led to some untenable conclusions lacking empirical proof. A discussion of the role of abortion versus splitting in the two whirls of the Brassicaceae androecium in interpretation of the gynoecium is beyond the limits of this paper; summarizing reviews are given by Cejp (1925) and Gu6d6s (1967). The section discusses valvate gynoecia only in the taxa of concern, although these concepts have also been applied to the paracarpous gynoeeia of Dilleniidae taxa (Salicales and Violales), Orobanchaceae, Orchidaceae, and others (see Meeuse, 1975). A. THE nl THEORY In this theory carpel number equals the number of placentae (n), and for siliquae with two placentae it is known as the "bicarpellary theory" (Fig. l 0a). According to the nI theory, bivalvate gynoecia are considered largely symplicate and consist of two congenitally fused car-

CARPEL NUMBER

181

Fig. 10. Schematic transections of bivalvate gynoecia (bold black lines are zones of dehiscence; white, sparsely dotted, and densely dotted areas indicate carpels according to the different theories).

182

THE BOTANICAL REVIEW

pels that are positioned in the transverse plane. The carpel tips end in carinal stigmatic lobes that are, correspondingly, also in transverse position. Stigmatic lobes in median position are explained as commissural stigmas produced by the fused carpel margins. All carpels are fertile and have marginal placentae. The placental regions that form the replum are the congenitally fused carpel margins; their more or less complex vaseulature consists of fused ventral bundles. The valves are secondary structures that are encircled by a continuous zone of separation tissue within the laminae of the carpels. Thus, valvate dehiscence is a special mode of loculicidal dehiscence. The Brassieaceae septum is explained in several ways, but is mostly considered to be a product of fusion of two placental outgrowths. Other interpretations are discussed below. The gynoecium of the Berberidaeeae is monomerous and consists of a carpel ascidiate up to the stigma. The massive placenta extends over the whole ascidiate part but, in some taxa, bears ovules only at its base. As in the taxa mentioned above, the zone of dehiscence is differentiated within the parenchyma of the carpel lamina and enables a modified Ioculicidal opening of the peculiar fruit (Rachenbalg). First applied to the Brassicaceae (Brown, 1817; de Candolle, 1821a), the nI theory is the oldest explanation of the gynoecial construction of the taxa under consideration. Valvate gynoecia were not treated as anything particular, and their interpretation does not differ from the general explanation of syncarpous gynoecia. This idea has gathered wide acceptance (see Appendix 1). In Brassicaceae and some other bivalvate taxa, several authors, especially of the nineteenth century, postulate a four-carpellate ancestor. The two carpels of the inner median whorl are thought to become suppressed or lost phylogenetieally. The occasional occurrence oftetravalvate fruits is interpreted as a reversionary character. It cannot be clearly understood from the literature whether the inner whorl is considered completely missing or reduced to a small amount of tissue integrated into the placental regions, in which case the authors would be adherents of the 2n theory (see below). B. THE nlI THEORY The concept termed the "nil theory" was created by Hochstetter (1847, 1848) for the Brassicaceae gynoecium and extended to Papaveraceae, Fumariaceae, and Capparaceae. In bivalvate taxa there are two carpels, but, in contrast to the nI theory, these carpels are supposed to occupy a median position. The replum is thus formed by their dorsal parts, and their tips are the median stigmatic lobes that are not of "commissural"--namely, double--nature. The transverse stigmatic lobes of some Papaveraceae members are thought to be likewise of dorsal origin, as they are replum borne. A median carpel position implies a median and thus dorsal placentation. Hochstetter considered the placentae and the septum to be products of the bifurcate floral axis. The valves are double structures, consisting of the fused lateral parts of the median carpels. Their midveins are believed to represent a suppressed transverse carpel pair (Hochstetter, 1848). Fruit dehiscence is loculicidal along both sides of the dorsal region of the carpels. Similar to Hochstetter's ideas, Spratt's (1932) concept became popular, but it was restricted to the Brassicaceae (Fig. 10b). The dorsal placentae were considered carpellary with insertion of the ovules along the midrib of the carpels. The septum is thus unambiguously false and, as a secondary outgrowth, connects the median parts of both carpels. This structure is thought to he comparable to that in Boraginaceae and Lamiaceae gynoecia. In Papaveraceae, the supposed forerunners in a phylogenetic line ending with Brassicaceae, the

CARPEL NUMBER

183

ovules had already moved from the carpel margins to the lamina, a phenomenon especially conspicuous in Papaver. The nII theory is widely rejected today. Nevertheless, two authors who have studied Brassicaceae gynoecia in detail, Roth (1957, 1977) and Eigner (1973), regard it not at all refuted. In "Fruits of Angiosperms" Roth (1977: 219) writes: "The ovary of Cruciferae still remains an open question to which no answer is found either by ontogenetic or by anatomical studies, unless the explanation of Spratt is accepted." Additional concepts claiming placentation along carpel midribs can be reckoned with the nII theory s.l. Lignier (1896a, 1896b, 1896c, 1896d) considered Fumariaceae and Brassicaceae carpels tripartite. A fertile median part is flanked by two sterile lateral lobes; the valves arise by fusion of the sterile lobes of both carpels. Later on, Lignier rejected his idea of trilobed carpels in the median plane. Goebel ( 1933:1916) wrote that the Brassieaceae septum as a fused placental outgrowth "umrahmt von den Mittelteilen der Fruchtbl~tter stehen bleibt, wenn die Seitenteile bei der Fruchtreife sich abl6sen." Obviously the replum is interpreted as originating in the dorsal carpel regions. As to Papaveraceae, Mohl (1836) postulated a more or less laminal placentation, especially for Papaver. Influenced by this hypothesis, some authors described the placenta of Papaver as inserted along the median line of the carpels (Tr~cul & Paty, 1845; Clos, 1862; Duchartre, 1867). C, THE 2n THEORY According to the 2n theory s.str., the gynoecia concerned consist of twice as many carpels as placentae. The term "tetracarpellary theory" is commonly applied to bivalvate gynoecia. In these taxa four carpels are present, mostly thought to be arranged in two dimerous whorls, with members of the two whorls differing in shape. They all can bear stigmatic lobes, or the stigmatic lobes of one pair may be reduced or absent. The second carpel pair is not believed to appear only sporadically as an atavistic reminiscence ofa tetracarpellary ancestor, as seen by some adherents of the nI theory, but is present in every bivalvate gynoecium. The 2n theory s.l. is maintained by several authors who note a conspicuous dimorphism of gynoecium units but do not interpret all of them as carpels. The valves are regarded as of carpellary or foliar nature, respectively, whereas the placental regions may be branched axes alternating with the phyllomes. Within the 2n theory there are several subtheories. Its originator, J. Lindley (1828), proposed three different hypotheses concerning Brassicaceae and Papaveraceae (Eschscholzia). Following are explanations of the two carpel pairs: A. The transverse carpels of the outer whorl have an expanded lamina; they are sterile and form the valves (Fig. 10c). Often they are designated "valve carpels" or "expanded carpels." The median carpels of the inner whorl have a reduced lamina and no locules of their own; they are fertile and form the replum. They are called "contracted" or "solid" carpels. Both carpel forms are modifications of the general carpel type. The Brassicaceae septum is mainly interpreted as part of the lamina of the median carpels, but is also said to be a placental outgrowth (Lindley, 1828, 1847) or an axial structure (Chodat & Lendner, 1897). More or less divergent ideas on the nature of the placental regions (e.g., interpretation as fertile branches) are nevertheless considered part of this variant of the 2n theory. Table I is a chronological listing of the supporting literature. B. The transverse carpels are completely developed and fertile (Fig. 10d). The median carpels are strongly reduced and have perhaps nearly disappeared except for a small radial strip

184

THE BOTANICAL REVIEW

Table I Chronological list of literature supporting variant A of the 2n theory (valves = sterile expanded carpels, replum = fertile solid carpels)

Papaveraceae Lestiboudois, 1823a, 1823b: refers not to "carpels" but to "valves" alternating with "trophosperm" = "cordons pistillaires" Bravais & Bravais, 1839 Griffith, 1847: Eschscholzia, and possibly other taxa Van Tieghem, 1868, 1875 Henslow, 1891: Eschscholzia Kerner, 1898 Dickson, 1935 Bersillon, 1955: refers to "dimorphic pistil elements" Kuusk, 1960 Fames, 1961 Melville, 1962, 1963: refers to a "tegophylr' pair = sterile phyllomes enveloping a pair of ovule-bearing axes

Fumariaceae Lestiboudois, 1823a, 1823b: refers not to "carpels" but to "valves" alternating with "trophosperm" = "cordons pistillaires" Bravais & Bravais, 1839 Eames & Wilson, 1928, 1930 Kuusk, 1960

Capparaceae Lestiboudois, 1823a, 1823b: refers not to "carpels" but to "valves" alternating with "trophosperm" = "cordons pistillaires" Fames & Wilson, 1928, 1930 Stoudt, 1941 Kuusk, 1960 Fames, 1961

Brassieaceae Lestiboudois, 1823a, 1823b: refers not to "carpels" but to "valves" alternating with "trophosperm" = "cordons pistillaires" Lindley, 1828, 1830 Kunth, 1831, 1833 Bravais & Bravais, 1839 Griffith, 1847 Fournier, 1864: the septum is thought to consist of two addorsed carpels; 1865a, 1865b: the placental regions that are connected by the septum are independent organs of probable axial nature Van Tieghem, 1868, 1875 Duchartre, 1870-1871 Huisgen, 1873: placentae here as in other families arising from "independent blastemes" sensu Hanstein, 1873 Henslow, 1888, 1891 Chodat & Lendner, 1897 Kerner, 1898 Gerber, 1905a, 1905b, 1905c, 1905d: the septum is an outgrowth of the median parts of the fertile carpels that is fused with the floral axis Fames & Wilson, 1928, 1930 Alyavdina, 1931 Dickson, 1935 Purl, 1941 Gazet du Chatelier, 1946 Yen, 1957 Kuusk, 1960 Fames, 1961 Melville, 1962, 1963: refers to a "tegophylr' pair = sterile phyllomes enveloping a pair of ovule-bearing axes Merxm011er & Leins, 1966, 1967 Eigner, 1973 Klopfer, 1974 Weymar, 1988: the nI theory is equally accepted Jacob et al., 1994

CARPEL NUMBER

185

of tissue; they are sterile but may end in stigmatic lobes. Lindley (1828) proposed this variant for Eschscholzia (Papaveraeeae), and later it was applied to Brassicaceae by Klein (1894), Martel (1898), and Motte (1957; see Gu6d6s, 1967). Klein explained the septum as a product of the sterile median carpels. Martel believed it to consist of the fused margins of the fertile transverse carpels that have laminar placentation. According to Motte, the cruciferous flower is a polyaxial system. The median carpel pair is thought to form the outer whorl, and each of its members subtends an "axis" that bears a median pair of stamens. C. Both carpel forms are fertile, although the median carpels are more or less reduced, compared with the transverse ones (Fig. 10e). The placentae are formed by the fused carpel margins, as is normal in paracarpous gynoecia. Because of the extremely reduced lamina of the median carpels, however, the two flanking placentae approach each other very closely and appear to be a homogeneous placental region. Lindley (1828) was the first to assume such a structure of the Eschscholzia gynoecium. He considered that the rudimentary median carpels might possess a narrow placental area of their own that could not be delimited from the welldeveloped placentae of the transverse carpels. This interpretation has been overlooked for a long time. Purl (1950), though having returned to the nI theory, demonstrated by means of teratological gynoecia in Crateva (Capparaceae) that, if there were 2n carpels, all of them were unambiguously fertile. The present author confirmed the occurrence of two fertile carpel forms in Papaveraceae and Fumariaceae (Gonnermann, 1979, 1980, 1982; Brtickner, n6e Gonnermann, 1982, 1984). D. In this variant, the ancestors of the Brassicaceae were thought to have pentamerous gynoecia, with one carpel becoming suppressed in the extant representatives. The remaining four are in diagonal position and display an identical shape (Fig. 10f). There are two zones of fusion in the transverse plane and two in the median plane; only the latter ones bear fertile placentae and form commissural stigmas. The septum is produced by the inward extension of the upper surface of the carpel margins. The fused margins in the transverse plane form the midribs of the valves. Placentation and septum formation are suppressed. Each valve consists of parts of two carpels. Dehiscence is loeulicidal. This variant of the 2n theory was maintained by Henslow (1880); it has had no other adherents. E. Gerber (1899a) proposed a tetramerous Brassicaceae gynoecium. Comparable to the origin of the four longer stamens by transverse d6doublement, it arose by median d6doublemerit o f a dimerous carpel whorl. The two outer carpels, with their dorsal regions in the median plane, are sterile and form the ovary wall. The two inner carpels, also in the median plane by origin, are addorsed and fuse back to back to form the septum. They are fertile, with their ovuliferous margins toward the ovary wall. Soon Gerber published this concept, he changed his mind and, instead, proposed six carpels (see below) and accepted variant A of the 2n theory. In the Berberidaceae gynoecium, in addition to a well-developed carpel, one or two reduced ones may be integrated. The predominantly trimerous flower construction is supposed to indicate a primarily tricarpellate gynoecium. Under this concept, the extant Berberidaceae have pseudomonomerous gynoecia. Morren and Decaisne (I 834) assumed a dimerous gynoecium in Epimedium, with one carpel sterile and one bearing the ovules along its median zone. This concept corresponds with variant A: a sterile valve carpel and a fertile carpel being more or less consolidated. The dorsal placentation is, however, unusual. More recent authors suppose one or two vestigial carpels to be integrated into the massive placental region. The veins assigned to these carpels are, however, involved in the vascular supply of the ovules. Thus, fertility of all carpels could be claimed in accordance with variant C. Such a kind of pseudomonomery is accepted by Chapman (1936), Eckardt (1937: still uncertain; 1963), Puri (1952),

186

THE BOTANICAL REVIEW

Kumazawa (1958, pers. comm. in Eckardt, 1963), Takhtajan (1959, 1987), Eames (1961), Kaute (1963), Berg (1972), and Huber (1991). D. THE 3n THEORY The occurrence of six carpels in a bivalvate gynoeeium has been assumed "only for the Brassicaceae. Again, the number of stamens was considered in this theory (see Duchartre, 1870--1871 and reply by Eiehler, 1872). According to Gerber (1899b, 1899c, 1899d, 1900a, 1900b), the Brassicaceae gynoeeium consists of two dimerous carpel whorls. The outer one is sterile and corresponds to the valves. The inner one is also sterile, and its members display a reduced lamina; they form the outer part of the replum and the median stigmatic lobes. At their bases the inner carpels fork in radial directions. This d~doublement gives rise to two addorsed carpels that fuse back to back, forming the septum. Their ventral parts are directed outward and bear the ovules; thus, only the derivatives of the inner whorl are fertile (Fig. 10g). Gerber assumed a similar construction of the Papaveraceae gynoecium. The concept of Yen (1959) recalls Gerber's hypothesis. He also postulated two carpel whorls in the Brassieaceae gynoeeium. However, the outer whorl consists of four sterile earpels. A transverse pair corresponds to the central parts of the valves. A median pair forms the lateral valve parts and the outer replum tissue. In the inner whorl, two addorsed fertile carpels with marginal placentation are fused with the median outer carpels along their inner surface. The septum tissue connects the dorsal parts of the inner fertile carpels. Martel (1900, 1902) maintained a very complicated organization of the crueiferous flower. He interpreted six whorls: two median sepals; two lateral sepals; two tripartite phyilomes, with the middle segment reduced to a vein running in the outer replum and taking part in the formation of the gynoecium beak, whereas the lateral segments become the four petals; two short transverse stamens; two tripartite phyllomes, with the middle segments forming the inner fertile replum part with its inverted bundles and the septum, the lateral segments being the four long stamens; and two transverse sterile carpels, the valves. Thus, the gynoecium would be combined from six components (3n theory s.l.). Regarding the sterile valve carpels, Martel's theory is approximately in accordance with variant A of the 2n theory. He emphasized, however, that the structure septum + replum must not be equated with two carpels but consists of several phyllome fractions. E. SAUNDERS'S THEORY OF CARPEL POLYMORPHISM Beginning with studies on Brassicaceae gynoeeia (1923), E. R. Saunders developed a peculiar theory of gynoeeial construction in angiosperms. This theory, called carpel polymorphism, was developed in a series of publications that climaxed in a two-volume book, "Floral Morphology" (1937). Though Saunders's concept was criticized (e.g., Parkin, 1926; Eames, 1931; Arber, 1931 a, 1931 b), she did not divert from it but defended it passionately. The basic idea of carpel polymorphism is that there is not a general carpel type but two main forms that may be fertile or sterile (see Saunders, 1925, 1929b). The first is the hollow or valve carpel. It corresponds to the only carpel type recognized so far and is considered the most primitive form. Its lamina is well developed and pinnately or, rarely, palmately veined. If it is fertile, the ovules are inserted in one row on each of its margins. The second main form is the solid or consolidated carpel. It may be a derivative from the valve carpel. It has no iam-

CARPEL NUMBER

187

ina and is reduced to a radial plate of tissue or, in the most extreme case, to a single or double vascular bundle. If solid carpels occur in combination with valve carpels, the former are fertile and the latter are sterile. Fertile solid carpels bear ovules either singly or in one to several rows along both sides of the midrib. A modification of the solid carpel is the very variable semisolid or pseudovalve carpel. It arises by lateral, not necessarily symmetrical, expansion of the solid carpel on both sides of its midrib. Externally resembling a valve carpel, when fertile the placentae are on both sides of the double vein in the midrib. As to the families concerned, Saunders (1923, 1925, 1926, 1927, 1928b, 1929a, 1929b, 1930, 1931, 1932, 1937) concluded the following: Papaveraceae: Most bivalvate gynoecia consist of two whorls, the outer valve carpels, sterile, the inner solid carpels, fertile (Fig. 11.1). Stigmatic lobes over the valves are of the 89+ 1 + 89pattern (fusion of stigma of valve carpels with one-half of the bifurcated stigmatic lobe of the neighboring solid carpels). Bivalvate Eschscholzieae: 20 carpels, outer whorl of 8 sterile valve carpels, inner whorl of 10 sterile solid carpels and 2 median fertile solid carpels (Fig. 11.2). Here the valves are compound, consisting not of one valve carpel, as in Chelidonieae, but of 9 carpels (5 solid, 4 valve, all sterile). Papavereae: alternating fertile solid and sterile valve carpels (Fig. 11.4). Platystemonoideae: at first thought to have fertile valve carpels only; later Platystemonwas described as having 9-12 sterile solid or valve carpels alternating with 9-12 fertile semisolid carpels (Fig. 11.3) and in Hesperomecon Greene 3 sterile solid carpels alternating with 3 fertile semisolid ones. Fumariaceae: Two lateral, sterile valve carpels and two median, fertile solid carpels in two dimerous whorls (Fig. 11.5). Capparaceae: Either 4 carpels (2 sterile valve, 2 fertile solid) in one whorl or two 4-8 merous whorls, outer valve carpels sterile, inner solid carpels fertile. Brassicaceae: One whorl of 4 orthogonal carpels in two pairs. Siliqua: lateral valve carpels sterile, median solid carpels fertile (Fig. 11.6). Silicula: lateral valve carpels sterile, median semisolid carpels fertile (Figs. 11.7 & 11.8). Biscutella L. is an exception with sterile, solid median carpels and fertile, lateral valve carpels (Fig. 11.9). Berberidaceae: Two carpels. At first several combinations were proposed (Nandina: both valve carpels, one fertile; Berberis, Mahonia,Podophyllum,Leontice, Bongardia: valve carpel plus fertile solid carpel; Epimedium, Jeffersonia: valve carpel plus fertile semisolid carpel). Later, the latter case was favored for all taxa (Figs. 11.10 & 11.11). Among more recent authors, only Emberger (1960) maintains the theory of carpel polymorphism. It is of historical interest only. F. OTHER THEORIES De Candolle (1821 b, 1824) believed that an axial cupule had a role in the formation of the

Papaver ovary, leaving only the carpel tips exposed. Clos (1865) assumed that the ovaries of Papaver and, perhaps, also the related Meconopsiscambrica (L.) Vig. were merely products of the floral axis, whereas in the Chelidonieae and Eschscholzieae two carpels apparently existed. Thus, the Papaveraceae were considered heterogeneous with regard to gynoecial morphology. Later, Tr6cul (1873) claimed axial nature for all papaveraceous gynoecia. Koch (1869) was convinced of an axial origin of the capparaceous gynoecia. In an early paper, Martel (1895) interpreted the flowers of Brassicaceae, Capparaceae ("Cleomaceae"), and Fumariaceae as polyaxial systems with four"buds" in the axils of the sepals. The two inner buds were thought to develop completely; their "phyllomes" are petals, sepals, and carpels. In con-

188

THE BOTANICAL REVIEW

Fig. 11. Theory of carpel polymorphism (1-4, Papaveraceae; 5, Fumariaceae; 6-9, Brassicaceae; 10, 11, Berberidaceae). 1. Glaucium: 2 fertile solid carpels, 2 sterile valve carpels. 2. Eschscholzia: 12 solid carpels including 2 fertile ones, 8 sterile valve carpels. 3. Platystemon: 9 fertile semisolid carpels, 9 sterile solid (or strongly contracted valve) carpels. 4. Papaver: 10 fertile solid carpels, 10 sterile valve carpels. 5. Pseudofumaria: 2 fertile solid carpels, 2 sterile valve carpels. 6. Cheiranthus: 2 fertile solid carpels, 2 sterile valve carpels. 7. Lunaria: 2 navicular fertile semisolid carpels, 2 sterile valve carpels. 8. Capsella: 2 fertile semisolid carpels, 2 navicular sterile valve carpels. 9. Biscutella: 2 closed fertile valve carpels, 2 sterile solid carpels. 10. Epimedium: 1 fertile semisolid carpel, 1 sterile valve carpel. 11. Nandina: 1 fertile semisolid carpel, 1 sterile valve carpel. Source: Redrawn from Saunders, 1930, 1931.

trast to this, the development of the two outer buds is retarded. They form the replum, the stigma, and, if present, the gynoecial beak. In Hypecoum (without replum) the two outer buds are considered completely aborted. Motte (1946) and Guyot (1962) also accepted a polyaxial origin for the Brassicaceae flower. However, the gynoecium was said to be bicarpellate and not complex, so these authors are considered among the supporters of the nI theory. Motte turned to the 2n theory later (1957; see above), the polyaxial interpretation of the flower remaining unchanged.

VII. Morphological Support of Carpel Number Theories The morphological bases of the theories about carpel numbers are compared and evaluated in the following section. For clarity the structures and phenomena are treated separately. It should not be forgotten, however, that they are closely interconnected and, in some cases, deterrnine each other.

CARPEL NUMBER

189

A. STIGMA SHAPE

1. Papaveraceae The forms of Papaveraceae stigma are diverse (see Gonnermann, 1980; Br(ickner, 1982; Karrer, 1991). The simple gynoecia of Platystemonoideae are choricarpous in the upper part, the carpels ending in free stigmatic lobes (Figs. 3.10 & 3.11). Homologues to these free lobes are the carinal stigmatic lobes over the valves in Hypecoideae and Papaveroideae (Figs. 12.1-12.3). The more complicated stigma forms, including the stigmatic disc of Papaver (Figs. 12.6 & 12.9), are also produced by the more or less centripetally bent, carinal stigmatic lobes, with the receptive area restricted to the margins in the most advanced taxa of Papaveraceae. The touching margins of the adjacent stigmatic lobes form the papillate, double "stigmatic rays." Due to their origin, they are situated over the placental regions and are thus commonly termed "commissural." However, it must not be overlooked that the rays correspond to the margins of carinal stigmatic lobes. In addition to well-developed carinal stigma parts, several taxa possess receptive areas over the placental regions. In bivalvate taxa, these stigma parts are arranged in the median plane. Frequently, only narrow bulges connect the carinal lobes (e.g., in Chelidonium). Some taxa, however, display conspicuous lobes over the placental regions, directed horizontally (Glaucium [Figs. 12.4 & 12.5]; Argemone) or downward (Canbya [Fig. 3.7]; species of Papaver). According to the nI theory, these stigmatic parts are termed commissural. Hunnemannia has a fiat tetragonal stigma with commissural and carinal areas. The stigma of Eschscholzia is tetramerous, consisting of two long and narrow carinal lobes and two identical but usually shorter lobes over the placental regions (Figs. 12.7 & 12.8). In Eschscholzia lobbii Greene, Karrer (1991 ) observed two free stigma projections over each placental region. Also, the carinal lobes may subdivide. Saunders (1925) described a specimen ofEschscholzia californica Cham. with 19 stigmatic projections, some of them rudimentary. Pteridophyllumis the only papaveraceous taxon with no earinal stigma parts. The two elongated receptive lobes are in median position over the placental regions (see Br(ickner, 1985). In a study of the receptive surface of angiosperm stigmas, Heslop-Harrison and Shivanna (1977) demonstrated that the stigmas of all examined Papaveraceae species, representing I 1 genera (including Hypecoum), uniformly belonged to the "dry" type, with the receptive cells being papillate and producing no distinct surface secretion at maturity.

2. Fumariaceae In the Fumariaceae, the stigmas are flattened in the transverse plane (Fig. 13; cf. Ryberg, 1960; Brtickner, 1984, 1993; Lid6n, 1986). The simplest forms consist only of more or less spreading carinal lobes (Fumarieae: Discocapninae, Rupicapnos sect. Tripteryx Pugsley, Cryptocapnos; Corydaleae: Corydalis edulis Maxim.). Asymmetrical forms having a receptive part and a membraneous sterile part (Ceratocapnos [Fig. 13.3]; Sarcocapnos p. p.) are also found. The majority of taxa, however, possess more complicated stigmas. The two carinal lobes, being the primary stigmatic parts, are close together in the center of the enlarged stigma. On their outside or below them one to several receptive bulges develop from secondary growth centers (Figs. 13. l, 13.2, 13.4 & 13.5). The resulting subdivision of the stigmatic area is an adaptation for pollination, as the stigma develops in close contact with the anthers (Fig. 29.5) and is a secondary pollen carrier, loaded with pollen in the preanthetic bud. This behavior is further correlated with a nonpapillate, "wet" stigma bearing a considerable

190

THE BOTANICAL REVIEW

Fig. 12. Papaveraceae stigmas. 1. Chelidonium majus, young bilobed stigma in bud. 2. Macleaya rnicrocarpa (Maxim.) Fedde, stigma at anthesis. 3. Stylophorum lasiocarpurn (Oliv.) Fedde, stigma at anthesis. 4. Glaucium flavurn Crantz, stigma in young bud. 5. Glaucium flavurn, stigma in fruit with horizontal lobes over the placental regions. 6. Papaver dubiurn L., stigma in young bud, stigmatic lobes still not merged into stigmatic disc. 7. Eschscholzia californica, developing stigma in bud with two longer projections in the transverse plane and two shorter ones (arrow) in the median plane. 8. E. californica, stigma at anthesis with the two lobes over the placental regions being shorter. 9. Papaver bracteaturn Lindl., stigmatic disc in older bud. Unlabeled scale bars = l mm.

CARPEL NUMBER

191

Fig. 13. Fumariaceae stigmas (1,2, 4, and 5, Corydaleae; 3, 6, and 7, Fumarieae). 1. Dactylicapnos torulosa (Hook.f. et Thomson) Hutch., stigma in young bud. 2. Dicentraformosa (Haw.) Walp., stigma at anthesis, loaded with pollen. 3. Ceratocapnos claviculata, dorsiventral stigma; the dark part is receptive. 4.Adluraiafungosa (Ait.) Greene ex Britton, Stems et Poggenb., stigma in preanthetic bud. 5. Corydalis ophiocarpa Hook.f. et Thomson, stigma in older bud (from Briickner, 1992b). 6. Pseudofumaria lutea (L.) Borekh., fully developed stigma (from Brtlckner, 1992a). 7. Fumaria capreolata L., stigma in older bud. All viewed from the median plane; unlabeled scale bars = 1 mm.

192

THE BOTANICAL REVIEW

amount of a fluid secretion, as was shown by Heslop-Harrison and Shivanna (1977) for representatives of five genera from both Corydaleae and Fumarieae. In several Fumarieae taxa (Pseudofumaria [Fig. 13.6]; Fumaria [Fig. 13.7]; Fumariola, Rupicapnosp. p.) there are short, delicate projections over the placental regions between the spreading carinal lobes. Goebel (1933) considers them functionless. They are homologues of the median stigmatic lobes in Eschscholzia (Papaveraceae).

3. Capparaceae The capparaceous stigmas are mainly sessile and discoid to capitate (Figs. 14.1-14.3). Distinctly elongated carinal lobes are seen in some taxa of Capparoideae (Neothorelia, Stixis, Tirania). Pentadiplandra also possesses five short stigmatic lobes. As far as is known, the capitate stigmas of Capparoideae are formed by the tightly packed, short carinal lobes (Capparis [Fig. 14.3]; Maerua; see Karrer, 1991). In Oceanopapaver, the two carinal tips subdivide into two or three projections at the end of stigma ontogeny (Karrer, 1991). The capitate stigmas of Cleomoideae (Cleome [Fig. 14.2]; Polanisia [Fig. 14.1 ]; Dactylaena Schrad.) are produced by union of the areas over the placental regions, which gives rise to a transverse line of fusion (Karrer, 1991). Under the nI theory, this stigma form is said to represent the commissural type, though the areas over the placental regions do not overtop the areas over the valves. According to Heslop-Harrison and Shivanna (1977), the stigmas investigated so far belong to the "dry" type, being nonpapillate (Capparis; Euadenia Oliv.; Cleome isomeris) or covered by unicellular papillae (Cleome; see also Polanisia in Fig. 14.1).

4. Brassicaceae In the majority of Brassicaceae taxa, the stigmas are capitate and slightly emarginate. Acute forms are rarer; they are seen in, for example, Malcolmia R.Br., Leptaleum DC., and Diceratella Boiss. Conspicuously bilobed stigmas characterize taxa ofHesperis L. (Fig. 14.4, of. Villani, 1902a, 1902b; Schulz, 1936).), CheiranthusL., Anastatica L., Cryptospora Kar. et Kir., and Lachnoloma Bunge. (Snogerup [1967] included Cheiranthus in Erysimum. The name Cheiranthuscheiri L. is, however, commonly applied to an ornamental plant often mentioned in this paper. It is used here instead ofErysimum cheiri [L.] Crantz.) If there are stigmatic protrusions, they are in the median plane over the placental regions. Capitate stigmas (Figs. 14.7-14.10) have a transverse slit. Thus, the stigmas of the Brassicaceae are commissural. In carinal position, the papillate stigmatic tissue runs somewhat downward and forms a V-shaped area as a part of the continuous receptive bulge. For Hesperis Weberling (1981: 200) depicted true carinal lobes separated by a median slit; the papillate bulges are recurvate over the valves. This observation was not confirmed by the present author; in the Hesperis material examined, all gynoecia had well-developed, acute stigmatic lobes over the placental regions, and there was no sign of a median splitting of these lobes (Fig. 14.4). Conspicuously bicerate stigmas characterize the fruits of Matthiola R. Br. The long protuberances in the median plane arise from the backside of the lobes over the placental regions. They attain their full length after anthesis, during fruit development (Figs. 14.5 & 14.6). Heslop-Harrison and Shivanna (1977) found in representatives of 22 genera a uniformly "dry" stigma type, characterized by unicellular papillae (Figs. 14.7-14.9).

CARPEL N U M B E R

193

Fig. 14. Stigmas in Capparaceae (1-3) and Brassicaceae (4-10). 1. Polanisia dodecandra (L.) DC., capitate stigma with cleft in the transverse plane. 2. Cleome hassleriana Chodat, capitate stigma with cleft in the transverse plane. 3. Capparis spinosa, capitate stigma. 4. Hesperis matronalis L., stigmas at anthesis viewed from the median plane (left) and the transverse plane (right). 5. Matthiola longipetala (Vent.) DC., stigma in bud with two stigmatic lobes viewed from the median plane. 6. M. longipetala, stigma in fruit viewed from the transverse plane, the backs of the stigmatic lobes bearing long projections. 7. Erysimum cheiranthoides L. 8. Thlaspi arvense L. 9. Hirschfeldia incana (L.) Lagr.-Foss. 10. Aubrieta deltoidea (L.) DC. (7-10, capitate stigmas with a more or less distinct groove in the transverse plane.) Unlabeled scale bars = 1 ram.

194

THE BOTANICAL REVIEW

5. Berberidaceae Most stigmas of the Berberidaceae are of considerable size (Fig. 15; see also Citerne, 1892). In several taxa the stigmas are plicate, arising from increased growth of the upper marginal area of the gynoecium (e.g. Podophyllum [Figs. 15.1-15.4]). This peculiar growth mode causes the margin to fold into a great number of"lobes." Vancouveria has a capitate stigma, the upper margin of which is differentiated into long papillae (Fig. 15.5). In Mahonia and Berberis the stigmas are discoid, with the receptive area covering the lower margin (Figs. 15.6 & 15.7). Goebel (1933) interpreted this form as terminal stigma on the top of a strongly revolute upper margin of the carpel. The stigma of Nandina consists of three or, rarely, four lobes with dentate margins (Figs. 15.8 & 15.9; cf. Eckardt, 1937). Both "dry" and "wet" stigmas are present in Berberidaceae (Heslop-Harrison & Shivanna, 1977). Representatives of the "dry" type are the stigmas of Nandina, Plagiorhegma (both nonpapillate), Bongardia, and Epimedium (both with multicellular papillae). "Wet" stigmas are produced in Berberis, Mahonia, Ranzania (with low to medium papillae; all taxa with sensitive filaments), Diphylleia, and Podophyllum (nonpapillate and with more surface fluid).

6. Discussion In syncarpous gynoecia, free stigmatic lobes are usually thought to represent the choricarpous apical region, and their number is expected to equal the number of fused carpels. In gynoecia of paracarpous organization the position of the stigmatic lobes alternates with the position of the (sub)marginal placentae, as displayed by many bivalvate members of the Papaveraceae and some Capparaceae taxa. As early as 1825--when the theoretical carpel concept was just coming into general acceptance--Mirbel suggested the possibility of stigmatic lobes occurring over placental regions. This fact was subjected to much interpretative controversy, which dealt mainly with Brassicaceae. The nI theory was used to explain the commissural stigmatic lobes in the sense of R. Brown (in Horsfield, 1838-1852), Howell (1842), and Moquin-Tandon and Barker Webb (1848, 1849). According to these authors, each carpel originally had a bifid stigma. The neighboring stigma branches of the touching carpels then fused in pairs over the placentae. Doell (1843) remarked that the carpels "often split" in the upper half; that is, the growth of the dorsal regions of the carpels is retarded compared with the development of the fused marginal parts. The following facts may corroborate this interpretation. I. Stigmatic lobes that, in the median line of the carpel, are emarginate to bifid occur in extant taxa with parietal placentation (e.g., Salix L., with two placentae and four separate stigmatic lobes). Such a stigma type could serve as a model for a phylogenetic prestage of a commissural stigma. 2. The ontogeny of commissural stigmas is characterized by proliferating flanks of the carpels that overtop their dorsal parts (in, e.g., Aristolochiaceae: Leins & Erbar [1985b] and Weber & Biswas-Brenner [1993]; Orobanchaceae: Hecht [1990]; Boraginaceae: Heliotropium L., Hilger [1992]). However, in early ontogeny the carpel borders are distinct in these taxa, which is different in Brassicaceae (cf. section VII.E.4). 3. Weberling (1981 ) also explained the commi ssural stigma of the Brassic aceae as a result of retarded growth in the dorsal regions of the carpel tips. It causes an apparently downward course of the receptive bulge on the valves, whereas the carpel margins (described as not yet fully united in Hesperis) are elevated to form the main stigmatic area. As Saunders (1932) demonstrated for Matthiola, the development of the stigmatic papillae starts in the carinal re-

CARPEL NUMBER

195

Fig. 15. Berberidaceae stigmas (1-4, Podophyllum hexandrum Royle). 1, 3. Developing stigma (beginning of lobing) in bud. 2, 4. Strongly convoluted stigma at anthesis in longisection and viewed from above. 5. Vancouveria hexandra, stylar apex flanked by two anthers. 6. Berberis candidula C.K.Schneid., receptive rim (arrow) around periphery of stigma. 7. B. regeliana Koehne, anthers of irritated stamens touch the receptive rim. 8, 9. Nandina domestica, stigmas with three and four indented lobes. Unlabeled scale bars = 1 mm (2, 4-8) and 0.5 mm (9).

gion and then extends to the commissural protuberances, which could indicate a primary dominance of the carinal stigma parts. 4. Homeotically transformed flowers of several Brassicaceae taxa produce free carpels that show accelerated growth of stigmatic tissue in marginal position (see section VII.F. 1).

196

THE BOTANICAL REVIEW

5. The additional commissural stigmatic lobes in Papaveraceae are mostly explained as secondary outgrowths of the united carpel margins. Howell (1842) interpreted the tetragonal stigma of Hunnemannia and the four-lobed stigma of Eschscholzia as an indication of a possible increase in carpel number, which is realized in the polyvalvate taxa. According to this concept, stigmatic lobes precede their appertaining carpels (which is unsound from an ontogenetic viewpoint). The nlI theory primarily explains bivalvate taxa with stigmas in the median plane. It maintains that vascularized stigmatic lobes must be identical with carpel tips. The highly unusual dorsal placentation that has to be accepted in the concept of two median carpels is considered much less meaningful than is stigma position. Hochstetter did not carry out detailed anatomical investigations, which accounts for his conclusion (1848: 172) that the carinal stigmas of the Papaveraceae "keinen Zusammenhang mit den Klappen haben, sondern auf dem Replum sitzen" (which means that they, as a part of the replum, are a product of the dorsal carpel parts like the whole replum). Had he known that in the bivalvate Papaveraceae the dorsal veins of the valves extended into the carinal stigmatic lobes, he would have arrived at a completely different interpretation. Moreover, he considered these stigmas circular, with the parts over the valves turned upward and the parts over the placental regions turned downward; the division in separate lobes would thus be only a virtual one. Spratt (1932) restricted her nil theory to the Brassicaceae and thus avoided the problem of interpretation of stigmatic lobes over valve tips. The 2n theory, except in variant D, also maintains that stigmatic lobes are carpel tips, but it does not question, except in variant E, the existence of carpels in the valve regions. Bivalvate gynoecia with stigmas over valves as well as over placental regions (Glaucium,Eschscholzia) are unambiguously tetracarpellate in this concept. Commissural stigmas do not exist, as the placental regions are not merely zones of fusion but contain complete carpels. By reason of their narrowness these carpels must be considered blade reduced (solid); nevertheless, their stigmatic lobes may be well developed. In the 2n theory, to explain gynoecia with stigmatic lobes only over the valves or only over the placental regions, one must postulate stigma reduction of either the solid or the expanded carpel pair. The variants of the 2n theory have different interpretions of stigma morphology. Variant A: In the Brassicaceae, the valve carpels may have lost their stigmas because they are sterile (Lindley, 1828; Kunth, 1833). This hypothesis fails in Papaveraceae with welldeveloped stigmatic lobes over the valves. Dickson (1935) supposes that these carinal lobes are products of fusion, consisting of the lobe of the sterile valve carpel in the middle and onehalf of the forked lobe of the fertile solid carpel on each side. This conclusion, based on the vascular pattern in the carinal lobes, is shared by Saunders (1937 and earlier). Variant B: The stigma shape is noted, but no interpretation is given. In Eschscholzia, LindIcy (1828) considers the solid carpels probably sterile, because their stigmatic lobes are shorter. Variant C: As both carpel forms are assumed to be fertile, reduction of one stigma pair cannot be explained by sterility. Moreover, the taxa possess a compitum, so the position of the receptive organs is rather irrelevant. Variant D: For the Brassicaceae, Henslow (1880) postulated commissural stigmas over the fertile zones of fusion in the median plane. Parolinia Webb. is cited to demonstrate a possible stigma formation in the transverse plane. However, in this taxon there are no receptive transverse protuberances, only elongated valve tips that are not homologous to stigmatic lobes (cf. Schulz, 1936).

CARPEL NUMBER

197

Variant E: Restricted to the Brassicaceae, of the assumed four carpels in median position only the two outer sterile ones are thought to produce stigmas. As for the Berberidaceae, stigma shape is not used to interpret carpel numbers. A maximum of three carpels is believed to exist. Whereas the three-lobed stigmas of Nandina may provide evidence of this, the occasional four-lobed ones do not conform to this hypothesis. The multilobed stigmas of several other taxa could indicate much higher carpel numbers. Ambiguity in stigma shape does not render the 2n theory more plausible than the nl theory. This is underlined by the following three points: 1. As van Heel (1978) demonstrated for the Malvaceae tribus Ureneae, strongly reduced, sterile carpels may nevertheless end in normally developed stylodes. The possibility that solid carpels are included in stigma-bearing placental regions is at least theoretically acknowledged. However, the gynoecium ontogeny of the Malvaceae differs fundamentally from that of the families under discussion. 2. The papillate stigmatic tissue forms a continuous zone. Thus, over the placental regions are receptive areas that can be interpreted as reduced stigmas of solid carpels. Stigmatic tissue circling the style top is, however, formed in many syncarpous gynoecia that do not terminate in free stylodes (e.g., Tulipa L.). This character in combination with solid carpels leads to Saunders's extreme interpretations of gynoecia. 3. The differing positions of the distinct stigmatic lobes (over valves and over placental regions; over valves only; over placental regions only) do not permit the conclusion for 2n carpels without additional hypotheses (different local reductions). In the cases studied, stigma shape does not give unambiguous information about carpel numbers. An extension of the stigmatic area, for example, by folding (Berberidaceae) or secondary formation of further receptive bulges (Corydalis and other Fumariaceae taxa), must be interpreted in terms of pollination biology. Thus, the 2n theory lost important support, as its creator Lindley (1828) based it on the stigma position of Brassicaceae and Eschscholzia. Saunders's theory is not consistent regarding stigma shape, and the stigmatic lobes are as variable as the whole carpels. The occasional splitting of the carinal lobes in Eschscholzia is taken as evidence of nine modified carpels in each valve. Otherwise, depending on the "division of labor" between solid and valve carpels, stigmatic lobes may be present or absent in all carpel types, and they may subdivide and fuse in parts with adjacent carpels. Finally, the 3n theory is not supported by stigma shape. B. ZONES OF DEHISCENCE The demarcation of the valves from the placental regions is given by the zones of dehiscence (Figs. 16 and 17; for the specific case of Berberidaceae, see below). These zones are discernible on the surface of the maturing fruit by a more or less deeply sunken outer epidermis, with conspicuously diminished cell size. The same structure is found in the inner fruit epidermis. In the outermost valve edges, there is a separation tissue consisting of one to three radial cell layers (Fig. 16.4). In the majority of fruits, these parenchymatous cells are much smaller in diameter than are the adjoining cells that belong to the valves or to the placental regions (Fig. 16.8; cf. Berg, 1972). They are elongated longitudinally and have delicate walls, and frequently there are wide spaces between them (Fig. 16.3). In Glaucium (Papaveraceae), the cells in the zones of dehiscence contain large calciumoxalate druses (Figs. 16.1 & 16.2). As Meakin and Roberts (1990a, 1990b) demonstrate in Brassica napus L., dehiscence is efletted by enzymatic autolysis of the cell walls within the separation tissue. Sclerenchymatous

198

THE BOTANICAL REVIEW

Fig. 16. Anatomy of zones of dehiscence. 1,2. Glauciumpulchrum Stapf (Papaveraceae), transection of placental region in fruit and enlarged detail; cells in the zones of dehiscence (arrows) contain calcium oxalate druses (dark bodies). 3. Corydalis cava (L.) Schweigg. et K6rte (Fumariaeeae), margin of a detached valve; intact inner epidermis (below), with some intercellular spaces in its marginal strip (middie), borders spongy separation tissue (above) (from Brtickner, 1993). 4. Dactylaena micrantha Schrad. ex Schult. et Schult.f. (Capparaeeae), transection of detached valve; zone of dehiscence low (bracket), consisting of small-celled tissue. 5. Alliaria petiolata (M.Bieb.) Cavara et Grande (Brassicaceae), fruit transection; protruding placental region with detached (left) and persistent valve (right). 6. A. petiolata, zone of dehiscence (arrows) along the sclerified valve edge. 7. l/ancouveria hexandra (Berberidaceae), fruit transection; zones of dehiscence (arrows) distant from placental region s.str. 8. I/. hexandra, zone of dehiscence with small-celled tissue of few layers. Scale bars = 0.1 mm and 1 mm (1,7).

CARPEL NUMBER

199

Fig. 17. Course of zones of dehiscence (bold black lines) in schematic, "unrolled" fruits of the families concerned, a, h, j-I. Valves detach completely but may be smaller, b, e, f, i, n, o. Valves detach incompletely, d, e, p. Detaching valve tips cause poricidal dehiscence, g. Forming of lateral slits. m. Unilateral dehiscence. For further explanations, see the text.

200

THE BOTANICAL REVIEW

areas adjoining the separation tissue (fiber caps of the placental venation; sometimes lignified valve edges [Fig. 16.6]) act as resistance layers that facilitate hygrochastic rupture of the fruit. The extension of the zones of dehiscence along the fruits varies within and between the families.

1. Papaveraceae In some taxa of Chelidonieae there are two zones of dehiscence, each turning back into itself, which have their upper and lower peaks below the stigma or the short style and close to the receptacle. The two valves "cut off" from the pericarp by these encircling zones detach completely (Fig. 17a; a type of"vollkommene Loehkapsel" [perfectly aperturate capsule] or "Schote" [siliqua], according to Winkler, 1940; "Fensterkapsel" [fenestrate capsule], according to Stopp, 1950; of. Endress, 1995: figs. 4D & 5). The valve dorsal veins are disrupted at the base and the tip of the valves by dehiscence. The valves occupy nearly the whole pericarpial area in these cases; the style-bearing replum persists as a fragile, narrow frame. In other Chelidonieae taxa, as well as in Eschscholzieae, the valves detach nearly completely but, in acropetal dehiscence, remain attached to the replum apically or, in basipetal dehiscence, adhere to the receptacle basally (Figs. 17b & 17c). Correspondingly, the valve dorsal veins tear off only at the top or the base of the valve. These "fast vollkommene Lochkapseln" (nearly perfectly aperturate capsules), sensu Winkler (1940), are not siliquae s.str., according to this author. In the Papavereae, there is a tendency toward shortened zones of dehiscence restricted to the upper half of the fruit. Completely demarcated valves are not seen; instead, only their apices are discernible. In dehiscence the apices bend back basipetally, thus opening more or less wide apertures (Fig. 17d). In the extreme, only the upper margins of indiscernible "valves" curve and form small pores below the style or stigmatic disc (Fig. 17e), enabling fractionated seed dispersal. Winkler (1940) calls this fruit type "unvollkommene Lochkapsel" (imperfectly aperturate capsule) leading to a perfectionated poricide capsule. As the valve dorsal veins are strongly reduced or absent in the taxa producing such fruits, dehiscence of the valve tips is not hindered.

2. Fumariaceae The zones of dehiscence are not completely circular. Dehiscence starts with lateral rupturing along the placental regions and progresses toward the style. In basipetal direction the valves may separate to the receptacle, and their bases detach more or less tardily, rupturing the valve dorsal veins (Fig. 170. The dehiscence may also be restricted to the upper third of the fruit (see Berg, 1969). The valve tips are not demarcated by a special separation tissue (Fig. 20.1). The valve dorsal veins, running into the style, additionally increase the stability of the tissue, so that the valve apices remain connected to the pericarp. If the valves do not detach apically and/or basally, the seeds are scattered through the lateral slits (Fig. 17g), which widen by rapid shrinking of the valve tissue.

3. Capparaceae Acropetal dehiscence with complete detachment of the valves as well as basipetal incomplete dehiscence occur in the Cleomoideae (Figs. 17h & 17i; cf. Ernst, 1963). According to

CARPEL NUMBER

201

Stopp (1950), there is an association between erect fruits and basipetal dehiscence, and another between pendulous fruits and acropetal dehiscence. The horizontally directed fruits of Cleome spinosa Jacq. are described as opening first along their adaxial side, followed by the opposite one. This phenomenon is not only due to more intensive desiccation of the upper, sun-exposed side. The anatomical background of a dehiscence without formation of a replum (Buhsia, Podandrogyne spp.) is still insufficiently known.

4. Brassicaceae In the majority of taxa, the fruits have encircling zones of dehiscence and open by completely "cut off" valves. Valve size is, however, very variable. Initially, the valves cover nearly the whole pericarp, and the replum is narrow. A morphological series of basipetal valve reduction can follow (Figs. 17j-171; cf. Zohary, 1948b; Stopp, 1950), with the final stage being minute, functionless, hardly discernible "valves" at the fruit base. In contrast to Papaveraceae, the dehiscent part of the fruit is progressively restricted in the basipetal direction. Furthermore, the seeds included in the indehiscent fruit part (stylar segment, beak, "Schnabel") cannot be liberated. In the Brassiceae, a peculiar, highly specialized region of dehiscence forms a jointlike connection between the upper valve margin and the beak base. This structure, termed "saddle articulation," with its components "inner" and "outer cog" (toothlike bulges at the valve tip), hollow "saddle groove" between these bulges, and "swivel head" at the beak base, fitting in the saddle groove, is responsible for the delayed detachment of the valves. It is described in detail by Bauch (1940-1941). A rare mode of dehiscence is realized in Leptaleumfilifolium (Zohary, 1948a). The dorsiventral fruits adopt a horizontal position and open along the adaxial side only. In contrast to comparable eases in the Capparaceae (see above), only the upper fruit side has a separation tissue between the placental region and the valve edges, whereas in the lower side a continuous sclerenchyma zone extends over the placental region and the indiscernible "valve edges" (Fig. 17m).

5. Berberidaceae In the valve-forming taxa, the zones of dehiscence do not directly flank the placental region but are differentiated at a considerable distance to it in the pericarp. For that reason, the broad, style-bearing fruit part with the placenta along its median line doe: not look like a "replum." However, this is only a matter of proportions. Whereas the fruits that dehisce more or less to the base (Figs. 7.2, 7.3, & 17n; cf. Berg, 1972; Endress, 1995: fig. 5) are easily comparable to the "nearly perfectly aperturate capsules" discussed above, the "Rachenbalg" of Jeffersonia opening with a transverse slit (Figs. 7.4 & 17p; Endress, 1995: fig. 4C) corresponds to a poricide capsule with just one pore.

6. Discussion A transverse section through a bivalvate fruit with broad placental regions demarcated from the valves by distinct zones of dehiscence strongly suggests the existence of four carpels. Saunders (1923: 456) wrote: "One feature.., which has always appeared to me difficult to reconcile with the view generally held [= nI theory, C. B.], is the considerable and often

202

THE BOTANICAL REVIEW

varying width of the tract of tissue (commissure) separating the valve boundaries." Her view is shared by the majority of the supporters of the 2n theory. Thus, dehiscence zones that cause the formation of valves are given great importance. But do these zones, encircling in many taxa, really delimit complete carpels? Or are the base and the tip of these carpels formed of tissue outside the valve? Lindley (1828), the "father" of the 2n theory, gave an ambiguous answer to that question in defining the Brassicaceae fruit: "Pericarpium formed of four confluent pistilla, of which two are placentiferous and furnished with stigmata, and two destitute of placentae and stigmata, but separable in the form of valves." This conclusion was drawn in comparison with the fruit of Eschscholzia (Papaveraceae), whose four "pistilla" (carpels) were all interpreted as stigma bearing but were, on principle, homologous to those composing the Brassicaceae siliqua. Thus, Lindley was aware of the fact that sterile carpels could end in (styles and) stigmas. However, these stigmas are not constituents of the detaching valves but are borne by the persisting replum. Consequently, the hypothetical "valve carpels" are more than the valves, for part of their tissue is also integrated in the replum. Unfortunately, explanations as imprecise as Lindley's still appear in the Brassicaceae literature (e.g. Jacob et al., 1994: 324; "Die beiden fertilen Fruchtbliitter bilden einen stehenbleibenden Rahmen, von dem sich die beiden sterilen ablOsen"). More logical interpretations of the nature of the zones of dehiscence have been given by several botanists. Prior to Lindley's paper, Lestiboudois (I 823a), unacquainted with the theoretical term "carpel" and considering the placentae independent from the valves, constructed the following morphological series within the Papaveraceae lineage: siliqua with complete dehiscence (Chelidonium)--capsule with incomplete dehiscence (Argemone)--poricide capsule (Papaver)-indehiscent fruit (Fumaria). Lestiboudois emphasized homology between these fruit forms and the cruciferous siliqua. To him, the existence of several (probably four) structural components was not necessarily reflected in the presence of valve-forming zones of dehiscence. Brassai (1838) gave the corresponding morphological series for the Brassicaceae. Beginning with siliquae with a narrow replum and large valves (Thlaspi), he demonstrated the tendency of basipetal diminution of the valves with a simultaneously enlarging beak and vigorously denied a homologization of the beak with a style, as "ein Griffel mit Samen ein Unding ist" (p. 312) to the complete absence of valves (via Raphanus L. to Bunias L.). Also, Brassai emphasized the identical organization of all cruciferous fruits, dehiscent or not. To him, the valves are "'assumenta' (gleichsam angeflickte Stticke)" (p. 313) and have no carpel identity at all. More than a hundred years later, Zohary (1948b) repeated the interpretation. He termed the valves "opercular structures" and, like his successors Eigner (1973) and Zimmerli (1973), pointed to the late differentiation of the separation tissue in gynoecium ontogeny. Gynoecial length may exceed 1 mm or more until the zones of dehiscence become visible. Parkin (1926) stressed that in increasingly syncarpous gynoecia the border lines between the individual carpels become more and more unrecognizable and that the zones of dehiscence are no longer fixed to these zones of carpel fusion but may adopt manifold courses. Concerning carpel numbers, Brown (in Horsfield, 1838-1852:111) stated that "an opinion founded on dehiscence only may be said to be a mere begging of the question." The interpretations of the course of dehiscence that are given by the individual theories can be summarized as follows (Fig. 18). 1. The nI theory: The zones of dehiscence form a line (often turning back into itself) within the blade of each carpel. Correspondingly, dehiscence is loculicidal (Winkler, 1940; Stopp, 1950). Because the longitudinal detachment of the valves frequently takes place in the carpel margins along both sides of the placental regions, Stopp (1950) spoke of "Plazentifragie"

CARPEL NUMBER

203

Fig. 18. Relationship of the valve-bordering zones of dehiscence to the assumed carpel borders according to the different theories, shown in a schematic Iongisection of an "unrolled" bivalvate gynoecium (valves, white; replum, including placental regions, dotted; ovules, white circles; postulated carpel borders, bold black lines).

(placental fragility) as an extreme case of loculicidy. In the Berberidaceae, the irregular ruptures in the upper part of the Bongardia fruit are also included in the loculicidal type of dehiscence. 2. The nII theory: This concept also maintains loculicidy, but valve separation takes place along the fertile median parts of the carpels. Spratt (1932) emphasized that such dorsicidal dehiscence is comparatively frequent in angiosperms. The circular zones of dehiscence would then cross carpel borders. Comparable compound valves are known in, for example, Campanulaceae (cf. Winkler, 1940). 3. The 2n theory, variant A: Many of the earlier authors interpret the zones of dehiscence as zones of fusion between fertile and sterile carpels. Lestiboudois (1823a,b) explained the continuous shortening of the zones of dehiscence in Papaveraceae by increasing fusion of placentae ("cordons pistillaires") and valves, the organization of the fruits remaining un-

204

THE BOTANICAL REVIEW

changed. Whether the sterile carpels form part of the style and beak is for the first time unambiguously answered in the affirmative by Eigner (1973). Consequently, he concludes (p. 398): "Die Trenngewebe stellen keine exakten Karpellgrenzen dar, was aufjeden Fall ffir die obere Abl6sungsstelle unterhalb des Griffels bzw. des Schnabels gilt"; the zones of dehiscence seem to have become "independent" in a certain manner. The authors who base their concepts on the vascular pattern only do not evaluate the course of dehiscence. 4. The 2n theory, variant B: If there are median sterile carpels, their borders are located within the replum. The zones of dehiscence then represent secondary submarginal structures of the two expanded fertile carpels; dehiscence would be loculicidal. 5. The 2n theory, variant C: Because the placentae are considered zones of fusion between carpels with marginal placentation, the carpel borders, as in variant B, are within the replum, but they are located in the region of ovule attachment. The zones of dehiscence run within the blade of the two expanded carpels. 6. The 2n theory, variant D: According to Henslow (1880), the zones of fusion of the four carpels are situated exactly in the median plane and in the transverse plane; all four carpels are of equal dimension. Loculicidal dehiscence takes place across the transverse zones of fusion. The carpels are asymmetrical, with a one-sided placenta, and a sterile crescent is formed by the zone of dehiscence. 7. The 2n theory, variant E: Similar to the nlI theory. Gerber (1899a) also postulated loculicidal dehiscence that, because the carpels are in the median plane, crosses over the transverse carpel borders. The zones of fusion with the enclosed addorsed carpels are not contacted by the zones of dehiscence. 8. The 3n theory: The concepts of Gerber (I 899b, 1899c, 1899d, 1900a, 1900b) and Yen (1959) differ concerning the zones of dehiscence. According to Gerber, the zones of dehiscence equate the zones of fusion between the four sterile carpels that form the fruit wall (valves and outer replum). According to Yen, in fruit the four carpels are of equal dimension, and dehiscence is loculicidal. Vertically, the zones of dehiscence flank the middle axes of the median carpels and, at the base and top of the valves, run horizontally across both transverse carpels. Martel (1902) defined the valves as sterile carpels; thus the zones of dehiscence are their edges. Delimitation of the four tripartite phyllomes that are believed to take part in gynoecium construction remains obscure. This concept is also primarily based on the vascular pattern. 9. The theory of carpel polymorphism: Saunders argues (1925: 133; 1926:301) that the course of dehiscence may be identical with carpel borders in some cases and need not be in others. The first case would be true ofgynoecia formed of two sterile valve carpels and two fertile solid carpels. An important characteristic attributed to a solid carpel is its external delimitation by a"double-line suture," the sunken epidermis over the zones of dehiscence along the valves. However, "zone of dehiscence = carpel border" applies only to the vertical sections and not to the basal and apical delimitation of the valve. The valve carpels take part in style and stigma formation. On the other hand, in Berberidaceae the zone of dehiscence deiimitates the complete valve carpel which is shorter than the fertile semisolid carpel and does not take part in style and stigma formation. In several other cases, dehiscence zones are considered secondary separation tissues that are independent of carpel borders. For example, the valvate fruits of Eschscholzieae are explained as having nine carpels per valve surrounded by only one circular zone of dehiscence. The valves of the cruciferous siliculae are said to consist of a strongly reduced valve carpel and the lateral parts of median semisolid carpels. Saunders's interpretation of the zones of dehiscence is rather arbitrary.

CARPEL NUMBER

205

As discussed above, it is clear that the course of dehiscence does not necessarily delimit carpel borders in valvate fruits, invalidating another aspect of the 2n theory. Independence of dehiscence zones is also shown by the monocarpellate pod of Carmichaelia R. Br. (Fabaceae), which is strongly reminiscent of the capsule of Bocconia (Papaveraeeae). A circular zone of dehiscence differentiates within the carpel blade on each side between the dorsal vein and the placenta, giving rise to two valves. After the valves are shed, a style-bearing replum remains that consists of the dorsal region and the placentiferous ventral region of the carpel (Fahn & Zohary, 1955). A siliqua-like capsule is thus produced by a single carpel. If one accepts the zones of dehiscence in valvate fruits as merely secondary separation tissue, how can the existence of sterile carpels be demonstrated? This hypothesis will then rest exclusively on the vascular pattern in the gynoecium (see section VII.D.6). C. STRUCTUREOF PLACENTALREGIONS In this article, the term "placental regions" refers to the placentiferous parts of the gynoecium or the pericarp, respectively, that in the families under discussion often have a remarkable width and a characteristic structure. Except in Berberidaceae, they are bordered by the zones of dehiscence; thus the placental regions form the replum. As discussed above, secondary separation tissues should not been taken a priori for organ-delimitating lines and, furthermore, may be present only in parts of the fertile ovary region. Therefore, the placental regions are not sharply bordered areas. Nevertheless, their peculiar structure is described and discussed here.

1. Papaveraceae In some bivalvate taxa, the placental regions are rather broad and have their own area of outer epidermis that produces the same trichomes as the outer valve epidermis (Fig. 19.5). The placental regions are usually sunken along the valve edges or may be convex, in some cases projecting beyond the level of the valve edges. The zones of dehiscence are then distinct on both sides of the placental region. Deeply sunken placental regions are triangular in transection and have on their outer edge only few rows of small epidermal cells; seen from outside, the double zone of dehiscence appears to be a single slit (Stylophorum diphyllum). In extreme cases the replum has no outer surface and is covered by the two fused valve edges. In transection, the outer pericarp has only one radial zone of dehiscence which bifurcates centripetally, like an upside down Y, and "cuts out" the inner part of the pericarp (vascular tissue, placenta) to form the replum (Roemeria of the Papavereae; Briackner, 1982). Voluminous placental regions are found in Glaucium (Fig. 16.1), Dicranostigma (Figs. 19.1 & 19.2), Sanguinaria, and Stylophorum lasiocarpum (Oliv.) Fedde (Figs. 19.4, 19.5, & 19.7). Their venation consists of several bundles that are in an arc or ring in the replum, with the inner ones usually inverted. In the more or less ripe fruit, this placental bundle system has, at least on the abaxial side, separate or confluent areas of sclerenchymatous fibers. The fiber caps may extend to the outer epidermis in some taxa. In narrower placental regions, the ring members are closely adjacent or, in the polyvalvate Papavereae, form a condensed placental bundle system represented by a strong concentric bundle. A single collateral bundle in the placental region occurs in few- to one-seeded gynoecia (Macleaya,Bocconia). There are connections between placental and valve venation, which is described in more detail in section VII.D.

206

THE BOTANICAL REVIEW

Fig. 19. Placental regions in Papaveraceae. 1, 2. Dicranostigma erectum K.-F.G~nther, part of fruit transection, and placental region enlarged. 3. Chelidoniummajus, young placental region with differentiating ovules. 4, 5. Stylophorum lasiocarpum, longisection and transection of fruit. 6. C. majus, transection of young placental region with ovules still undifferentiated. 7. SGlophorum lasiocarpum, transection of placental region in fruit. 8. Papaverdubium, Iongisection ofpreanthetic gynoecium, polyovulate placental regions deeply intruding. 9. Meconopsis cambrica, transection of preanthetic gynoecium with six polyovulate, intruding placentae. Scale bars on the stcrr = 1 ram; on the light micrographs = 1 mm (1, 7) and 0.1 mm (2, 6).

CARPEL NUMBER

207

Usually, the placentae consist of loose parenehyma enclosing wide intercellular spaces. In polyvalvate taxa especially large air spaces are formed within the placentae that deeply intrude into the locule. Gu6d6s (1979: 221) compared such bulky placentae with "some sort of marginal callus" of the carpels. They seem not to be "continuations of the carpel blade itself," though they appear early in gynoecium development (Cass & Fabi, 1990). Romneya has a multichambered gynoecium, as the placentae are congenitally fused in the center. The Glaucium placenta undergoes a peculiar postfertilization development. Its parenchyma produces a bipartite outgrowth that, in the center of the locule, more or less tightly touches the outgrowth of the opposite placenta, forming a false septum (Fig. 16.1). The placental epidermis differs distinctly from the inner valve epidermis in the shape and orientation of cells. Often the cells are vesicular to papillate, which is characteristic oftransmiring tissue. In Glaucium, however, there are occasionally some recognizable rows of cells between the two parts of each outgrowth that look like a strip of inner valve epidermis. Possible interpretations of this phenomenon are discussed below. In many-seeded gynoecia the ovules are inserted in one row (Macleaya cordata) or two to several rows (up to 20 rows in Papaver [Fig. 19.8; Karrer, 1991; Endress, 1995: fig. 4F]), depending on the width of the placenta (Figs. 19.1, 19.3-19.5, 19.8, & 19.9). The number of rows has been counted in transverse cuttings of the gynoecia; whether the ovules are really inserted in distinct vertical rows or scattered over the placental surface is not sufficiently known (see irregularities in Chelidonium [Fig. 19.3], and Sanguinaria [Lehmann & Sattler, 1993]).

2. Fumariaceae The placental regions are very similar to those in Papaveraceae (Figs. 20.1-20.6). They are relatively wide in Dicentra; correspondingly, the placental bundle system consists of several components. Adlumia and Corydalis sect. Duplotuber Ryberg (Fig. 20.4) are furnished with conspicuously convex placental regions that, on the outside, protrude over the valve edges. Frequently the placental regions are deeply sunken and have a narrow back with a reduced or lacking outer epidermis (Corydalis [Fig. 20.6]; Pseudofumaria). In Corydalisuniflora (Sieber) Nyman the placental regions lie within the pericarp and are completely covered by the fused valve edges. In transection, the zone of dehiscence has the form of an inverted Y (Briickner, 1993; cf. Roemeria of the Papaveraceae). A similar situation is apparently present in Daetylicapnos scandens (D. Don) Walp. and Cysticapnospruinosa(Bemh.) Lid6n (Fukuhara & Lid6n, 1995). The placental bundle systems are often represented by strong concentric bundles that give off inner (inverted) sectors as ovular traces and then permanently reconstitute. They may be surrounded by a closed or discontinuous sclerenchymatous sheath or accompanied by an abaxial fiber cap, respectively. Except in Dicentra, the placentae hardly intrude into the Iocule; nevertheless, they mostly consist of very spongy parenchyma with large central air spaces (Fig. 20.5). The formation of these spaces starts before fertilization (Zimmerli, 1973). Rarely, the ovules are inserted in several rows (Dicentra [Fig. 20.2]; Dactylicapnos sect. Dactylicapnos; Cysticapnos vesicaria; Prain, 1896; Fukuhara & Lid6n, 1995); in most taxa one or two rows of ovules are formed (Fig. 20.3).

3. Capparaceae In dehiscent gynoecia, most placental regions are strongly convex and, in transverse section of the fruit, distinctly thicker than the valves, but not very broad (Fig. 20.8). Their outer

208

THE BOTANICAL REVIEW

Fig. 20. Placental regions in Fumariaceae (1-6) and Capparaceae (7, 8). 1. Corydalis cava, dehisced fruit with the valves remaining attached to the replum apex (most of the seeds shed). 2. Dicentraformosa, longisection of gynoecium showing placenta. 3. Corydalis ophiocarpa, longisection of gynoecium showing placenta (from Brfickner, 1992b). 4. C decumbens (Thunb.) Pers., transection of protruding placental region (from Brtickner, 1993). 5. C pallidiflora, transection of sunken placental region (from Brtiekner, 1993). 6. C pallida (Thunb.) Pers., transection of the deeply sunken placental region with almost no outer epidermal area. 7, 8. Polanisia graveolens, gynoeeium in longisection (placental view) and transection, showing conspicuously protruding placental regions. Unlabeled scale bars = 1 mm.

CARPEL NUMBER

209

epidermis shows the same indument as the outer valve epidermis. In the majority of taxa, the vascular bundles form more or less perfectly concentric placental bundle systems. These bear sclerenehymatous fiber caps on the outer and inner sectors, the inner one directed toward the placenta being distinctly weaker (Fig. 21.1). Briquet (1914) observed a ringlike configuration consisting of a strong outer bundle with a fiber cap, two lateral bundles with phloem accompanied by fibers, and two weak, inverted placental bundles. The central tissue between the outer and lateral bundles is described as sclerenchymatous. Furthermore, divisions of the vascular tissue into two components lying, in transection, side by side or on the same radius (in the latter ease, the inner part is inverted) have been observed (Stoudt, 1941; Puri, 1945). In several taxa the placentae intrude deeply into the ovary center. Their extended tips are sterile. At the time of ovule initiation, these tips fuse postgenitally (surface fusion according to Sattler, 1977) and thus form continuous septa. The fusion takes place either basally and apically (Courbonia, Maerua, Morisonia, Oceanopapaver) or along the whole length of the placentae (Cadaba ssp.; Capparis as in Fig. 21.2; Steriphoma). Concurrently, schizogenous air spaces develop in the septum parenchyma that destabilize the tissue (Leins & Metzenauer, 1979). As the accrescent seeds develop, the unstable septum of Crateva ruptures after anthesis, making the fruit one chambered again (Raghavan & Venkatasubban, 1941a; Narayana, 1965). Ruptured septa have also been observed in Capparis (Narayana, 1962). On each placenta, the ovules are borne in a few (often two; Fig. 20.7) rows that are inserted close to the ovary wall. About six rows are reported for Cladostemon A. Braun et Vatke (Fig. 5.11B); in Crateva and Capparis up to ten rows are formed, the whole surface of the placentae except their central parts bearing ovules. In Oceanopapaver the intruding septa not only consist of placental parenchyma but also contain the placental bundle systems; the ovules are inserted at a distance from the ovary wall, suggesting axial placentation (see Karrer, 1991).

4. Brassicaceae Frequently, the placental regions are broadly convex and in several cases project considerably over the valves, and the valve edges (radial zones of dehiscence) are narrow (Figs. 16.5 & 16.6). Deeply sunken placental regions without a separate outer epidermis are also not uncommon. An extreme is reached with the placental regions being completely covered by the fused valve margins and the zones of dehiscence forming an inverted Y, as previously mentioned for papaveraceous and fumariaceous taxa. Leclerc du Sablon (1884) frst described this phenomenon in Sisymbrium acutangulum DC. Briquet (1915) reexamined the species and found a broad replum covered by its own epidermis; so Leclerc's material was probably incorrectly determined. The zone of dehiscence shaped as an inverted Y, running radially to the placental bundle system and then forking, was demonstrated in Camelina sativa (L.) Crantz by Briquet (1915) and in C microcarpa Andrz. ex DC. and Nesliapaniculata (L.) Desv. by Eigner (1973). In several taxa the valve margins appear "hitched" to the flanks of the placental regions, so that the latter look somewhat pear shaped in cross-section (e.g., Lepidium L., Nasturtium R. Br.). The venation of the placental regions consists of a strong, normally oriented bundle and one or two inner inverted veins that give offthe ovular traces. In fruit the outer bundles bear a selerenchymatous fiber cap, and the tissue in the center of the placental bundle system becomes lignified. The development of the bundle configuration is described in section VII.D.4. The placentae form a septum by radial intrusion and postgenital fusion in the center of the ovary. In the majority oftaxa the ovary is perfectly two chambered. Incomplete (Fig. 21.3) or rudimentary septa are seen in about 30 genera (Endress, 1992). The chloroplast-containing

210

THE BOTANICAL REVIEW

Fig. 21. Placental regions in Capparaceae (1, 2), Brassicaceae (3-5), and Berberidaceae (6-8).

1. Cleome hassleriana, transection of placental region in fruit (xylem of concentric bundle system, dark; sclerenchymatic fiber caps, bright). 2. Capparis spinosa, transection of preanthetie gynoecium, seven intruding placentae fused postgenitally in the center. 3. Cheiranthus cheiri, fruit with front valve removed; septum partly discontinuous along the middle. 4. Thlaspi arvense, transverse longisection of gynoecium, placental view with septum halved. 5. T. arvense, transmitting tissue within the septum very loose and vanishing, the epidermises persisting as a double layer (pollen tubes penetrate the epidermises and grow toward the ovules). 6. Podophyllurn hexandrum, longisection of gynoecium showing bulky placenta. 7. Epimediurn alpinum L., longisection of gynoecium with ovules inserted in two rows. 8. Mahonia aquifolium (Pursh) Nutt., longisection of gynoecium with subbasal placentation. Scale bars on the stereomicrographs = 1 ram and on the light micrograph = 0.1 mm.

CARPEL NUMBER

211

placental parenchyma extends into the septum and, in its center, is converted to transmission tissue. Large air spaces develop in the parenehyma before fertilization (Fig. 21.5); the tissue fades, and in the fruit the epidermises of both sides of the septum touch. Stomates are found near the gynoecium wall. For further details of septum structure, see Hannig (1901). On each placenta the ovules are inserted parietally, forming two rows that are separated by the septum (Fig. 21.4).

5. Berberidaceae The typically broad placental region has a slightly convex to concave back and, seen from outside, is not very distinct from the ovary wall or pericarp, respectively. Its tissue is usually loosely parenchymatous. The placental bundle system frequently consists of many distinct bundles in a ring or, more rarely, of a single strong bundle. At the periphery a faint sclerenchymatous zone may be formed. The wall-borne placenta protrudes more or less deeply into the locule. In the many-seeded taxa, the ovules are attached in two (Fig. 21.7) to several rows (up to ten in Podophyllum; Fig. 21.6; Endress, 1995: fig. 4E). Subbasal to basal placentation (Fig. 21.8) is secondarily derived from wall-borne placentation (Terabayashi, 1985a). The placental parenchyma is interspersed with large air spaces, as is the loose pericarpial tissue below the horizontal vascular bundle zone.

6. Discussion In considering carpellary composition of the gynoecium special attention has always been paid to the erueiferous septum, though septa are also found in the gynoecia of Capparaceae and Papaveraceae taxa. The beginning of carpel-theoretical interpretations is marked by de Candolle's (1813) "Th6orie 616mentaire," in which a peculiar gynoecium member common to Nymphaeaceae, Papaveraceae, and Brassicaceae is postulated. This hypothetical pistil, called "siliquelle," is said to never occur singly and to be composed of three longitudinal parts. The middle part is sterile, whereas the two lateral parts have laminar placentation on their inner surface distant from the margin. Two such siliqueiles fused along the outside of their marginal parts would produce a more or less completely two-chambered gynoecium with intervalvular placentae; each placenta then had a double nature. Later, de Candolle (1821a) wrote of two united "carpels," which are not fully congruent with the siliquelle. The ovules are now considered margin borne. In the Brassicaceae, the united placentae are separated by a double septum membrane that is thought to originate by centripetal extension of the outer carpel epidermis (epicarp). Lestiboudois (1823b) objected to the idea that the Brassicaceae siliqua is a product of fusion of two complete pistils. According to him, each of the pistils must have been bivalvate and the septum thus produced by fusion of two valves, one from each of the adjoining pistils. Lestiboudois rejected this concept because of floral symmetry and comparison with polyvalvate fruits of Papaveraceae. The theoretical background of this strange-sounding discussion is that de Candolle (1813) believed the Brassicaceae flower to have originated by condensation of three flowers reduced to varying degree and, furthermore, that the theoretical term "carpel" had been introduced only recently and was still not clearly defined. Lestiboudois guessed correctly (1823b: 197) that in this case de Candolle had in mind fusion not of complete pistils but of pistil elements. Lestiboudois also discounted the idea that the septum consists of two bent and fused carpel margins. He argued that there are no signs of a fusion of two parts (notch, suture) in the middle

212

THE BOTANICAL REVIEW

of the placental regions, that the dehiscence is septifragous instead of septicidal, and that placentation would be axial when carpel margins are bent inward. Septum origin by epicarp extension was not discussed. Lestiboudois considered the placentae (cordons pistillaires) independent structures that form the replum and, by radial extension, the septum. The zone of contact of the two outgrowths is marked by a visible longitudinal line along the middle of the septum in many taxa. Lestiboudois's response to de Candolle began the debate on gynoecium organization in the Brassicaceae. According to Mirbel (1825b), the Brassicaceae fruit is formed by two "valves" with united margins. The placental venation is connected with the suture of the valves. The loculedividing septum is a placental outgrowth, and, by reduction, a siliqua of the papaveraceous Chelidonium type arises. Furthermore, Mirbel (1825a) pointed out that the Brassicaceae septurn is formed in the young ovary, whereas the septum of Glaucium (Papaveraceae) develops only after fertilization and thus evidently is a different character. Brown (1826) emphatically claimed authorship of the bicarpellary interpretation of the cruciferous gynoecium, published in 1817, and he deplored de Candolle and Mirbel for not citing his paper. In his explanation the septum is a bilamellar outgrowth of the ovary wall. Like Lestiboudois, he mentioned a longitudinal "vein" in the middle of the septum of many taxa. This vein laterally gives off"tubes" that run to the margins of the septum (especially conspicuous in Farsetia Turra), which recalls the venation of a leaf. The apparent occurrence of a septum vein caused Bernhardi (1838a, 1843) to doubt the concept that a placental outgrowth ("dissepimenturn spurium") produces the septum. Fournier (1864, 1865a, 1865b) was also of the opinion that bundle-like as well as laticifer-like structures occur in the septa of several taxa, though several years earlier Treviranus (1847) had correctly interpreted the cords as transmitting tissue extending to the stylar canal. Later, Capus (1878), Hannig (1901), and Calestani (1917) demonstrated the transmitting nature of this fibrous tissue, which is free of vascular elements. That the Brassicaceae septum is not merely a double lamella (as was supposed by de Candolle [ 1821 a] and Brown [ 1826] and depicted by Lindley [ 1828]) was demonstrated by Tr6cul (1843). He gave a histological description of the placental intrusions and showed that the apparently bilayered construction is due to the disintegration of the loose parenchyma between the two epidermises. Furthermore, he detected stomates in the epidermal area close to the ovary wall. Schleiden (1843), Treviranus (1847), and Hannig (1901 ) confirmed Tr6cui's observations. Despite increasing knowledge of septum morphology, no universal interpretation of its nature emerged. The following possibilities are taken into consideration by the supporters of the nI theory. 1. The septum is a double structure, formed by outgrowths of both parietal placentae. This is the most common opinion. 2. Authors who considered the placentae axes also ascribed an axial nature to the septum (Agardh, 1836; St.-Hilaire, 1841; Schleiden, 1843; Treviranus, 1847; Payer, 1857). Treviranus compared the increasing schizogeneous cavities to the central medullar cavity of axes. Fournier (1865a) voiced similar ideas. Kozo-Polj ansky (1945), who accepted the placentae as fused carpel margins, nevertheless believed the greater part of the septum to be of receptacular origin. 3. Recalling de Candolle's siliquelle, Suringar (1883), van Tieghem (1906), and Guyot (1962) maintained that united carpel margins form the septum and that placentation is laminal and distant from the margins. 4. The involute carpel margins have a strictly marginal placentation. The septum is a product of the hypertrophied outsides of the carpels; that is, the morphologically lower surface of

CARPEL NUMBER

213

the phyllomes (Celakovsk~, 1902: see fig. 26.2A; Raghavan & Venkatasubban, 1941b: their interpretation is said to rest on an obligatory occurrence of stomates on the carpel outside; Rohweder, 1959-1960; Guides, 1964, 1966b, 1967). This concept resembles the epicarp hypothesis of de Candolle (1821 a). Under the modem nlI theory, the septum can only be a placental outgrowth. Hochstetter (1841), its founder, supposed axial nature of septum and placentae. The 2n theory allows several possible explanations for the septum (see section VI.C): that it is a placental outgrowth (with the receptacle possibly taking part); that it is a product of the median carpel pair (either fertile or sterile, in normal position or addorsed); that it consists of the united margins of the transverse carpel pair; or that it is formed by the upper surfaces of the carpel margins in the median plane according to variant D. In the 3n theory the septum is made of two addorsed carpels (Gerber) or phyllome fractions (Martel). Saunders considers the septum a true dissepimentum consisting of solid carpels. None of these interpretations can be unambiguously verified or refuted by septum anatomy. A comparison of the placental regions in all families concerned makes it probable that the locule-intruding tissue that loosens after fertilization should be equated with a placental outgrowth. This opinion is confirmed by ontogeny (see section VII.E.4). Homologies between the complete Brassicaceae septum and the intruding but unfused placentae of Papaver have been pointed out (see Velenovsk~,, 1910). Likewise, the septa of Capparaceae and Brassicaceae are most probably of common origin (Leins & Metzenauer, 1979). Broad placental regions with complex venation (e.g., in Glaucium of Papaveraceae) do not look like fused carpel margins, for there is no sign of a compound nature and, in the middle of the placenta, there may be a strip of typical inner valve epidermis (characteristic of the inner carpel lamina, not of a placenta). Thus, such a placental region has easily been compared to a (slightly contracted) "normal" carpel. Under the nI theory, however, the epidermal strip could also be ascribed to the carpel margins s.str., the placentae being somewhat submarginally inserted. Therefore, this structure is also not unambiguous proof of additional carpels. To what extent the much-discussed placental bundle systems can contribute to a decision will be elucidated in the following section. D. VASCULAR PATTERN The course of vascular bundles has been considered of enormous importance for detecting floral organization (Bauplan). This was especially upheld by van Tieghem (1868, 1875) and his followers, and for this reason many of the theories that assumed more than two carpels in a bivalvate gynoecium are based almost exclusively on venation. Thus, there is a necessity for a detailed analysis of the vascular pattern in gynoecium and fruit, which is rather similar in the families under consideration. The bivalvate members are described first; the vascular pattern is traced acropetally from the gynoecial base to the stigma.

1. Papaveraceae Above the stamen traces the floral axis stele consists of several more or less confluent bundles forming a ring (Fig. 22.I. 1). In the transverse plane, the dorsal veins of the valves (possibly accompanied by some veinlets) diverge from the bundle ring and enter the valve bases (Fig. 22.I.2). In the median plane the remaining areas of vascular tissue form the placental bundle systems. Their marginal elements turn somewhat inward. In Glaucium this curving is relatively slight and, in each placental region, gives rise to a convex, open arch of mostly dis-

214

THE BOTANICAL REVIEW

Fig. 22. Vertical course of vascular bundles in the base of the gynoer (IV) or fruit (I-Ill). The ascending arabic numerals mark an acropetal sequence of transection levels. In 1.2-I.4 the zones of dehiscence demarcating the valve bases are shown as broken lines. I. Sanguinaria eanadensis (Papaveraceae). II. Dieentra spectabilis (Fumariaceae). IIl, IV. Cleome gynandra (Capparaceae; IV redrawn from Purl, 1945). Scale bars for l-Ill = 1 mm.

CARPEL NUMBER

215

Fig. 23. Vertical course of vascular bundles in the base of the gynoecium (lI) or fruit (I, III). The ascending arabic numerals mark an acropetal sequence of transection levels. In 1.3 the zones of dehiscence demarcating the valve bases are shown as broken lines. I. Alliariapetiolata (Brassicaceae). II. s sativa Mill. (Brassieaceae; redrawn from Puri, 1941). IlL Vancouveria hexandra (Berberidaeeae). Scale bar for I and III= 1 mm.

216

THE BOTANICAL REVIEW

tinct bundles; the lateral ones are partially inverted. In other taxa, the marginal bundles are markedly incurved, thus approaching each other (Fig. 22.I.3). Their xylem faces the xylem of the middle bundles of the complex, which maintain their original position. The result is a circular group of bundles with the (referring to the floral axis) adaxial ones inverted. Occasional localized contact of bundles belonging to different placental regions has been observed in Dicranostigma lactucoides Hook.ill. et Thomson (Gonnermann, 1979, 1980). The inversion of the inner bundles may be constant along the whole length of the ovary (e.g., Dicranostigma [Fig. 19.2]); however, orientation of the bundles may change above the base of the Iocule, with the xylem afterward turned toward the center of the gynoecium (Sanguinaria). A more compact configuration of placental bundle systems develops via an intermediate concentric stage (Chelidonium, Hylomecon, Stylophorum, Eschscholzia, Hunnemannia). In the base of the gynoecium the vascular tissue forms, in the median plane, amphicribral bundles with a central pith. Below the base of the locule these bundles disintegrate into circular systems consisting of a strong, normally oriented bundle in the abaxial position and several weaker, inverted bundles. Also, in Hunnemannia the vascular tissue of the valves passes through a concentric phase, giving rise to four amphicribral bundles in the base of the gynoecium. The concentric systems in the transverse plane then become dorsal valve bundles with accompanying side veins; inversion of bundles does not appear within the valves. Side branches of the placental bundle systems contribute to the vascularization of the valves. Usually they depart from the strong normally oriented bundles; more rarely (e.g., in Stylophorum diphyllum) inverted bundles are also in contact with the valve venation. The valve veins that have originated from the placental bundle systems commonly run in the valve margins parallel to the zones of dehiscence or, in Macleaya, first curve in an obliquely basipetal direction and then turn upward. Ovular traces are further derivatives of the placental bundle systems; they diverge from the inverted parts. The vascular pattern of the few- to one-seeded gynoecia is greatly simplified. Four collateral bundles--valve dorsals and placental bundles--enter the base of the gynoecium in Macleaya and Bocconia. Inverted bundles are not formed. Branches of all four bundles supply the valve venation. Within the valves, the main bundles are mostly parallel. Branching and anastomoses occur with differing frequency in the taxa. A conspicuously reticulate vascular pattern is found in the valves of Sanguinaria (Fig. 24. l) and Stylophorum diphyllum. Apparently recurrent bundles departing from strong veins in the top of the gynoecium have been described in Eschscholzia and Dendromecon (Ernst, 1962a). According to Saunders (1928a, 1937), these bundles are ascending but have lost their basal connection (however, see below, on the origin of pseudodorsal veins in polyvalvate taxa). In the Chelidonieae, the lateral valve bundles terminate in the valve tips, and only the two dorsal veins enter the style. In the Eschscholzieae, whose valves do not detach apically, several valve bundles reach the style. The vascular tissue derived from the valve venation supplies the stigmatic lobes in the transverse plane. The configuration of the placental bundle systems changes at the top of the gynoecium. The inner, inverted bundles disappear, and the normally oriented vascular tissue disintegrates into two (to several) components. If there are no stigmatic lobes in the median plane, the halves of the placental bundle systems pass into both transverse stigmatic lobes. In the simplest case, each stigmatic lobe thus receives three bundles: one-half of the venation of the neighboring placental regions and, in the middle, the valve dorsal vein (the 89+ l + 89pattern described by Dickson [1935] and Saunders [1937]). However, if stigmatic lobes are present in the median plane, they are supplied by the median elements of the disintegrating placental bundle systems (Glaucium, Eschscholzia).

CARPEL NUMBER

217

Fig. 24. Vascular pattern of fruits (I and 2, Fapaveraceae, 3, Fumariaceae; 4 and 5, Capparaceae; 6-8, Berberidaceae, 9, Rutaceae). 1. Sanguinaria canadensis, lower third of a valve, showing reticulate higher-order venation. 2..4rgemone platyceras Link et Otto, macerated dehisced fruit, valve venation formed by horizontal branches (arrow) of marginal valve veins, stemming from the placental bundle systems, that may fuse in the middle of the valve. 3. Corydalis gotlandica Lid~n, venation of valve with replum parts still adhering (from Br(lckner, 1993). 4. Polanisia dodecandra, reticulate valve venation with forking dorsal vein (arrow). 5. Cleome hassleriana, placental region (center) with connected valve veins (arrow). 6. Epimediurn sp,, unilaterally dehisced fruit with pinnate valve venation ending at the zone of dehiscence (arrow pointingto the position of the separation tissue still intact). 7, 8. Leontice leontopetalum, fruit venation viewed from dorsal side and placental region. 9. Dictamnus albus L., basal transection of a carpel; crescent-shaped placental bundle system derived from amphicribral bundle being homologous to the fused ventral veins. Scale bars = 1 mm.

218

THE BOTANICAL REVIEW

The bivalvate Pteridophyllum has reduced valve dorsal veins that end blindly within the valve, as do the other valve bundles. The concentric bundles of the placental regions are the only veins that enter the style. They supply the two stigmatic lobes that occupy a position in the median plane, an exceptional case in the Papaveraceae (see B~ckner, 1985). In the polyvalvate Papavereae, the vascular pattern deviates in several respects from that described above for the Chelidonieae. Here, valve dorsal veins are rudimentary or absent. Exceptions are some species of Meconopsis that display comparatively well developed dorsal bundles that terminate, however, in the valve tips (Ernst, 1962a). During the formation of the placental bundle systems below the locule bases a network of diffuse delicate veinlets may arise that interconnects all placental bundle systems (Papaverrhoeas, Meconopsis horridula Hook.ill. et Thomson; of. Gonnermann, 1980). In Romneya gynoecia the central column is vascularized by fine traces derived from the placental bundle systems. They give off veins to the placentae and join the placental bundle systems below the stigma; vascular tissue remaining in the center ends blindly (Dickson, 1935; Bersillon, 1955; Ernst, 1962a). In Meconopsis horridula and probably Romneya, the placental bundle systems are collateral throughout. A concentric intermediate stage is seen in Meconopsis cambrica, Roemeria, and Stylomecon. The placental bundle systems in Argemone and Papaver remain concentric up to the lower border of the dehiscent region. The valve venation of the polyvalvate taxa originates in the placental bundle systems. Lateral branches of the latter enter the valves below the dehiscent region. In the valve lamina they give rise to higher-order anastomosing veins. Within Argemone valve edges, two strong derivatives of the placental bundle systems parallel the latter. There are many horizontal branches from them that frequently fuse in the middle of the valve (Fig. 24.2). This can be regarded as a precursor to the formation of"pseudo-dorsal veins" characteristic of Roemeria, Stylomecon, Meconopsis cambrica, and Papaver. Here, in the upper part of the gynoecium, two strong horizontal veins emerge from the placental bundle systems; they meet and fuse in the middle of the valve. The basipetal extension of the dehiscent region is demarcated by this fusion, for dehiscence stops at the transverse bundle bridge. The combined vascular tissue forms a thick, recurrent vein that locally splits into two or more strands that may reunite at lower levels; the tip of the main bundle usually extends to the base of the gynoecium. This peculiar structure was observed as early as 1847 by Griffith, who mentioned its recurrent course. On the other hand, Henslow (1891) and Arber (1938) postulated a de novo origin of"true" dorsal veins, whereas Saunders (1928a, 1937) assumed a secondary loss of the basal connection with the stele. Later, Ernst (1962a) precisely described the origin of these bundles and coined the term "pseudo-dorsal vein." Style and stigmatic lobes are supplied by the placental bundle systems only. As already described, they divide at the top of the gynoecium and vascularize the neighboring stigmatic lobes above the valves. If additional stigmatic lobes are present over the placental regions (Argemone, Canbya), the median components of the placental bundle systems terminate in these projections.

2. Fumariaceae In the majority of taxa, the base of the gynoecium is entered by two strong bundles in the median plane and two somewhat weaker ones in the transverse plane. The latter become the valve dorsal veins. They branch immediately above the valve base and, with parallel, higherorder branches, extend the length of the valve (Fig. 24.3). Rarely, the dorsal veins branch below the valve base; in these cases, more than one bundle enters the valves. If, in Dicentra

CARPEL NUMBER

219

spectabilis, the course of the valve bundles is analyzed in serial transections of the gynoecium from base to top, the valve dorsal veins seem to arise de novo below the valve bases, become stronger acropetally, and then branch (Figs. 22.II.2 & 22.II.3). According to Fukuhara and Lid6n (1995), Cysticapnos vesiearia also has only two median bundles in the fruit base, but at a level somewhat higher about five intensively ramifying valve veins appear at once. This phenomenon is discussed below in light of ontogeny. The placental bundle systems are combined from several portions of vascular tissue. In Dicentra, which has the broadest placental regions, an arc or circle of several distinct bundles is formed, with the inner ones being more or less inverted. In the majority of taxa, however, the venation of the placental regions is represented by a compact, concentric bundle. Inner sectors diverge as ovular traces; the remainder of the bundle is then collateral but reorganizes the concentric condition soon. Lateral valve bundles are also supplied by the placental bundle systems; they may anastomose with the branches of the valve dorsal veins. Occasionally, within the valves recurrent are higher-order veins that may originate from the valve venation as well as the placental bundle systems (Pseudofumaria; Brfickner, 1992a). In the valve tips venation reduces to the dorsal bundles. The lateral valve bundles join together and then fuse with the placental bundle systems or, more rarely, with the dorsal veins; blind termination has also been observed. The valve dorsal veins and the placental bundle systems enter the style. The former bend toward the margins of the transverse stigmatic area. The placental bundle systems bifurcate; their halves turn outward and join the valve dorsal veins. The combined vascular tissue may ramify within the whole stigmatic area, terminating in tracheid groups.

3. Capparaceae In the gynophore, there is a ring of more or less confluent vascular tissue. At the base of the ovary the valve dorsal veins depart from the ring and immediately branch laterally or are already accompanied by branches emerging from the ring. In the valves the bundles continue to ramify and form a markedly reticulate pattern in some taxa (e.g., Polanisia dodecandra [Fig. 24.4]). The valve dorsal veins may be inconspicuous and indistinguishable from the rest of the bundles. The remaining tissue of the circular stele, being fractionated by valve dorsal vein gaps, gives rise to the placental bundle systems. As in the previous families, the marginal areas swing inward to the center of the gynoecium base. At that transitional stage localized contact between all placental bundle system components may occur but is not always observed (see Cleome gynandra L. [Figs. 22.Ili & 22.tV.3]). The former marginal parts of the bundle groups are then in a position opposite the middle parts and have an inverted orientation. Thus, in each placental region a circle of distinct bundles (C. hassleriana [Fig. 21.1 ] and others; of. Briquet, 1914) or, more frequently, a completely concentric bundle is found (Puff, 1945; Murty, 1953). According to Stoudt (1941), in Crateva tapia L. two normally oriented bundles run parallel in the placental regions. The ovular traces arise from the flanks of the placental bundle systems. Lateral branches to the valves are also given off(Fig. 24.5) and anastomose with the valve venation. In Oceanopapaver, all higher-order gynoecial bundles consist only of phloem (Schmid et al., 1984). All lateral bundles terminate in the valve tips. If there are well-developed valve dorsal veins, they enter the style (Crateva, Maerua, Oceanopapaver, Capparis ssp.). In Polanisia, the valve dorsal veins ramify in the lower half of the gynoecium, losing their identity (Fig. 24.4). Stoudt (1941) wrote that the valve venation extends to the base of the style, but this is true only of higher-order bundles. The Capparaceae display a similar tendency of reduction of

220

THE BOTANICAL REVIEW

the valve dorsal veins, as do the Papaveraceae, which Karrer (1991) emphasizes. For taxa with several fine veins entering the valve base, her statement "dorsal mediani absent" may be too rigorous, however: the dorsal veins are either inconspicuous or ramify immediately when leaving the stele. The inverted parts of the placental bundle systems terminate in the tip of the ovary. The normally oriented parts fork (in Polanisia in the middle of the ovary) or split into several bundles. Either the resulting veins terminate in the style or the halves diverge into the stigmatic regions above the valves. Each more or less distinct stigmatic lobe thus receives two halves of different placental bundle systems and, if present, the appertaining valve dorsal vein. Frequently all the vascular tissue forms a confluent circle within the style and then vanishes. Two peculiarities in venation deserve special mention. First, in the placental regions of Oceanopapaver there is only one strong obcollateral (i.e., inverted) bundle (Schmid et al., 1984), and other normally oriented bundles do not occur. The derivation of this configuration from the stele has not yet been described. Karrer (1991) does not mention the inversion. The second interesting case has been observed in Crateva adansonii DC. (Puri, 1945, 1950 as C religiosa; see Jacobs, 1964). The placental bundle systems of this taxon consist of two opposing areas, the inner ones inverted. This arrangement is not atypical, but there is no zone of contact between inverted and normally oriented bundles, not even in the base of the gynoecium. According to Puri (1945, 1950), the inverted bundles result from the disintegration of a confluent ring of vascular tissue with the xylem turned outward; this tissue originates de novo in the gynophore. It is surrounded by a similarly confluent ring of normally oriented vascular tissue from which the dorsal "valve" veins diverge below the locules (the indehiscent fruit shows no demarcation of valves). The remaining tissue of the outer ring forms the normally oriented outer parts of the placental bundle systems. An similarly separate inner vascular ring has been observed in the base of some fruits of Dicranostigma lactucoides (Papaveraceae; see Gonnermann, 1979, 1980). At a somewhat higher level, however, it contacts the lateral parts of the forming placental bundle systems. It is interpreted not as a de novo structure but as a recurrent part emerging from the tissue of the placental bundle systems. Indeed, if the inverted Crateva bundles were an independent de novo formation, they might be induced by the extremely active intercalary meristem that produces the gynophore (Eyde, 1975). In the remaining gynophore-bearing taxa, however, such isolated bundles have not yet been observed.

4. Brassicaceae When the staminal traces have diverged, four strong bundles remain in orthogonal position at the very base of the gynoecium. These subdivide, giving rise to a more or less confluent circle of a number of veins. In the transverse plane, two bundles emerge from the circle to become the valve dorsal veins; their gaps may close again. The median parts of the circle constitute a normally oriented bundle that may be flanked by further bundles. Their number is correlated to the width of the placental region (e.g., Matthiola incana, with several bundles arranged in a slightly convex arc; Saunders, 1923). As has already been described, the lateral parts of the two arcs in the median plane pass inward and, while becoming inverted, adopt a position opposite the strong middle bundles. At this stage, varying degrees of fusion between the bending bundles can be observed. In the simplest case (e.g., Cardamine hirsuta L.), below each placental region two bundles curve inward and remain there in an inverted but more or less separate state; the placental bundle systems then consist of three parts. More frequently, the two inverted bundles join, giving rise to a bipartite placental bundle system with an outer

CARPEL NUMBER

221

Fig. 25. Hypotheses concerning the phylogeny of the Brassicaeeae gynoecium. 1. Fames and Wilson's 2n theory, variant A (after Fames & Wilson, 1930). 2. Puff's 2n theory, variant A (after Puri, 1941). 3. Puri's nl theory (after Purl, 1945). For further explanations, see the text.

normal and an inner inverted bundle facing each other. As the bundle configuration forms, contact between the veins belonging to different placental regions may occur (see Earnes & Wilson, 1928; Puri, 1941). The center of the gynoecium may remain free of vascular tissue (e.g., in Eruca sativa [Figs. 23.II.3 & 23.I1.4]; Cheiranthus cheiri), or an X-shaped area of vascular tissue may form across the gynoecial base, which, at a somewhat higher level, disintegrates into four inverted bundles that afterward join in pairs (Alliariapetiolata [Figs. 23.I.3 & 23.I.4]; Brassica oleracea L.; B. rugosa Prain; Capsella bursa-pastoris; Matthiola annua [L.] Sw. and others). Similar vascular reticulation below the locules is observed in Papav=eraceae and Capparaceae (see above). In the very base of the gynoecium thin veins emerge from the placental bundle systems and enter the valves, running along their margins. They may originate from the strong middle bundle itself or from separate bundles flanking the middle bundle (Eruca sativa [Figs. 23.II.3 & 23.II.4; Brassica campestris L.), or they may depart from the afterward incurving and inverting bundles (Brassica oleracea). Detailed descriptions of the development of the bundle configuration in the placental systems and its basal derivatives were made by Puri (1941); most of the patterns mentioned above come from that work. Additional valve bundles arise by ramification of the valve dorsal veins. The placental bundle systems are also continuous with lateral valve veins. Bundles of this origin may form the major part of the valve venation, if the valve dorsal veins are reduced. In Lunaria, the latter lose their identity by forking and anastomosing in the lower half of the gynoecium (cf.

222

THE BOTANICAL REVIEW

Saunders, 1925, 1929a). Also, Cardamine is described as having valve dorsal veins that end slightly above the middle of the valve (Treviranus, 1847). In the valve tip, the lateral valve bundles may join the dorsal ones, in which case only one valve bundle enters the style. Zimmerli (1973) described an apical bifurcation of the valve dorsal veins, whereby one branch terminates in the valve tip and the other runs into the style. Puri (1941) observed additional valve bundles in the stylar base. The inverted bundles vanish apically in the placental regions, and the normally oriented parts enter the style, in several taxa thereby dividing the placental bundle systems in two. All the bundles may fuse to form a ring of vascular tissue in the style. The stigmatic lobes over the placental regions are supplied by components of the placental bundle systems.

5. Berberidaceae A ring of several more or less confluent bundles enters the base of the gynoecium. In the long stipe of the Gymnospermium albertii gynoecium the ring expands from a strong amphicribrai bundle. In the base of the reduced gynoecium of Achlys there is only one concentric bundle. In Ranzania the ring consists of six concentric bundles. Several bundles depart from one side of the ring below the locule. These departing bundles are designated "valve bundles" here, for the sake of comparison with the families already described; however, distinctly demarcated valves are formed in only a few taxa. Terabayashi (1977, 1978, 1979, 1981, 1983a, 1983b, 1983c) described the following patterns of valve bundle formation: 1. Several valve bundles, including a distinct dorsal vein, leave the basal bundle ring. These bundles ramify bi- to tridimensionally within the gynoecium wall. The dorsal vein and some stronger lateral veins enter the style and take part in vascularizing the stigma with numerous dichotomous endings. (Diphylleia, Podophyllum, Dysosma.) 2. Several valve bundles, including a dorsal bundle, are formed. The venation is, however, weakly developed, and commonly only the dorsal vein reaches the stigma. (Mahonia, Berberis [in some taxa there is no distinct dorsal vein], Ranzania.) 3. Several valve bundles, including a dorsal bundle, are formed, all of which ramify intensively. The resulting venation terminates below the zone of dehiscence (Jeffersonia, Plagiorhegma). In A ehlys, the three valve bundles likewise end below the nondehiscent suture. In the indehiscent gynoecium of Nandina the valve venation terminates above the middle. 4. Only the valve dorsal vein leaves the bundle ring. It branches immediately and terminates, like its derivatives, below the zone of dehiscence. (Epimedium, Vancouveria.) 5. Several valve bundles arise; however, a dorsal one is not distinct. The valve venation terminates below the style. ( Caulophyllum, Leontice, Gymnospermium, Bongardia.) These patterns are not absolutely constant. For example, in some fruits of Vancouveria venation fits pattern 3 (Fig. 23.III.4) better than 4. The bundle that is continuous along the dorsal side of Leontiee fruit could be designated a dorsal vein but may become distinct only during the development of fruit. After the valve veins diverge, the bundles that remain in the gynoecial base reorganize into a more or less complete ring. The components closest to the valve bundles course away from them toward the opposite side and, with the ring components there, form a smaller ring system below the placental region. The orientation of its outer (1-)2-3(-4) bundles (Fig. 24.8) remains unchanged, with the phloem directed abaxially and the xylem adaxially, whereas the inner ones are more or less clearly inverted. The most compact placental bundle systems appear as strong periphloematic-concentric bundles. The origin of the ovular traces varies in the following ways:

CARPEL NUMBER

223

1. In taxa with basal placentation (Caulophyllum, Leontice, Gymnospermium, Bongardia), the inner 2-9 bundles of the ring serve directly as ovular traces. They are independent of the outer components of the placental bundle system. 2. In taxa with the placenta extending upward along the ovary wall, the ovular traces are of a different origin. In the lower and middle regions of the placenta the inner bundles of the ring give off the ramifying ovular traces, thereby becoming exhausted. In the upper placenta the outer (peripheral) bundles of the ring provide the branching ovular traces but are not used up. Massive placentae display a well-developed vascular network consisting of the branches of inner (in the lower half) and outer (in the upper hal0 components of the ringlike placental bundle system (Diphylleia,Podophyllum). Lateral components may also supply ovular traces (Ranzania) or run parallel to the outer bundles and terminate blindly above the fertile part of the placenta (Mahonia, Berberis). 3. The origin of the ovular supply is less difficult to trace in taxa with a reduced number of seeds. In Nandina each of the two outer components of the placental bundle system gives off an ovular trace at approximately the middle of the gynoecial height. More rarely, the ovular traces emerge from the inner components that otherwise serve as sterile wall veins or join the outer ones. In the single-seeded Achlys, the ovular trace departs from the only bundle of the placental region. The Berberidaceae differ from the other families in that all bundles of the placental bundle system, not just the inverted ones, are sources of ovular traces. In this, all except the outer, strongest bundles are gradually used up in the acropetal direction. In many taxa these outer bundles also give off several bundles that are horizontal in the gynoecial wall and undergo ramification. These derivatives of placental bundle systems are clearly separated from those of the valve venation, for they do not cross the zone of dehiscence (Epimedium [Fig. 24.6]; Vancouveria;Jeffersonia; Plagiorhegma;Achlys; see Berg, 1972). In Nandina a similar situation is found, though no external demarcation line is visible. In all taxa the outer components of the placental bundle system enter the style and vascularize the extended stigmatic area. They ramify dichotomously and, with valve bundles, if present, form a dense vascular zone. The endings of the numerous bundles may curve outward, which is especially pronounced in Ranzania. Lateral components of the placental bundle systems may also enter the stigma in Nandina. The data on the vascular pattern in the Berberidaceae gynoecium come chiefly from Kaute (1963) and the comprehensive works by Terabayashi (1977, 1978, 1979, 1981, 1983a, 1983b, 1983c). For further details, as well as examples of pattern variation, see Chapman (1936) and De Maggio and Wilson (1986).

6. Discussion The basic characteristics of the gynoecial vascular pattern are largely similar in all five families. In bivalvate taxa, the acropetal course of the bundles can be summarized as follows (Fig. 26.1). From a ring=shaped receptacular stele two distinct bundles at first emerge as the valve dorsal veins (Fig. 26.1B). The remaining vascular tissue rearranges into two complexes that vascularize the placental regions (Figs. 26.1D & 26.1E). Here lateral bundles curve in= ward and become inverted (Fig. 26.1C). The gynoecial venation of the Berberidaceae develops in a similar manner, however, there is only one valve dorsal vein and one placental bundle complex. The fact that four separate bundle groups enter the base of a bivalvate gynoecium has long been known. As early as 1847, Treviranus relatively exactly described the bundle course in Brassicaceae fruits. However, neither he nor other contemporary botanists observed

224

THE BOTANICAL REVIEW

Fig. 26. 1. Schematic depiction of the vascular pattern of bivalvate gynoecia traced in acropetal direction from the receptacle-gynoecium transition (A) to the unilocular ovary base (E). 2. Phylogeny of a brassicaceous placental region according to Celakovsk~'s nl theory (redrawn from Celakovsl~, 1902): A. Touching involute carpel margins, the outside (= morphologically lower side) of the marginal area s.str, indicated by dotted line; outside outgrowths become septum units; B. United carpel margins with fused vascular bundles and homogeneous septum, carpel borderline indicated by broken lines; 3. Carpellary composition of a placental region according to the 2n theory, variant C, by Gonnermann-Brilckner (redrawn from Gonnermann, 1980). Valve carpels, white; solid carpel, hatched; zones of dehiscence, dotted.

CARPEL NUMBER

225

the weak inverted bundles, which decades later were given great importance. Van Tieghem (1868, 1875) first mentioned the incurving marginal parts of the vascular complexes in the median plane and interpreted them as an indication of an additional median carpel pair with involute margins. Van Tieghem founded the so-called anatomical method, based on a detailed analysis of the vascular pattern, which has been widely used to elucidate ground plans. As with each new feature that became available for broad studies, it was hoped that the key to deeper understanding had been found. "A floral anatomist would need only to consult the bundles of a flower to obtain its deepest secrets, whereas the poor general taxonomist must struggle with a host of characters" (Schmid, 1972: 433). Several rules for interpreting the bundle configurations reflect the idea that vascular tissue within the organs is less variable than their external form. The vascular system of the flower was compared to the skeletal system in animals (e.g., Eames [1931: 148] speaks of its "slowto-change skeleton"). Three postulates characterize the concept of vascular conservatism presented by Henslow (1888) and later maintained especially by Eames and his school: l) In fused organs the vascular bundles may still be separate; their degree of fusion indicates their degree of evolution; 2) As vascular patterns change more slowly, an organ reduction or loss may be recognized by the rudimentary bundles that are still present; and 3) The orientation of vascular bundles (direction of xylem and phloem) is useful for the interpretation of homologies. The carpel is ascribed three main veins as basic characteristics (Henslow, 189 l). Its middle bundle, the dorsal vein, emerges from the stele separately and at a level somewhat deeper than the marginal or ventral bundles that supply the ovules. The ventral bundles may fuse in united carpel margins but leave the stele separately and sometimes at a conspicuous distance from each other (cf. Eames & Wilson, 1930). The independence of the four vascular complexes in the bases of bivalvate gynoecia and the abundance of the vascular tissue in the median plane has caused speculation about the presence of carpels in all four regions (2n theory). That the valve tissue belonged to transverse carpels was beyond question. In addition to dorsal veins there are lateral bundles, so the claimed three-trace structure is complete without bundles of the placental regions. To Klein (1894), the unusual massiveness of the latter indicates incorporation of a reduced carpel pair, because such strong bundles did not seem to be common in fused carpel margins. Later, Eames and his followers paid special attention to the weak inverted strands radially opposed to the thick median bundles. Three bundles in the median plane proved, according to the "anatomical method," the existence of a carpel in each placental region. For Brassicaceae, the inversion of the "ventral bundles" of the "fertile carpels" was explained in an intricate manner (Fig. 25.1, after Eames & Wilson, 1930). The phylogenetic starting point is a gynoecium with four fertile, open carpels in two whorls, with the xylem of all the vascular bundles toward the iocule (Fig. 25.1A). Then the inner median carpels become closed, and the ventral bundles are thereby inverted. The margins of the outer lateral carpels fuse with both sides of the dorsal regions of the closed inner ones and lose fertility (Figs. 25.1B & 25.1C). The locules of the inner closed carpels diminish, so they become "solid" (Fig. 25.1D). The ovules can no longer be formed within the dwindling locules; they become extruded to be secondarily inserted on the outside of the solid carpels (Fig. 25.1E). Although Eames and Wilson (1930) emphasized that the crossing of the ovules through the carpel wall was a phylogenetic occurrence and not an ontogenetic one, their postulate has never found acceptance. (Only Kuusk [I 960] described tissue remains on the ovule surface, which, she thought, stemmed from the passage through the wall.) The septum is formed by the expanded veinless margins of the solid carpels that join in the center of the gynoecium. Consequently, it is not a placental outgrowth but consists of carpel wall tissue.

226

THE BOTANICAL REVIEW

Alyavdina (1931) agrees with Eames and Wilson's (1928, 1930) interpretation of the replum venation. Dickson's (1935) discussion is similar; however, she starts from a marginal fusion of open carpels as the first phylogenetic step. An alternating contraction of the fused components leads to solid carpels not in the sense of Saunders's "solid carpel" (see below); the development of concentric and inverted bundles is due to their condensed venation. Only the solid carpels remain fertile, whereas the adjacent expanded ones become sterile. The Brassicaceae septum is a mere placental outgrowth. Dickson's variant of the 2n theory, like that of Eames, is essentially based on the three-trace concept of the carpel. The venation patterns of the "expanded" and "solid" carpels are said to be fundamentally identical. Puri (1941; Fig.25.2) was strongly influenced by the ideas of Eames and his collaborators. However, he realized the necessity to reconcile these ideas with a fact already mentioned by Hannig (1901) but overlooked since: that the marginal bundles of the "valve carpels" originate from the vascular complexes in the median plane; that is, from the venation ascribed to the solid carpels. Furthermore, he observed the X-shaped network formed by all four "ventral bundles" of the solid carpels in the gynoecium base (see Figs. 22.IV.3 & 23.I.3). His interpretation of the venation in this region is that the bundles are still of stelar nature and cannot be designated true carpel bundles. Only at a higher level do they become carpellary, even if they are undoubtedly the same bundles that are distinct in the stele. Puri's explanation of the origin of the septum is simpler than Eames and Wilson's. It was not thought to consist of the carpel margins because the margins, including their (inverted) bundles, would have retracted from the center (Fig. 25.2A). At this stage the margins are still unsealed, so that the ovules can shift outward through the opening (Fig. 25.2B). The septum is, at the gynoecial base, of receptacular origin; its upper part is formed by placental outgrowths fused in the center (Figs. 25.2C-25.2E). In his Capparaceae studies Stoudt (1941) interpreted the inverted bundles of the placental regions in the same way, as ventral bundles of fertile carpels. In the opinion of the present author (Gonnermann, 1979, 1980; Briickner, 1982, 1984), the complex structure of the placental bundle systems served as an essential support for the 2n theory. At that time, she shared the opinion ofDickson (1935) to a great extent but did not accept the existence of sterile carpels. Based on the fact that parietal placentae are usually inserted along fused carpel margins with united venation, she interpreted the lateral parts of the placental bundle systems as integrated components of the venation of the bordering valve carpels (Fig. 26.3). The fact that valve bundles emerge from the margins of the placental bundle systems, inconsistent with the 2n theory until then, subsequently became unimportant. Likewise, Dickson's (1935) difficulties in interpreting the vascular pattern of stigmatic lobes over the valves were overcome. As each of these lobes receives a branch of the apically bifurcating placental bundle systems, Dickson (like her predecessor Saunders) postulated, split stigmatic lobes of the "solid" carpels that fuse with the lobes of the "expanded" carpels (89+ l + 89combination). For the present author, the apical branches of the placental bundle systems represent vascular tissue belonging to the valve carpels, whereas the venation of the reduced solid carpels disappears at lower levels. The 2n concepts of van Tieghem (1868, 1875), Henslow (1891 ), Klein (1894), Eames and Wilson (1928, 1930), Dickson (1935), Puri (1941), and the present author (Gonnermann, 1979, 1980) are relatively similar in their interpretation of the gynoecial vascular pattern. Saunders's (1923 and later) theoretical four-carpellary construction of the siliqua is based on completely different ideas. Although she considered the course of the vascular bundles essential and stated (1929a: 13 l): "The character of the 'vascular anatomy of the base of the ovary' lies at the foundation of the theory of carpel polymorphism and has been used.., as the ultimate test of carpel number," she rejected the claim that three traces must supply a carpel by

CARPEL NUMBER

227

reason of homology with the venation of lateral appendages of the axis. Her three carpel types vary considerably regarding the vascular pattern: 1) Valve carpels have a conspicuous midvein and pinnate, rarely palmate, venation; 2) Solid carpels are reduced to a single (possibly complex) bundle running from the base to the tip of the gynoecium; and 3) Semisolid carpels have a double midvein, from which the ovular bundles diverge, with lateral branches extending to the ovary wall. These carpel types, the existence of which Saunders vehemently proclaimed, have been rejected by nearly all morphologists, especially the solid and semisolid ones. Saunders's statement (1926: 299): "External structures which are restricted to the course of vascular cords running a separate course from the base of the gynoecium upward (= cords of solid carpels) enable us to judge the carpel number from outward inspection," is an unproved postulate. Moreover, this idea is not new. As early as 1848, Hochstetter (1848: 158) emphasized: Dass ich als die Grundlage eines Blatts bei den Gefiisspflanzen einen Gef'~issb~ndel betrachte, der einen Hauptnerven zu bilden imstande ist. Ich nehme so viele Bl~itter an, als Hauptnerven da sind, welche in dem Wirtel herrschen, weswegen i c h . . , sagte, in den beiden Klappen der Cruciferen-Frucht, worin ein deutlicher Nerv bis zur Spitze durchlaufe, mOchte schon gewissermassen ein zweites Paar von Bl~ittern angelegt sein, die abet nicht zur Vollendung gekommen seien--im Fruchtblattkreis kOnnen als vollendete Hauptnerven nur diejenigen betrachtet werden, welche an ihrer Spitze zur Narbe sich entwickeln. Thus, he equated a vertical vascular bundle continuous from gynoecial base to stigma to a phyllome and an "incomplete" vascular bundle that does not reach the stigma to an imperfect phyllome with retarded growth. Henslow (1891:167, 168) expresses a similar opinion regarding the valve dorsal veins and, possibly, further valve bundles of Eschscholzia: "This appears to show that the cords.., are potentially marginal or placentary;.., it looks as if other cords might be marginal as well, but are now arrested so as simply to supply the wall of the ovary." ("Marginal cords" are understood as the two closely approached lateral veins of an extremely reduced, needlelike carpel.) According to Saunders's ideas, the vertical venation of a solid carpel is allowed to consist of two or more bundles that may vary in arrangement and degree of fusion. This "range of variation" caused Eames and Wilson (1930: 639) to criticize: "It appears.., that the solid carpel may have any anatomical structure, or any form whatso= ever, and any function which suits the occasion. And when.., valve and solid carpels cannot be made to explain the structural condition, because of some peculiarity of secondary venation, the convenient and flexible semisolid carpel can always he adjusted to the situation." Saunders believed that carpel polymorphism occurred in all angiosperms. She distinguished several types of valvate fruits, three of them solely for Brassicaceae (cf. section VI.E). Siliculae differ mainly from siliquae in that the placental bundle systems give offconspicuous horizontal branches to the valves, the valve dorsal veins being poorly ramified or reduced. As the definition of solid carpels does not allow the vein to produce such branches, semisolid carpels with median placentation are postulated, the lateral borders of which are demarcated by the tips of the side branches emerging from the median venation. In practice, however, there is no difference in venation between siliqua and silicula. The problem lies in the interpretation of the secondary branching pattern. Saunders declared (1929a: 135): "Only as a part of post-fertilization development does further secondary venation make its appearance. Nevertheless one may presume that this later development reflects the character of the carpels"; that is, that higher-order bundles help in recognizing carpel borders. However, here she ignores the general occurrence of lateral branches from the placental bundle systems that

228

THE BOTANICAL REVIEW

are also seen in fruits to which she ascribes solid carpels. On the contrary, Eames and Wilson (1928: 265) stated: "In fruits generally, the vascular system is that of the ovary; the number and course of the main bundles is not changed, but branches are added, and these may become numerous and strong. These branches may show little respect for carpel limits." That vascular bundles may indeed connect postgenitally fused organs of separate origin was observed by Briechle-Mack (1993: in postanthetic ovaries of pseudosyncarpous Annonaceae) and U. Hofmann (pers. comm.: in the roof of the pitfall-trap flower ofCeropegia, Asclepiadaceae). The importance of the inverted bundles in the placental bundle systems has also been debated. Saunders did not pay particular attention to them, for in a cylindrical solid carpel all bundles present are expected to turn the xylem toward its center. Bifurcation of the placental bundle systems in the gynoecium apex and divergence of the branches into the stigmatic lobes of the "valve carpels" cannot be passed over so easily. Saunders can explain this phenomenon only through the 89+ 1 + 89pattern (bifurcation and fusion of stigmatic lobes), which certainly is an unsatisfying construction. As for the Papaveraceae-Eschscholzioideae, the sum of five "solid" and four "valve carpels" per valve, based on the number of vein-bearing longitudinal ribs, has also been generally refuted. Saunders's interpretation of the gynoecial venation was influenced by many subjective ideas that cannot stand up to critical reconsideration. | The course of the vascular bundles has also been used in other concepts (of. section IV) to iOterpret carpel number and position. Because of their poor acceptance these ideas are treated briefly here. In the placental regions of a teratological siliqua of Sisymbrium (Brassicaceae), Gerber (1899a) found an increased number of inverted bundles opposite an outer, normally oriented bundle group. The arcuate arrangement of the inverted groups suggested to him the presence of vascular tissue from two additional carpels. Apparently turning their backsides toward the center of the fruit, they would be united back to back, thus forming the septum. This recalls the interpretation of the septum as two addorsed carpels by Fournier (1864); however, consistent with the tetracarpellary theory, he considered the placental regions the marginal parts of these carpels. Fournier's concept is based mainly on a misinterpretation of the ramified transmitting tissue in the center of the septum as dorsal vascular tissue of a carpel pair. As ovular traces are given off exclusively by the inverted bundles, the addorsed carpels were the only fertile components of the gynoecium. Eames and Wilson (1928) criticize this concept because of the lack of vascular tissue representing the dorsal carpel veins in the center of the septum. However, Gerber apparently considered the middle part of the group of inverted vascular bundles to be the dorsal vein, whereas the septum tissue is parenchymatous throughout. To explain such an unusual carpel position, Gerber first postulated a radial d~doublement of two carpels situated in the median plane (2n theory). Later (1899b, 1899c, 1899d; 1900a, 1900b), based on the zones of dehiscence that were then interpreted as carpel borders, he proposed the existence of four carpels in the gynoecium wail. He did not abandon the idea of two addorsed carpels forming the septum; furthermore, he no longer applied it to teratological cases only, as he had found inverted vascular tissue, though in smaller amounts, in normally developed crucifer fruits. Because the inverted bundles terminate below the stylar region, style and stigmatic lobes were thought to have been produced by the four sterile carpels only. In Gerber's concept, six gynoecium components was consistent with the six-membered androecium of Brassicaceae. However, there is no proof of an inception of additional carpels rotated by 180 ~ Gerber had overestimated the inverted bundles, and with continued study he became convinced of this fact. Consequently maintaining the idea of vascular conservatism, he had not been able to explain the variability of the vascular pattern in the placental regions of the Brassicaceae.

CARPEL NUMBER

229

Martel analyzed the venation of the whole flower, and to him the connection of the bundles in the stele is of great importance. Because the bundles may join in different patterns (see Norris, 1941; Sands, 1973), Martel first considered the flower a system of pluriaxial structures (1895). Later (1902) he claimed "phyllome fractions" that, strictly speaking, are represented by individual vascular bundles. This manner of analysis, decades later also applied by Motte (1957), rests on the inadmissible assumption that the floral stele represents only the sum of the already determined organ traces. To the contrary, Eames and Wilson (1930: 649) state: "The interpretation of the anatomy of any appendage can.., only be made on a basis of study of the actual departure of the trace from the stele." Sands (1973), a student of Melville, investigated the bundle course in the receptacle of some papaveraceous and fumariaceous taxa and concluded, much as Melville (1962, 1963) had, that the vascular pattern does not agree with the classical carpel theory but speaks in favor of the gonophyll theory. Sands, like several predecessors, emphasized the separate origin of valve dorsal veins and placental bundle systems. However, he uses this fact to deduce not additional carpels but "tegophylls" and "fertile axes" (cf. section V). This kind of view is applied not only to vaivate taxa but to angiosperm gynoecia in general. In his early papers, Lignier (1896a, 1896b, 1896c, 1896d) stressed the basic assumption that all Brassicaceae and Fumariaceae floral appendices have three longitudinal veins (fusions possible) and tend to be triiobate. Perfect dimery of alternating whorls provided, the three-veined carpels must occupy a position in the median plane and thus bear the placentae along the midvein. Later, Lignier recognized that he had made unsound generalizations based on highly specialized floral features, especially of Fumariaceae, and false conclusions regarding homologies; hence he dissociated himself from his earlier concepts. Spratt (1932) presented her concept of the nII theory in a short and precise, though not incontestable, manner. In young gynoecia of silicula type, she pointed to the lack of valve dorsal veins, whereas the strong, normally oriented bundles of the placental bundle systems always extend to the stigma. These bundles show no sign of a double nature, as would be expected in fused carpel margins, but rather can be compared to the midrib of a leaf, as they protrude outward. The inverted bundles are considered basal side branches of the carpel "midrib," their torsion due to their function as placental veins. The valve midveins are said to often be double, which indicates touching carpel margins. Had Spratt extended her comparative studies to bivalvate Papaveraceae taxa with well-developed, vascularized stigmatic lobes over the valves, she would have faced serious difficulties in applying her nII concept. The conclusion that a carpel must have a midrib as prominent as that in a green, pinnately veined leaf is untenable. The postulate of placentation along the middle line of the carpel is nearly as extraordinary as the insertion of an addorsed carpel whorl by Gerber. Like Spratt, Tr6cul (1873) tried, but was unable, to recognize a repetition of the pinnate vascular pattern of the leaf within the carpel. Side branches of the placental bundle systems running across the zones of dehiscence into the valves but remaining separate from their dorsal veins are not typical of pinnately veined leaves, and a further contrast to leaf venation is the absence of valve dorsal veins. Instead, reticulations in the pericarp appear to be a continuation of the receptacular bundle network. Thus, Tr6cul did not support any of the n and 2n theories and considered the Papaveraceae gynoecium to be axial in nature. As Sprotte (1940) demonstrated later, however, the puzzling modes of origin of the secondary venation are absolutely typical for carpels. Statements on pseudomonomery in Berberidaceae gynoecia are essentially based on studies of the bundle course by Chapman (1936), a student of Eames, and Kaute (1963), a student of Eckardt, whose work she continued (cf. Eckardt, 1937). Both stressed the vascular patterns

230

THE BOTANICAL REVIEW

in teratological cases (see section VII.F). (However, Terabayashi, who published the most detailed analyses ofberberidaceous floral venation [ 1977 and later], avoided any speculation on carpel numbers.) Chapman, especially, attempted to identify, according to the basing principle of vascular conservatism, the individual veins of the placental bundle systems as members of the "promoted" carpel or the supposed two or three "diminished" carpels, respectively. Regarding the interpretation of the inverted bundles, she is rather inconsistent. On one hand, they are thought to represent, because of their direction, the joined ventral bundles of the reduced carpels. On the other hand, when present in higher numbers and in a weaker state, they are explained as a result of the considerable thickness of tissue in the placental region. Kaute did not advocate a definite number of incorporated carpels. In her scenario there is also a problem of determining when stelar bundles become true pistil bundles. Is Kaute's conclusion that the ovular traces originate from the basal venation of the valve correct, or are there axial bundles that become placental bundles after the valve venation emerges from the stele, as was the opinion of Leinfellner (1956)7 Terabayashi has demonstrated that all members of the placental bundle systems are potentially able to produce ovular traces. Thus, the explanation of one or two bundles as dorsal veins of reduced carpels seems to be unfounded. The complexity of the berberidaceous placental bundle systems challenges the 2n theory to include intricate hypotheses concerning fusions and reductions. How are these various interpretations of the vascular pattern addressed by the nl theory? Many of its supporters are skeptical of or refute the "anatomical method," believing that position and quantity of vascular tissue depends essentially on functional aspects. Thus, Braun (1874: 46) pointed to the fact that fused carpel margins, due to their special placentiferous function, undergo an early and intensive development. This causes the production of the strongest vascular bundles, the branches of which curve downward toward the middle line of the carpel ("class hier die st~irksten Geflissbtindel gebildet werden, deren Zweige riickl~iufig nach der Mitteilinie des Fruchtblattes sich erstrecken"). Comparable complex placental bundle systems exist in many angiospermous taxa. Puri (1945) diverged from the 2n theory when he found inverted bundles originating from the stele in an identical manner in taxa that hitherto had not been ascribed any reduced carpels. Thereafter he explained the inverted bundles no longer as marginal strands of solid carpels but simply as placental veins developing from the last stelar tissue in the receptacle. Nevertheless, he considered the inversion something peculiar, because, in his opinion, the "typical" parietal placentation is caused by open carpels fused margin to margin and should thus display normal orientation of all vascular bundles. He therefore interpreted the inverted bundle position as a relic of an ancestral axial placentation in a bicarpellate gynoecium with the marginal bundles turning their xylem away from the center. His phylogenetic diagram of the Brassicaceae gynoecium (1945) shows his changed opinion (Fig. 25.3), and he applies it to Capparaceae, Papaveraceae, and Fumariaceae. His acceptance of Chapman's concept of pseudomonomery in Berberidaceae (1951, 1952), however, makes one question whether he studied berberidaceous gynoecia himself. Celakovsk~ (1894) was convinced that the "anatomical method" is not, on the whole, reliable, especially when used as the only criterion. Nevertheless, he admitted, concerning the Brassicaceae (1902: 89): "Trotzdem muB auch die rationelle Morphologie der anatomischen Structur, insbesondere dem Geftissbtindelverlauf Rechnung tragen, freilich in Uebereinstimmung mit anderweitig sicher festgestellten Thatsachen, wie hier mit der Zweibl~ittrigkeit des Fruchtknotens." A phylogenetic origin is evoked to explain the inversion of the inner placental bundles. Celakovsky proposed a fusion of involute carpel margins, with the phloem oriented toward the morphologically lower side of the carpel becoming inverted by inrolling of the carpel margin. The ovules are thereby strictly margin borne, and the septum is produced

CARPEL NUMBER

231

by the united outsides of the carpels (Fig. 26.2). Ou6des (e.g., 1964, 1965, 1966b) confirms Celakovsk~,'s hypothesis concerning the organization ofBrassicaceae gynoecia and the mode of bundle inversion. Berberidaceae gynoecia have been interpreted as monocarpellate by the same means; the numerous members of the placental bundle system are said to belong to the congenitally fused carpel margins and are thus more or less inverted (van Tieghem, 1875; Ou6d6s, 1977). Ou6d6s emphasized that the pattern of the lower ovules being supplied by inner bundles and the upper ovules by outermost bundles of the placental bundle system is also seen in other families that certainly do not have solid carpels (e.g., Rutaceae; Gu6d6s, 1973). Some authors (e.g., Karrer, 199 l) do not report inverted placental bundles, possibly due to delayed differentiation. According to Eames and Wilson (1930), Raghavan (1939), and Raghavan and Venkatasubban (1941b) they appear only when the ovules are fairly well developed, and in most Brassicaceae the ovular traces join the placental bundle systems only after shedding the petals. (However, in Papaver the ovular traces become differentiated into xylem and phloem well before fertilization; they develop from the placental bundle system toward the nucellus. See Cass and Fabi [ 1990].) The inverted members are thus secondary venation, and it is possible that during their differentiation from procambial strands xylem induction is influenced by the better-developed xylem of the strong, normally oriented bundies. Fisher (1971) has shown that during procambial splitting or origin of additional bundles due to an active intercalary meristem, when the newly arising bundles are close enough to a differentiated bundle, their xylem pole turns toward the xylem of the neighbor bundle. Another problem that needs detailed examination is the direction of the basal connection of the inverted bundles. Traditionally, acropetal series of gynoecium transections are analyzed, and therefore most authors describe the course of the vascular bundles from the gynoecium base to the style and stigma, to trace the transition of the receptacular stele to the gynoecial venation. Acropetal description of the bundle course is also used in this article to accentuate the foibles of rigidly following one direction. The seemingly acropetal run of a bundle in a fully differentiated gynoecium or a nearly ripe fruit does not necessarily correspond with the direction of its differentiation in ontogeny. During apical and marginal growth of the carpel the primary venation develops acropetally, indeed, but the procambial strands giving rise to higher order bundles need not repeat this direction. As Fisher (1971) demonstrated in monocotyledon leaves, these bundles may originate in the zone of basal insertion, extend into the intercalary meristem, and continue through it; they may also, however, arise de novo in the upper region of the intercalary meristem and develop in (apparently) basipetal direction. Bundles of the latter origin may terminate blindly in the leaf base when the activity of the intercalary meristem ceases. In this case, a description of an acropetal course would not give an exact picture of the true direction of development. Abruptly ending bundles would then be ascribed a de novo origin in the organ base followed by upward differentiation. Because carpels (as well as syncarpous gynoecia)---comparable to the leaves of monocotyledons--are characterized by pronounced basiplastic growth, the carpel base or gynoecium base, respectively, retains its meristematic nature longest and differentiates latest. Therefore, the basipetal connection of the vascular bundles to the stele and not their acropetal emergence from the stele should be studied. Applying Fisher's findings to the arrangement of the inverted bundles in the placental bundle systems, another interpretation is conceivable; namely, that they originate in the apex of the primordial ovary, develop basipetaily, and in several taxa fuse at the base with the normally oriented strands, thus producing a concentric bundle or, rather frequently, join the stele separately; in rare cases (Crateva of the Capparaceae: Puri, 19,45, 1950) abrupt ending is also possible. This kind of description does not imply any aspects of phylogeny. Even primary bundles like the valve dorsal veins of Dicentra

232

THE BOTANICAL REVIEW

spectabilis (Fumariaceae) may fail to connect to the stele (Fig. 22.II.2) and instead end in wide tracheids. The explanation for this phenomenon is clear from gynoecium ontogeny (see section VII.E.2). When style and stigma are already well developed and vascularized, the delayed differentiation of the ovary is still in progress. In young ovaries, the bundles already differentiated in the style are not perfectly connected basally. The placental bundle systems attain it in any case; the valve dorsal veins may fail, as the example demonstrates. Whether this is a general phenomenon in Dicentra spectabilis has not been discovered because of the small number of fruits investigated so far. The degree of variability in vascular patterns is insufficiently known but seems to be higher than formerly expected (e.g., Cleomegynandra of the Capparaceae [Figs. 22.III & 22.IV]; see also Trzaski et al., 1997; Weiss and Trzaski, 1997). Eggers (1935) and Arber (1938) offered a purely functional interpretation of inverted placental bundle members. The former assumed that the inversion is due to the morphological nature of the placenta and to the late origin of these bundles. The latter considered it the result of a special type of bundle branching, required because the ovules to be supplied are positioned on the same radius as the bundle that supplies the ovular traces. Inner sectors split from a concentric bundle are naturally inverted. However, to Arber these concentric bundles do not represent the contracted venation of solid carpels but, being single strands, are a mere variant of the collateral bundle type (see Arber, 1930). This opinion is confirmed in the taxa discussed here by the occurrence of concentric bundles in regions other than the placental regions. In Hunnemannia (Papaveraceae) two such strands disintegrate into the valve venation. In the pistil base of Ranzania (Berberidaceae) all the bundles are concentric; these bundles should be interpreted in the way that concentric placental bundle systems are. Whether the strong inverted single bundle supplying the placental regions of Oceanopapaver (Capparaceae) should also be considered a variant of the collateral type has not been resolved yet; however, Schmid et al. (1984) assume a purely functional origin for the inversion. Because the inverted bundles are not proof of solid carpels, the question remains whether extended, multimembered placental bundle systems of some Papaveraceae taxa (e.g., Glaucium, Dicranostigma, and Sanguinaria) may, despite their size and subdivision, be considered products of merged carpel margins, There are clues to corroborate that interpretation. Comparable placentae are formed in the tricarpellate gynoecium of Hesperomecon (Papaveraceae-Platystemonoideae), which displays basipetal sutural dehiscence. Except for Saunders (from 1930), no 2n theoretician expected carpel dimorphism in that gynoecium. The gynoecial venation of this rare monotypic genus has been described by Lignier (1911), Saunders (1930), and Ernst (1967). In the unusually broad placental regions there are many vascular bundles of different thicknesses. At the ovary base they display a ringlike arrangement with four to five stronger, normally oriented members in the outer position and weaker veins inside (Lignier does not mention the orientation of the inner members, but it is expected that they are inverted). As Lignier points out, these placentae appear to be independent carpels. At a somewhat higher level, the vascular tissue extends, with the bundles side to side in a broad row; all give off ovular traces. At about the middle of the ovary the placental bundle systems divide into halves, their members still supplying traces to the ovules. In the gynoecial apex two strands from each placental bundle system remain, and they enter adjacent stigmatic lobes, which also contain the dorsal carpel veins. In this case the IA+ 1 + 89combination of the stigma bundles consists of dorsal and marginal bundles from a single carpel. The line of dehiscence is formed between the bisected portion of the placental bundle systems; the vascular tissue may tear downward. Saunders (1930) and Ernst (1967) figure that the placental bundle systems give offnumerous ramifying branches, ending in the ovular traces. In transverse sec-

CARPEL NUMBER

233

tion, the bundle configuration depicted by Lignier would be seen--a configuration as intricate as that in the placental regions ofUlaucium, Dicranostigma, or Sanguinaria. Because the edges of the placental regions in Hesperomecon are not demarcated by sunken zones of&hiscence, and dehiscence is sutural, the placental regions have not been interpreted by morphologists as separate solid carpels. Furthermore, even free carpels may have complex venation in their fused margins. This is clearly seen in Dictamnus L. (Rutaceae), where the concentric ventral bundle in the carpel base acropetally disintegrates into a crescent-shaped, muitimembered bundle ring (Fig. 24.9) that somewhat higher splits into the two groups of vascular tissue that supply the carpel margins. The inner members of the groups are inverted and supply the ovular traces (Briickner, 1991b). The complexity of this placental bundle system is certainly not due to solid carpels. In summary, considering all the theories postulating more than n carpels, the one published by the present author (Gonncrmann, 1980, 1982; Briickner, 1982, 1984) is most consistent with the vascular pattern. However, the nearly ripe fruits used in her investigations did not reveal the direction of bundle differentiation in its functional context. By traditional acropetal analysis of the bundle course, some patterns have obviously been overestimated in a phylogenetic sense. Furthermore, several postulates of fusion and reduction processes are necessary to "incorporate" solid carpels especially in the narrow placental regions with simple venation that are present in some taxa. Although the existence of more than n carpels cannot be refuted through the vascular pattern only, the nI theory is more direct in its explanations. In the end, differing opinions as to 2n or 3n versus n carpels reflect the difference in opinion about the concept of vascular conservatism. Supporters of vascular conservatism may revive variations of the 2n theory, though probably not as extreme as Saunders's, and not necessarily restricted to the taxa discussed here. The most extreme opinion held by opponents of vascular conservatism is that vascular bundles are simply induced where they are needed. Therefore, veins cannot be used to demarcate "congenitally fused" organs. Between these two extremes the majority of morphologists believe that vascular patterns must not generally be considered conservative. For a more detailed criticism of the concept of vascular conservatism, see Arber (1933), Goebel (1933), Carlquist (1969), and Schmid (1972). E. ONTOGENY OF THE GYNOECIUM The first comparative analyses of gynoecium ontogeny in the families concerned were published by Payer (1857). His excellent three-dimensional drawings were later complemented by study of serial sections. With this method the section plane must be precisely adjusted to avoid misinterpretations, and only since use of the scanning electron microscope in combination with serial sections became widespread have more precise presentations of developmental processes been possible.

1. Papaveraceae The gynoecium develops from the discoid floral apex that transforms into a plug stage (Pigs. 27.1 & 28. l) and gives rise to a gynoecial girdling primordium. In the bivalvate taxa, the ring becomes elliptical, with its largest diameter in the transverse plane. According to Payer (1857), it arises from two separate primordia that immediately fuse by interprimordial growth. (The same process is said to join the higher-number primordia of the polyvalvate taxa.) This observation was not confirmed by Schumann (1890), Bersillon (1955), Sattler (1973), or Karrer (199 l), who describe the initiation of the continuous girdling primordium as

234

THE BOTANICAL REVIEW

Fig. 27. Early gynoecium ontogeny in Papavei'aceae-Chelidomeae 0-4, Macleaya microcarpa; 5-8, Chelidonium majus). 1. Gynoecial plug. 2. Two transverse growth centers arise on the gynoecial plug. 3. Fusion of the growth centers gives rise to the gynoecial ridge; slight basal bulge in the median plane. 4. Development of the gynoecial tube; the narrow median parts (future placental regions) becoming overtopped by the transverse ones (future valve backs). 5. First signs of the gynoecial ridge and basal septum. 6, 7. Older stages with distinctly septate base, the median parts still appearing relatively wide. 8. Transverse parts of the gynoecial tube overtopping the narrow median ones and forming stigmatic lobes above the future valves (bud rotated by 90~

CARPEL NUMBER

235

Fig. 28. Early gynoecmm ontogeny (1-4, Eschscholzia californica [Papaveraceae]; 5-8, Dactylicapnos torulosa [Fumariaceae]). 1. Gynoecial plug. 2, 3. Gynoeeial ridge, obliquely lateral view (low septum distinct) and top view; median parts (future placental regions) wider than transverse ones (future valves). 4. Older stage showing four growth centers, the median ones still more bulky than the transverse ones but becoming overtopped by the latter; gynoecial tube still widely open. 5. Slightly concave floral apex (bud rotated by 90~ median plane in horizontal position). 6. Four-edged gynoecial ridge. 7. Slightly older stage with signs of basal septation. 8. Gynoecial tube with constricted apical opening and overtopping transverse parts.

236

THE BOTANICAL REVIEW

the starting point in gynoecial development. However, in early gynoecium ontogeny of Macleaya two separate transverse primordia are seen (Fig. 27.2; cf. Ronse Decraene & Smets, 1990: figs. 41 & 42). These discrepancies show that the degree of variability in primordial initiation needs further examination. Undoubtedly, however, the annular stage predominates early gynoecium ontogeny in any case. In several taxa a very low ridge is formed across the ring base (e.g., Chelidonium [Fig. 27.7]; Eschseholzia [Fig. 28.2]), termed "synascidiate" by Karrer (1991). Werle-Sprengel (1993b), however, denies the occurrence of a "true" septum resulting from meristematic activity of the carpel margins. She interprets the ridge as continuation of the placental bulges-that is, a placenta-induced structure--and proposes that no synascidiate zone exists in the Glaucium, Dicranostigma, Eschscholzia, or Dendromecon taxa studied. A "congenital" septation of the ovary extending to the stylar region is typical for Romneya (Bersillon, 1955; Karrer, 1991), so Karrer calls the whole gynoecium synascidiate. Totally symplicate gynoecia characterize many other taxa. Growth centers soon develop in the future placental regions at the gynoecium rim. Temporarily they may considerably overtop the future valve backs (especially pronounced in Papavereae [Karrer, 1991] and Hypecoum [Ronse Decraene & Smets, 1992: figs. 40 & 41]). Schumann (1890) describes a successive initiation of the two placental bulges in Glaueium and Chelidonium, which has not been confirmed by later observations. During the course of development, growth becomes accelerated in the regions of the valve backs, resulting in their permanently overtopping the placental regions. The tips of the protruding areas produce the carinal stigmatic lobes that either remain choricarpous or fuse postgenitally. According to Sattler (1973), in Chelidonium the ovules are initiated in an acropetal direction. Payer (1857) observed basipetal initiation in Maeleaya eordata, whereas in Chelidonium, Glaucium, and Eschscholzia the first ovules were found halfway along the vertical extent of the placenta, and subsequent initiation proceeded both acropetally and basipetally. The latter mode has also been confirmed for Papaver rhoeas (Cass & Fabi, 1990). In Hunnemannia, the gynoecial tube is still wide open at the time of ovule inception; developing ovule primordia can be observed for some time (Karrer, 1991). In the remaining taxa, the apical opening has already constricted when the ovules are initiated within the young ovary. Finally, the stigma attains its definite shape. If there are stigmatic lobes over the placental regions, their growth centers may be initiated in an early stage of gynoecium ontogeny and thus give rise to well-developed organs (Eschseholzia [Fig. 12.7]; of. Karrer, 1991). Late differentiation of stigmatic areas superposing the placental regions is also found, which lack conspicuous growth centers at the upper margin of the gynoecial tube (Glaucium,Argemone). Narrow stigmatic bulges over the placental regions start to develop shortly before the final differentiation of the dominant carinal lobes (see Sattler, 1973). Ira style is formed, it elongates at about the time of stigma differentiation or, possibly, somewhat earlier. The zones of dehiscence become discernible rather late in ovary ontogeny (Zimmerli, 1973). Finally, short gynophores develop in some taxa (Bocconia, Macleaya, Papaver). Of the Platystemonoideae only Platystemon has been analyzed, the ovary of which is initiated as girdling primordium (Payer, 1857; Bersillon, 1955; II'ina, 1968; Karrer, 1991). Gynoecium ontogeny is fundamentally identical to that of the rest of the family. The gynoecial ridge is shallowly stellate, as the backs of the numerous carpels protrude. Postgenital sealing of the carpel margins starts relatively late, and only part of the ovules are thus enclosed in the individual locules, whereas others protrude through the ventral slit into the central cavity of the ovary. Early accelerated growth of the dorsal regions of the carpels leads to long, free tips

CARPEL NUMBER

237

that act as stigmatic lobes. The intercarpellary zones of fusion, though massive structures in the ovary primordium, differentiate only a little and at anthesis are parenchymatous and inconspicuous, thus causing the choricarpoid appearance of the gynoecium.

2. Fumariaceae In contrast to Papaveraceae, the floral apex is not plug-shaped but plane to shallowly concave (Figs. 28.5 & 29.6) and is enclosed by the young androecium. A continuous gynoecial ridge is initiated that is first elliptical but soon turns tetragonal because of space constraints (Fig. 28.6; Payer, 1857; Eichler, 1865; Buchenau, 1866; Ronse Decraene & Smets, 1992; Brtickner, 1992a, 1992b). Zimmerli (1973) observed accelerated growth in the transverse and median parts of the ring primordium simultaneously giving rise to four bulges. These growth centers are, however, not separate but ring borne (Fig. 29.3). The median bulges remain distinct only in the Fumariaeae (Pseudofumaria [Fig. 29.7], Fumaria, probably also Fumariola, and Rupicapnosp. p.) and may clearly overtop the transverse rim parts in early stages ofontogeny (Ronse Decraene & Smets, 1994: fig. ld). Schumann (1890) described a short gynoecial tube with two tips each in the transverse and median planes also for Dicentra spectabilis (Corydaleae). Judging from the shape of the differentiated stigma, however, median bulges are unlikely to occur except in teratological cases. In Dicentraformosa with a similar stigma construction, it is clearly only the transverse parts of the ring wall with accelerated growth (Payer, 1857; Ronse Decraene & Smets, 1992). The base of the young gynoecium can only be viewed in the earliest stages of ring formation. In the taxa observed, a distinctly synascidiate zone is hardly recognizable (Fig. 28.7). Whether the ovaries of all taxa are symplicate is still insufficiently known. In Dicentra and Corydalis a low septum at the gynoecial base was reported (Arber, 1931 c), as was a comparable structure at the fruit base ofPseudofumaria (Brfickner, 1992a). To date, neither the constancy of this feature nor the time of its development is known; perhaps it is a secondary structure of late origin. The top of the gynoecial tube is tightly enclosed by the already well developed androecium and becomes compressed transversely, thus transforming the apical opening into a narrow transverse slit, which makes it impossible to look into the differentiating ovary (Figs. 28.8 & 29.4). Development of the protruding transverse regions is much more intensive than that of the median parts of the gynoecial tube (Figs. 29. l & 29.4). The former become the primary stigmatic appendages. Below these bulges, secondary growth centers appear, a process that Goebel (1933: 1925; my translation) described as "ramification of the carpel tips" (Figs. 29.2 & 29.5). They enlarge the future stigmatic area. Simultaneously, the primary bulges may approach each other so tightly that the apical opening between them virtually disappears (Fig. 29.2). The secondarily accelerated growth in the transverse plane gives rise to the vertical, variously shaped stigmatic plates or branched stigmas, respectively, that bear along their margins additional receptive bulges in varying numbers (see section VII.A.2). Gynoecial tubes exhibiting median growth centers likewise show regions of secondary growth below the primary stigma appendages. They give rise to the bulky stigma base that encircles the style (Fig. 29.9, see also Fig. 13.6) and lift the well-developed transverse stigmatic appendages that remain distant from each other. Between these spreading transverse lobes are two median protuberances, which were already perceptible in early gynoecial development but which, due to retarded growth, became tiny, possibly functionless, appendages. In the Fumariaceae, the elaboration of the stigma considerably precedes differentiation of the ovary. Stylar growth begins long after the transverse secondary growth centers have been

238

THE BOTANICAL REVIEW

Fig. 29. Gynoecium ontogeny in Fumariaceae (1 and 2, Dactylicapnos torulosa; 3-5, Corydalis ophiocarpa; 6-9, Pseudofumaria lutea). 1. The gynoecial tube opening has become a transverse slit; primary transverse stigmatic lobes are already discernible (bud rotated by 90~ 2. Young gynoecium; style initiation separates stigmatic and ovary regions. Primary stigmatic lobes approach each other with the apical slit becoming vertical; growth centers arise at the backs of the primary lobes forming secondary lobes (cf. Fig. 13.1). 3. Gynoecial ridge with two transverse growth centers overtopping the median parts (dent on the left due to artificial damage). 4. Transverse growth centers forming the future, primary stigmatic lobes; their backs are bulged by accelerated growth. 5. Primary stigmatic lobes approaching each other with constriction of the apical opening; secondary growth centers developing horizontally. (For further development of the stigma, see Briickner, 1992b, and Fig. 13.5.) 6. Top view of young bud with the floral apex giving rise to the gynoecial ridge. 7. Transversely elongated gynoecial ridge showing two median and two transverse growth centers. 8. Accelerated growth of the two transverse growth centers leads to overtopping the lower median ones. 9. Elaboration of the stigma; below the spreading transverse stigmatic lobes a bulge is formed that surrounds the slender style (cf. ripe stigma in Fig. 13.6). The median appendages are well developed, and the narrow apical slit is recognizable between them. Sources: 4 and 5, Brfickner, 1992b; 6-9, Brfickner, 1992a.

CARPEL NUMBER

239

active. At this time, zones of dehiscence are not yet discernible along the outside of the ovary (Fig. 29.2). Direction of ovule differentiation has rarely been mentioned in the literature. Payer's (1857) illustrations show a basipetal sequence. In species of Dicentra, the ovules are said to be attached to the placentae in four vertical rows, inception of the two outer rows being earlier than that of the two inner ones. In Fumaria, each placenta bears one (Saksena, 1954) or two ovules (Payer, 1857: basipetal inception is depicted; Buchenau, 1866); however, only one develops into a mature seed. The remaining taxa with one-seeded fruits also usually produce two ovules per ovary (Lid6n, 1986).

3. Capparaceae As in the Papaveraceae, the floral apex is a flattened plug before gynoecial inception. In the bivaivate taxa, again, an elliptical extension in the transverse plane is seen. The gynoecial primordia appear more or less separately: either two in the transverse plane or several in taxa with several placentae; undivided ring primordia have also been observed (Fig. 30.1; see Payer, 1857; Eichler, 1865; Narayana, 1962, 1965; Leins & Metzenauer, 1979, Erbar & Leins, 1997; Ronse Decraene & Smets, 1997a, 1997b). Raghavan (1939), Raghavan and Venkatasubban (1941b), and Ronse Decraene and Smets (1997a, 1997b) have described some exceptional taxa in which the two carpel primordia arise in the median plane. Consequently, the placental regions are differentiated transversely. The position of valves and placentae are thus shifted by 90 ~ compared with the typical case. A less pronounced rotation of gynoecial position--by about 45~ also occur (Endress, 1992; Ronse Decraene & Smets, 1997a, 1997b). In Crateva, Raghavan and Venkatasubban (1941 b) observed successive inception ofprimordia at the flanks of an oblique plug. Generally, a continuous gynoecial ridge arises very soon (Fig. 30.2). Very low synascidiate zones may appear (Karrer, 1991); however, the gynoecial tube develops predominantly in symplicate manner. The placental regions are differentiated early and may initially display accelerated growth, overtopping the future valve backs (Capparis, Podandrogyne). In other taxa, the upper margin of the gynoecial tube remains plain, without any bulges (Cleome gynandra, Dactylaena). Frequently, however, accelerated growth (sometimes delayed) of the valve regions occurs, so that their tips finally overtake the placental regions. The tips form carinal stigmatic lobes that may more or less fuse postgenitally (Capparis, Maerua, Oceanopapaverwith further splitting of the two lobes). However, in Cleomoideae capitate stigmas are produced mainly by the median tube parts over the placental regions, even if the transverse parts initially developed faster. (For a detailed description of these growth processes, see Karrer, 1991.) Payer (1857) states that Polanisia has carinal stigmas and Capparis commissural ones. Apparently, his conclusion is deduced from the fact that transverse bulges form on the gynoecial ridge in Polanisia, whereas Capparis shows an initial growth acceleration of the placental regions. As Karrer (1991) demonstrates, early growth patterns do not absolutely determine the final stigma elaboration: in Polanisia the main portion of the stigma is supplied by the area over the placental regions, whereas Capparis produces carinal stigmatic lobes. The early-forming placental regions look as massive as those of the Papaveraceae (Fig. 30.3). In certain stages of development they touch each other (Fig. 30.4) and may, in some taxa, fuse postgenitally in the ovary center (cf. section VII.C.3). Payer (1857) described a peculiar mode of ovule inception in Cleome and Polanisia. Initially, two rows of few ovules are produced acropetally; then additional ovule primordia are interspersed among the first formed. On the radially intruding placentae of Capparis, initially two ovule rows differentiate

240

THE BOTANICAL REVIEW

Fig. 30. Early gynoecium ontogeny (1-4, Polanisia dodecandra [Capparaceae]; 5, Erysimum cheiranthoides [Brassicaceae]; 6, Thlaspi arvense [Brassicaceae]; 7-8, Podophyllum hexandrum [Berberidaceae]). 1. Gynoecial plug with two transverse growth centers. 2. Plain gynoecial ridge without distinct septation. 3. Young gynoecial tube with dominant median parts (future placental regions). 4. Touching placental regions close the apical opening of the gynoecial tube (bud rotated by 90~ 5. Gynoecial tube with wide median parts slightly overtopping the lateral ones and touching at the base (prestage ofseptum formation); outer epidermis starts to form stellate hairs (left). 6. Gynoeeial tube with wide median parts that, compared with the transverse ones, display accelerated growth. 7. Gynoecial tube with bulky placental region. 8. Enlargement of 7.

CARPEL NUMBER

241

close to the ovary wail, followed by one or two more in centripetal direction. Within each row, the first ovules are formed halfway along the vertical extent of the placenta. Ovule initiation then progresses both acropetally and basipetally (el. Papaveraceae). Gynophore elongation takes place after final differentiation of the now somewhat bulbous gynoecium, with the zones of dehiscence already discernible in the dehiscent taxa (Karrer, 1991). According to Raghavan and Venkatasubban (I 941 b), gynophore development starts after the appearance of the ovule primordia and only gains considerable length before anthesis (Ronse Decraene & Smets, 1997a). Meristematic layers at the gynoecial base form by cell division and subsequent cell extension the organ so characteristic of the family. 4. Brassicaceae

Contrasting observations about the earliest stages of gynoecial differentiation have been reported. Early authors described the appearance of two transverse growth centers at the flattened floral apex (Payer, 1857; Eichler, 1865; Wretschko, 1868; Hannig, 1901). However, Eggers (1935), Alexander (1952), and Sattler (1973) described an elliptical girdling primordium as the starting structure in gynoecium development. Sattler (1973) did not discount that there may be initially four free primordia--two median, two transverse-- that, during inception, become interconnected by interprimordial growth. On the other hand, Merxmiiller and Leins (1966, 1967), Eigner (1973), and Zimmerli (1973) emphasized the development of two median growth centers as the characteristic mode of gynoecial initiation in the family. Zimmerli (1973) also stated that the level of cell divisions is highest in the transverse plane, result•ng in an inconspicuous rise of that region. In any case, an elliptical to tetragonal rim is formed, the median parts of which are interconnected by a receptacular bulge, thus giving rise to a low synascidiate zone (Hannig, 1901; Eigner, 1973; Sattler, 1973). The gynoecial tube develops in a symplicate manner (cf. Sessions, 1997), with four bulges recognizable on its margin, as was described by Griffith as early as 1847. The future placental regions are not overtopped by the transverse parts of the ring wall (see Bennett et al., 1995: fig. 5a). The protuberances of the median parts form the commissural stigmatic lobes; they are interconnected by transverse bulges. In many siliculaforming taxa, the gynoecial tube has no protruding parts on its rim and later forms a capitate stigma (cf. Eigner, 1973). The massive placental regions intrude into the ovary cavity (Figs. 30.5 & 30.6). Soon after their initiation two rows of ovule primordia appear along each of them. Simultaneous with initiation of ovule primordia the placental tips grow between the ovule rows centripetally and fuse with each other through a basipetal interdigitation of the epidermal cells. In most cases the postgenitally formed septum reaches the basal bulge of the synascidiate zone and merges with it (Eigner, 1973; Sattler, 1973). In the fully differentiated gynoecium the suture along the septum middle becomes virtually indiscernible. Remainders of both cuticulae are, however, recognizable when examined through a transmission electron microscope (Boeke, 1971). The zones of dehiscence are formed relatively late. Zimmerli (1973) observed the beginning of their differentiation in 1 mm high gynoecia, and Polowick and Sawhney (1986) saw first signs of their appearance in a 2 mm high gynoecium in which the style had become distinct from the ovary. The histological differentiation of the zones of dehiscence, culminating in the formation of the saddle articulation of the Brassiceae, continues during the development of fruit (Eigner, 1973). Floral ontogeny of gynophore-bearing taxa has not been studied yet. It can be assumed, however, that, as in Papaveraceae and Capparaceae, the extension of the gynophore starts when the gynoecium is essentially differentiated.

242

THE BOTANICAL REVIEW

5. Berberidaceae The gynoecium is incepted as a ring primordium at the dome-shaped floral apex, with the placental region recognizable from the beginning and initially somewhat overtopping the rim (Payer, 1857; Baillon, 1861-1862; Citerne, 1892; Kaute, 1963; Brett & Posluszny, 1982; Endress, 1989; Feng & Lu, 1998; Figs. 30.7 & 30.8). This massive growth center, which is mostly in an adaxial position, causes an oblique protrusion of the gynoecial dome. Payer (1857) considered it the tip of the floral axis surrounded by a broad carpel primordium and inferred from that an axial nature of the placenta. Because only periclinal cell divisions take place in the subepidermal layer of the gynoecial dome, Kaute (1963) interpreted the placenta as an independent primordium that is only united with the adaxial region of the ring wall. When the gynoecial tube increases in height the placental region becomes overtopped. No vertical suture is recognizable along the outside of the young gynoecium. Its apical opening allows observation of the differentiating ovule primordia (Fig. 15.3; cf. Brett & Posluszny, 1982; Endress, 1995: fig. 4G). In many-seeded ovaries they appear in a basipetal direction, the uppermost being the oldest (Payer, 1857). The rim of the gynoecial tube differentiates into the stigmatic area. It bends outward and frequently extends by folding, resulting in a convoluted stigma in several taxa. In Podophyllum, one lobe is formed over the placental region, whereas the remaining part gives rise to several folds (Figs. 15.1 & 15.3; cf. De Maggio & Wilson, 1986). Citerne (1892) and Eckardt (1937) describe the margin of the young gynoecial tube of Nandina to be three bulged: one growth center demarcates the placental region; the other two flank it. These protuberances give rise to the three-lobed stigma of the species. The scanning electron microscopy study by Feng and Lu (1998), however, does not mention any conspicuous bulges topping the gynoecial tube. The ontogeny of dehiscent gynoecia has not been studied in detail yet. Thus, no data are available on the time of differentiation of the zone of dehiscence, and stipe or gynophore development is insufficiently known. It may be assumed, however, that both zones of dehiscence and stipe are late-arising, secondary structures. Transections of the gynoecium or fruit base of some taxa (Epimedium, Vancouveria) show two locule bases separated by a low "septum" (Figs. 23.III.4 & 23.Ili.5; Kaute, 1963; Terabayashi, 1979). This structure superficially resembles that in the other families discussed: the base ofa gynoecial tube is two chambered because of a bulge across the floral apex. However, the nature of this bilocular zone is very different from a true synascidiate zone ofa bicarpellate gynoecium with the septum uniting two placental regions. In berberidaceous taxa, tissue bridges the placental and dorsal zone of the valve. Under the nl theory, if one ascidiate carpel is assumed to produce the gynoecium, its basal septation would be unusual. The gynoecium is not explained any better by the 2n theory, for the septum would be connected with the dorsal region of a carpel. In the center of the septum the bulk of the vascular tissue splits into a placental bundle system and valve venation. The impression ofa "septation" may be caused by a slight depression of the locule on both sides of that massive bridge of tissue. Detailed ontogenetic investigations are necessary to determine the time of initiation of the bulge.

6. Discussion Fundamental similarities are exhibited during ontogeny of the gynoecium in all of the families discussed. Characteristically, a girdling primordium is initiated at the more or less plane top of a plug-shaped floral apex (elliptical in bivalvate taxa); frequently, development

CARPEL NUMBER

243

of the placental regions is accelerated to a considerable degree. In no case are carpel borders clearly distinct. How, then, can the growth processes observed be explained by the individual carpel theories? In several bivalvate taxa, gynoecium ontogeny starts with the initiation of two transverse bulges (Capparaceae, Macleaya of the Papaveraceae). These structures can easily be interpreted as two carpel primordia in transverse position, fitting the nl theory. Through intercarpellary fusion of the marginal meristems the gynoecial ring arises soon. Even if its rim is plane and has no transverse protuberances, as is seen in some taxa, its transverse parts could be considered "free carpel tips," with the carpel flanks being fused at an angle of about 180 ~ (Werle-Sprengel, 1993a, 1993b). Because carpel primordia, and especially their ventral regions, cannot clearly be delimited from the oblique floral apex (see Rohweder & Endress, 1983:100, 107), the patterns of early initiation may have different interpretations, all of them equally plausible. Gynoecial initiation starting with the median parts attracted special attention. Concerning the Brassicaceae, Merxmiiller and Leins (1967:124) wrote: "Wollen wir weiterhin an der gebr~iuchliehen Vorstellung zweier, und zwar transversaler Karpelle... festhalten, miil]ten wir demnach hinnehmen, dab die Karpelle mit ihren R~indern zuerst entstehen. Ein solcher Entstehungsmodus eines Blattorgans dtirfte jedoch ohne Beispiel sein"; that is, the initiation of a phyllome beginning with the appearance of its margins is questioned. The authors accepted the 2n theory for Brassicaceae but failed to reconcile the fact that in the closely related Capparaceae-Cleomoideae, with a nearly identical gynoecium structure, two transverse primordia are the first to arise. As for the Papaveraceae, they cited Payer's (1857) observations, in which transverse growth centers support the nI theory. However, Payer stated the same for Capparaceae. As mentioned above, all of the families exhibit considerable growth acceleration in the growth of placental regions, at least temporarily. In the siliqua-forming Brassicaceae, this acceleration continues during ontogeny, so the gynoecial tube has median protuberances that finally differentiate into the commissural stigmatic lobes. In the siliculaforming taxa this pattern is modified, for the whole ring primordium grows steadily and produces a capitate stigma. In the remaining families, the regions of the future valves overtake the placental regions; therefore, the apices of the former appear as free carpel tips from that time. Usually they transform into distinct, carinal stigmatic lobes, but they may also be integrated into capitate stigmas predominated by the commissural areas (e.g., Polanisia of Capparaceae). Thus, during ontogeny the transverse and median parts of a bivalvate gynoecium undergo growth oscillations in all taxa that vary only in intensity and timing. Therefore, at some point definite growth centers arise at the rim of the gynoecial tube, or it may remain plain through steady growth. Carpel number cannot be apriori deduced from these bulges. This confirms the statement by Deroin (1997: 56): "Carpel is both a morphological and a functional unit, but not always a morphogenetical one, especially when it is implied in a congenital fusion." (He continues: "The compound nature of the primordium is then revealed only by the vascular organization at anthesis"; the possible incertitude of that postulate has already been discussed in section VII.D.6). Moreover, the actual initiation of organs may considerably precede their physical appearance in the form of bulges perceptible through a scanning electron microscope (see Endress, 1992: Sl18). In the gynoecial plug of several Brassicaceae taxa most cell division is concentrated in the transverse parts, a diminished degree is seen in the median ones, and only few cell divisions occur in the diagonal areas (Zimmerli, 1973). However, the median parts of the gynoecial plug, not the transverse ones, are the first to be elevated. Zimmerli assumes that carpel initiation has already started at this stage of floral development.

244

THE BOTANICAL REVIEW

Is there any reason, then, to consider the median parts of the gynoecial tube as individual carpels? Hanstein's school interprets early-arising placental regions as separate developmental centers ("blastemes") that are not necessarily reduced carpels but are supposed to be of an omnipotent nature initially. Blastemes were explained by the fact that their growth is more accelerated than is that of true "carpel sites," which represent only "connecting strips between the main developmental centers," that "carpel" growth overtakes the median bulges much later; and that blasteme (placental region) procambium is already differentiated before evidence of it emerges in the carpel proper (Hagen, 1873; Hanstein, ! 873; Huisgen, 1873). This concept has been applied not only to the taxa compared here but also to, for example, Reseda L. (Resedaceae of the Capparales). The developmental pattern of the gynoecium in this genus is very similar indeed (detailed by Sobick, 1983). In R. luteola L., on a gynoecial plug, a ring meristem arises with the young placental regions clearly overtopping the carpel backs. Afterward, the latter overtake and form long, free tips. In other Reseda species the growth of the carpel tips is accelerated compared with that of the placental regions. Although gynoecial development is thus easily comparable with the patterns discussed above, Reseda has never been ascribed more than three carpels except by Hanstein's school and Saunders (1923, 1926, 1937). In Pseudofumaria (Fumariaceae) median bulges are present at the beginning of gynoeeium development, well before differentiation of the placental regions. Eventually they become ancillary stigmatic appendages. Considerably later, but still early in development, protuberances arise over the placental regions in Eschscholzia (Papaveraceae), which transform into well-developed, receptive stigmatic lobes. According to Saunders (1925), the subdivision of the Eschscholzia stigma changes over the course of the season. At the beginning of the flowering period most splitting of the stigmas occurs; in addition to the median lobes they bear several appendages in the carinal position. At the end of the flowering period, however, the median lobes are often lacking, and the two carinal lobes are no longer subdivided, being the only parts of the stigma that remain; the ovaries do not show structural differences. Perhaps the seasonally induced lack of the median stigmatic lobes is a local phenomenon (four-parted stigmas were found in the majority of cultivated E. californica plants in early November 1994). Occasionally, Pseudofumaria gynoecia also lack median protuberances (see Briickner, 1992a: fig. 16). This variability in stigma structure suggests sporadically diminished growth of the median parts rather than absence of whole gynoecial components (carpels), for the organization of the ovary remains unaltered. The expanded commissural stigmatic lobes in some Papaveraceae taxa (Glaucium, Argemone) develop late during stigma elaboration, when the ovary is largely differentiated. These median structures are obviously produced through secondary growth processes to enlarge the receptive stigmatic surface. A further considerable increase occurs, moreover, only after anthesis in the developing fruit. Proponents of the 2n theory emphasized the somewhat tetragonal shape of the ring primordium in bivalvate taxa, having massive bulging edges in the transverse and median planes (see Eigner, 1973). However, the free space at the floral apex is essentially determined by size and position of the next older primordia; in this case, the neighboring androecial whorl restricts the gynoecium site. In the Fumariaceae, the distinctly tetragonal shape of the ring is caused by the arrangement of the already well developed anthers. This also occurs in Brassicaceae. Similar observations have been made in Lauraceae (Endress, 1972), in which a threeedged primordium of a single carpel is surrounded by a trimerous stamen whorl, and in Solanaceae (Parashar & Singh, 1986), in which the five edges of the gynoecial ring primordium, consisting of two carpels, alternate with the five stamen primordia.

CARPEL NUMBER

245

With the whole range of variation of gynoecium ontogeny in mind, the nI theory has a clear overall advantage, whereas the 2n theory seems better to explain isolated modes of ontogeny (e.g., in siliqua-bearing Brassicaceae taxa). At least in closely related taxa, an identical ground plan of the gynoecium is to be expected, however; thus, hypotheses are necessary to explain 2n carpels in all forms of gynoecia present. In reference to Brassicaceae, Celakovsk~, (! 876) wrote that although the median bulges could be explained as independent carpels, such an explanation is neither necessary nor probable. Unfortunately, much of the old literature has been overlooked by the more recent botanists. Certainly, the existence of some tissue belonging to extremely reduced carpels within the placental regions cannot be absolutely ruled out. IfReseda were examined for such--for example, in Saunders's mode---a new interpretation not only of valvately dehiscing gynoecia but paracarpous gynoecia in general could result. In some cases the reduced carpels would be revealed by their free tips; in many others they would be totally unrecognizable. In .4chlys (Berberidaceae) Endress (1989) found that despite chaotic phyllotaxis of the androecium the gynoecium occupies an amazingly constant position; namely, with the placental region turned toward the inflorescence axis. The development of the adaxial half of the flower is distinctly retarded. If there were two carpels in the gynoecium, the reduced one should also be in adaxial position--that is, in the placental region-whereas the abaxial position of the normal carpel would correspond to the larger abaxial part of the flower. This might argue for the 2n theory; Endress, however, supported the nI theory. Van Heel (1978; see sec. VII.A.6) demonstrated for the Malvaceae tribe Ureneae the ability of strongly reduced carpels to produce functioning stigmatic lobes. In contrast to the taxa discussed here, the diminished carpels are initiated as easily discernible primordia that differ little from those of the unreduced fertile carpels. Only later is there a clearly observable retardation of development of one carpel whorl, not seen in the taxa under concern. Therefore, an even greater carpel reduction has to be assumed--which, of course, is fully hypothetical. Thus, the 2n theory has a weak ontogenetical base. Gynoecium ontogeny can be applied to other theories as follows: 1. The nI theory: Two carpels confluent through meristem incorporation and meristem fusion sensu Hagemann 0975) and Werle(-Sprengel) (1984, 1993a, 1993b) can give rise to all variations in the developmental pattern seen in bivalvate gynoecia. Accelerated growth of the united carpel margins (placental regions) compared with that of the carpel backs has a functional background and seems to be rather frequent in paracarpous gynoecia. Obviously, the single Berberidaceae carpel is fully ascidiate and has an especially massive placenta. 2. The nII theory: Carpel borders are not demarcated in either the transverse plane or the median plane, and the existence of median carpels has not been refuted. As stated previously, dorsal placentation would be an extraordinary exception and difficult to explain. Protuberances arising by accelerated growth of parts of the gynoecial ring primordium (here the future placental regions) need not be identical with carpel tips, as is clear from Reseda luteola. Eigner (1973: 400) wrote (after having pleaded for four carpels in the Brassicaceae): "Trotzdem scheinen mir die Argumente nicht so stichhaltig zu sein, als dab sic die Sprattsche Ansicht zweier medianer Karpelle v611ig ausschliel3en k6nnten"; that is, he feels unable to disprove the concept by Spratt (1932), but, like the former author, he is too fixated on the Brassicaceae and has not compared the developmental patterns of other bivalvate taxa. 3. The 2n theory: Ontogeny can neither prove nor refute variants A, B, and C, which are, however, much more complicated than the nI theory. Variant D is highly improbable. Diagonally positioned growth centers would have to be expected then, but exactly in these positions the most insignificant development has been found. Variant E cannot be confirmed either: there is no indication that median carpel dddoublement leads to the formation of an addorsed carpel pair.

246

THE BOTANICAL REVIEW

4. The 3n theory: This theory is not based on ontogenetic evidence. In each of the placental regions there should be two growth centers, but this is not the case. 5. Saunders's theory: This author also ignores ontogeny but would surely have interpreted developmental patterns in the sense of carpel polymorphism. There are no ontogenetic indications for several carpels composing the "compound valve" said to exist in Eschscholzieae. 6. Other theories: Clos (1865) and Trgcul (1873) had in mind the ontogeny of polyvalvate Papaveraceae gynoecia (especially in Papaver)when they were interpreted as products of the axis. The reason is the initiation as a ring primordium without free carpel tips. Clos referred to the hypothetical Stengelfruchtknoten(axis ovaries) coined by Schleiden (1843) and supposed similar conditions, but only regarding the polyvalvate taxa. As Payer (1857) had described two transverse primordia as the initial stage of bivalvate gynoecia, Clos accepted their foliar nature, too. Tr6cul, however, considered all Papaveraceae gynoecia transformed axes. More than a century after the publication of Schleiden's paper, a similar concept was proposed by Comer (1958, 1963) for syncarpous gynoecia in general, which are interpreted as intercalary tubes bearing rudimentary, free carpels at their upper rims. The lack of sharp delimitation between caulome and phyllome was discussed in section V. Therefore, the neutral descriptive terms proposed by Sattler (1973, 1974) should be preferred to describe gynoecium ontogeny. However, they are explicitly not meant for tracing carpels in syncarpous gynoecia in all cases, and in this respect they do not help to answer the question under consideration. F. TERATOLOGY Flowers that deviate from the normal structure (terata) are often striking and, especially in the eighteenth and nineteenth centuries, attracted the attention of many botanists. It was hoped to deduce, from aberrant cases, the type of organization and perhaps also the phylogeny of the normally developed form. The approaches to teratological phenomena, however, fall into two camps, which Arber (1931 b: 200) characterized: "On the one hand are ranged those who consider abnormalities as of no interest or importance, and who generally avert their eyes from them altogether. The opposing camp contains those who regard deviations from the normal as providing one of the happiest hunting-grounds in which to track down ancestral characters." These approaches are reflected in the emphasis placed on floral teratoiogies in the families under discussion. Bemhardi (1843: 64-65) expresses his hope that, through a teratological flower, the ground plan of the Brassicaceae flower might sometime be unambiguously elucidated: "Es k 6 n n e n . . , nur fiber die Staubf~iden und die Karpellen verschiedene Ansichten gehegt werden, und so lange nicht eine ungew6hnlich gebaute Blfithe unerwarteten AufschluB gewahrt, werden dieselben sich wohl schwerlich vollkommen vereinigen." Tr6cul (1873: 140), however, warns against the foundation of conclusions on teratological structures: "Quelque s6duisant que soit I'examen des formes vari6es que peuvent prendre les parties de la fleur par des d6veloppements anormaux, il est urgent de renoncer aux conclusions illusoires qui en ont 6t6 d6duites, puisque des monstruosit6s diverses peuvent conduire aux avis les plus oppos6s." In the same sense, Lignier (1904) urges Gerber to use arguments deduced from terata only with greatest restraint; Gerber (1904e) agrees with Lignier regarding that aspect he had formerly much overestimated. Terata play an important role in the discussion of carpel numbers. In this article, however, only gynoecial structures are taken into consideration; omitted are pathological forms that, according to Heslop-Harrison (1952), are merely quantitative products of uncontrolled growth and have little morphological significance. Remaining are morphogenetically transformed organs that may allow conclusions concerning homologies. In all families discussed

CARPEL NUMBER

247

here similar terata have been observed. The individual phenomena, occasionally present in the taxa, will be considered below.

1. Homeotic Phenomena The term "homeosis" was coined by Bateson (1894) and redefined by Leavitt (1909); in more recent times Sattler (1988, 1994) has discussed comprehensively its significance in botany. Homeosis is the positional replacement of one structure with a different structure. This replacement may be total or partial; in the latter case only parts of the character set of the given structure are transferred to the new position. Partial homeosis is the much more frequent phenomenon, because structures that seem to be totally transformed may nevertheless show transitions to the replaced structure, at least in ontogeny. Thus, partial homeosis is expressed in intermediate mosaic organs (developmental hybridization sensu Sattler, 1988). The characters transferred may be structural as well as functional. Many teratological eases may be explained by homeosis. This is also true of malformed flowers that have been found in many taxa of the families under concern. They are discussed below. a. Gynoecia in Virescent Flowers Virescence of flowers (frondescence, phyllody, phyllomania, chloranthy, antholysis) means the more or less pronounced occurrence of foliar characters within the floral whorls. This phenomenon is an example of partial homeosis if compared with a vegetative pbyllome; absolute identity with green leaves with the size and shape of the taxon has never been found. The homeoticaUy transformed organs display a more or less bracteose appearance. The striking inflorescences of such plants, which deviate conspicuously from the normal habit, have attracted the botanists' interest for centuries, and many works, especially from the nineteenth century, analyze this phenomenon. As to the families concerned, virescent berberidaceous flowers have never been described. The observed cases in Capparaceae and Fumariaceae are very scarce, and this mode ofhomeosis is seldom seen in Papaveraceae, either. In several taxa of the Brassicaceae, however, virescent flowers have repeatedly been found. This unequal distribution of cases may partly be due at least to the fact that Brassicaceae usually occur in big populations and may be cultivated; virescent plants thus may be more frequent and conspicuous than in the other families. The factors that cause virescence are not fully known. Frequently, malformed plants have been infested by parasitic fungi of the Oomycota order Peronosporales (Albugo candida = Cystopus candidus, White Rust of the Brassicaceae; Peronospora parasitica). Also, the activities of parasitic animals (aphids, mites, larvae of beetles) can set off virescent development, as was experimentally demonstrated by Peyritsch (1882) and Gallaud (1926). Janse (1929) postulated a variation in quality of a "growth-enzyme," not actually a single substance but an intricate multifactorial network, that causes phyllody by retarded differentiation. Furthermore, spontaneous mutative effects play a role: at the end of the nineteenth century De Vries (1894, 1895, 1896 [cited from Penzig, 1921]) succeeded in proving some heredity of virescent disposition. A treatment with synthetic auxines (2,4-D) enables experimental induction of phyllody (Gu6d6s & Dupuy, 1964). A "green" appearance of the flower is also displayed by several homeotic mutants ofArabidopsis thaliana (L.) Heynh., a plant that has been the subject of intensive genetical studies of floral development. A broad range of ectopic expression of leaf or bract characters in all floral whorls is seen in single, double, and triple mutant strains (Irish & Sussex, 1990; Coen & Meyerowitz, 1991; Clark & Meyerowitz,

248

THE BOTANICAL REVIEW

1994). In particularly "green" looking flowers the petals are homeotically converted or lacking and have thus lost their showiness. Appendix 2 lists chronologically relevant original papers dealing with virescence in members of the families concerned. Surveys on observations of spontaneous cases are also given by Engelmann (1832), Moquin-Tandon (1842), Masters (1886), Worsdell (1915/1916), Penzig (1921), and Meyer (1966). The degree ofvirescence within the cruciferous inflorescences increases acropetally. The structure of the lowest flowers may appear to be totally unaffected. The gynoecia display gradually progressing homeosis and may be arranged in the following series of transformations (see also Gagnepain, 1900): 1. A long gynophore is produced, elevating the gynoecium conspicuously above the perianth (also seen in Argemone of Papaveraceae). Several authors interpret the gynophore as proof of a close relationship between Brassicaceae and Capparaceae, with a gynophore being consistently formed in the latter. 2. The ovary is shortened (silicula form also in normally siliqua-bearing taxa) and bloated; the top is especially dilated. The ovary wall is wrinkled. 3. The septum is reduced or lacking. The ovules are transformed into ascidiate to flat leaflets that may retain traces of integuments and nucellus. Celakovsk~ studied the morphology of such ovules in detail and published the results in several works (e.g., 1875, 1884). 4. The carpels separate in basipetal direction. In strongly virescent gynoecia they are replaced by two green leaves, free to their bases. In Argemone the number of phyllomes observed is three or four. At this stage there are almost no signs of ovule development. If there are still septum rudiments, they originate from the outside (= morphological underside) of the phyllomes. The more leaflike the carpels are, the more their dorsal veins (midribs) are developed. This phenomenon is especially striking in taxa whose normal carpels have no dorsal vein (Argemone). Rohweder (1959-1960) studied in detail the course of the vascular bundles in the placental regions of increasingly virescent gynoecia of Barbarea (Brassicaceae). In largely closed ovaries, the typical concentric placental bundle systems were found at least at the base. With progressing homeosis, the placental bundle systems were transformed into two strong bundles that, running in parallel direction, supplied as lateral veins the margins of the apically free phyilomes. Virescent gynoecia provide some evidence of carpel numbers. The overwhelming majority of bivalvate phyllodic gynoecia consist of two transversely positioned phyllomes with marginal placentation. Deviations from that composition are very scarce. According to Penzig (1921), Wirtgen found in a virescent flower of Corydalis cava (Fumariaceae) four "leaflets" replacing the ovary, and Chodat (1888) described a flower of Capsella bursa-pastoris(Brassicaceae) with a four-merous, cross-shaped structure in place of the ovary. These unusual structures can be ignored because of the abundance of diphyllous pistil transformations. In the virescent flowers of Argemone mexicana that Joshi (1933, 1939) observed, the number of leafy organs corresponds to the number of valves in normal gynoecia. Hence it is not surprising that all authors who have dealt with phyllody maintain the nI theory. This concept is further confirmed by the gradually changing vascular pattern in the basipetally splitting placental regions. For the 2n and 3n theories, only Gerber (1900a, 3n theory) gives his view on two-part virescent gynoecia: that the two carpel pairs forming the placental regions vanish acropetally, enabling the sterile transverse carpels to separate. However, this opinion is challenged by the following: the occurrence of rudimentary ovules along the free margins of the phyllomes, and

CARPEL NUMBER

249

Fig. 31. Carpellomania in a cultivated plant ofArgemoneochroleuca Sweet. 1. Postanthetic flower with the stamens feminized to varying degrees. 2. Same flower enlarged; feminized stamens flattened with bristly backs, smooth inner surfaces, a wavy stigmatic area, and marginal anatropous ovules. Scale bars ---0.5 cm.

250

THE BOTANICAL REVIEW

the vascular pattern, with the bundles of the supposed dwindling carpels not vanishing as well but becoming the lateral veins of the free phyllomes. If fertile replum and sterile valve carpels are assumed (2n theory, variant A), a splitting of the fertile carpels and merging of the halves with the margins of the sterile carpels must be postulated to explain the transformation. However, this is a superfluous theoretical complication that corresponds to Saunders's 89+ 1 + 89 pattern in carpel polymorphism. Variants B and C demand the complete disappearance of the replum carpels in virescent gynoecia--which is rather improbable. The number of free phyllomes in strongly transformed gynoecia may not necessarily reflect carpel number in normal ones. An example is the flowers ofPrunus serrulata Lindl. (Rosaceae), which are monocarpellate but, in doubled state, usually produce two free leaflets instead of a pistil (see Gu6d6s, 1966a; Weberling, 1981). This phenomenon can, however, be explained as a teratologicai increase in carpel number due to the process of doubling. Applied to the bivalvate taxa of interest here, the concept would demand that at least four or six leaflets (for 2n or 3n carpels) or even the double number (by analogy with the Prunus case) replace the gynoecium. Thus, the nI theory cannot be invalidated by this approach either. b. Petaloid Carpels In the taxa under consideration, this rare transformation is known only in species of Papaver. Goethe (cited from Clos, 1862) observed petaloid stigmatic lobes in Papaver somniferum. Crepin (1866) described petaloid carpels in P. rhoeas, and Godron (1871-1872) studied the same phenomenon in a doubled form ofP. somniferum. In addition to totally homeotic organs, specimens were found with only the carpel tips being free and colored, whereas the green basal parts were fused and bore bulky ovuliferous placentae; several of the ovules proved to be fertile. The placental bulges split acropetally, became sterile, and continued as narrow white margins along the petaloid carpel parts, here apparently having a stigmatic character. Godron (1871-1872) considered this structure evidence of the nI theory, for the placentae are clearly borne by the carpel margins. Indeed, there are no hints of additional reduced carpels within the placental regions. c. Carpelloid Stamens A rather frequent spontaneous teratology in di- and monocotyledonous taxa is the occurrence of carpel features in the androecial members. Stamen primordia develop into hybrid organs producing stigmatic papillae or well-developed stigmatic lobes and ovules. The pollen sacs may remain more or less unchanged and contain fertile pollen or display all stages of reduction. In the last step, "perfect" carpels replace the stamens. If (nearly) all stamens of a flower, espeeially a polyandrous one, are thus transformed, the phenomenon is called "pistillody," "carpellody," or "carpellomania" (e.g., Masters, 1886; Penzig, 1921; Gu6d~s, 1972: detailed phenotype analysis). The transformations are caused mainly by recessive single-gene mutations. Moreover, certain environmental factors additionally influence sex in the androecium. One factor is nutrition: in Papaver mutants, female characters are expressed only in the best-nourished individuals, whereas dwarfed plants never show carpelloid stamens (De Vries, 1906). Furthermore, some mutations proved to be temperature sensitive. A certain temperature regime is needed to produce carpelloid stamens: temperatures too low or too high alter the phenotype, and the primordia differentiate into more or less normal stamens (see Polowick & Sawhney, 1987; Bowman et al., 1989). Application ofphytohormones (gibberel-

CARPEL NUMBER

251

lic acid) can induce stamen transformation in nonmutants (Rylski, 1986) as well as normal development of otherwise carpelloid stamens in mutants (Sawhney & Greyson, 1973). An experimental shift of the photoperiod regime at the time of flower initiation may also give rise to a transformation of stamen primordia into carpels (Fisher, 1972). The occurrence of carpelloid stamens varies in frequency in the families under discussion (see the literature survey in Appendix 3). No case has been known in Fumariaceae yet; changes in the structure of the stamens tightly enclosed by the perianth may, however, easily escape observation. Berberidaceae and Capparaceae hardly tend to produce homeotic androecial members. Strikingly transformed organs were described in a Podophyllum population (Sawyer, 1926) and occasionally observed in flowers aberrant as a whole and then inserted at secondary, stamen-borne axes (Marchand, 1863-1864) or parts of extra florets in the axils of the sepals (Morini, 1891). In Papaveraceae, carpelloid stamens are found almost exclusively in the Papavereae, especially in Argemone and Papaver. Carpellomania ofPapaver taxa is a frequent phenomenon. In the Brassicaceae, Cheiranthus cheiri displays the total homeotic replacement of stamens by free or united carpels so frequently that in de Candolle's "Prodromus" (1824: 135) the variety gynantherus is based on that form ("antheris nempe in carpella mutatis").As to other cruciferous taxa, the spontaneous occurrence of the phenomenon is much rarer. However, several experimentally generated Arabidopsis mutants develop identical hybrid organs, as is described below. The structure of carpelloid stamens in Papaveraceae and Brassicaceae is analyzed below (see also Briickner, 1996a). i. PAPAVERACEAE In the family, two categories of feminized androecia have been known: compound structures and free organs. Compound structures: There are two types of compound carpelloid androecia, which have been observed only in species of Papaver. The more common type is seen in flowers with a regularly developed pistil surrounded by a large number of diminutive pistils that replace the androecium (Fig. 32.1 ). In transection, the additional pistils show few (frequently one to four) placentae that slightly intrude into the locule but usually bear numerous ovules which may be fertile (Figs. 32.1C & 32.1D). The number of stigmatic rays at the lobed stigmatic disc corresponds to the number of placentae; the spreading lobes conspicuously extend downward. The diminutive pistils are borne, singly or in small groups, on long stalks (Figs. 32. I A & 32. I B) that are homologous with the lower regions of the filaments (in Appendix 3, literature on this phenomenon is marked "DP"). Because such abnormal single-flower "infrutescences" have attracted much interest, several horticultural cultivars have been selected from Papaver somniferum in the last century ("polycarpum," "polycephalum," "monstr(u)osum," "Hen and Chickens"). Hoffmann (1877, 1878, 1881 ) and De Vries (1901 / 1903, 1906) have studied the pattern of inheritance of that feature. Another, rarer type of compound structures develops through lateral fusion of few to many transformed stamens. They form a disrupted to more or less perfect carpelloid sheath around the central pistil that projects through the apical opening (see De Vries, 1906). Several whorls of encaptic sheaths have also been observed (Magnus, 1876). Free organs: This occurs frequently in individuals of Papaver and Argemone species and was observed once in Macleaya (Gris, 1858). A multistaminate flower displays a transformation series from apparently normal stamens in the outermost whorls via intermediates to apparently perfect, but open, "carpels" in the inner whorls (see Guides, 1969). The following steps of stamen transformation are seen: The filaments appear normal, but have, instead of

252

THE BOTANICAL REVIEW

Fig. 32. Carpelloid androecia. 1. Papaver species: A. P. somniferum (redrawn from Groenland, 1860), capsule surrounded by stalked diminutive capsules of androecial origin; B-D, P. orientale L. (redrawn from Mori/;re, 1862); B. Group of diminutive pistils with united bases; C. Longisection of a diminutive pistil; D. Transection of diminutive pistils having one to four placental regions less bulky than in regular ovaries. 2. Perfectly earpelloid stamens: A. Argemone ochroleuca; B. Papaver nudicaule L.; C. Cheiranthus cheiri (redrawn from Duchartre, 1870-1871). 3. Semischematic transection of a "double" gynoeeium (redrawn from Sirks, 1924) with the inner one (hatched) of regular state and the outer one formed by six joined carpelloid stamens; the lateral veins of the latter show varying degrees of fusion, septum precursors are distinct. 4. Papaver rhoeas, semischematic transection of a tetrasporangiate anther with three vascular bundles (hatched) and two developing ovules (dotted) between the pollen sacs of each theca (redrawn from Carano, 1911). 5-7. Cheiranthus cheiri (redrawn from Duchartre, 18701871). 5. Androecium replaced with six free, carpelloid organs. 6. A compound "outer" gynoecium consisting of four fused units of androecial origin encloses the regular pistil (stigmatic tips protruding); valve-demarcating zones of dehiscence are present. 7. Semischematie transection of the compound structure depicted in 6, showing fusion of the four earpelloid stamens with each other and with the central gynoecium. Source: Brfickner, 1996a.

CARPEL NUMBER

253

one, three vascular bundles that fuse in the filament base (Fig. 33.3). Stigmatic papillae arise at the tip of the connective; with proceeding feminization, it elongates into a stigmatic lobe. In Argemone, an increasing number of bristles appear at the broadening connective back; normally, these bristles characterize the carpel exterior (Figs. 31.1 & 31.2). The first ovules arise at the anther base between the pollen sacs in the stomium region of each theca (Figs. 32.4 & 33.7). These areas enlarge gradually to form placentae. The pollen sacs shorten acropetally, the two outer vanishing first. The filaments broaden and flatten, the placentae extending downward along their edges. Of the three veins, the two laterals supply placentae and ovules. A unilateral carpellization of an anther has also been observed in Argemone, with one theca becoming a ovuliferous placenta and the other being in the normal state with two fertile pollen sacs (Fig. 33.4). Schimper (1829) found transformed Papaver androecia in which, among normal stamens, several "carpels" were interspersed in addorsed positions, turning their ovuliferous ventral side toward the perianth. Magnus (1876, 1877) also described addorsed carpelloid structures that, along their lower halves, were fused with the central pistil. Whereas he considered these structures outgrowths of the gynoecium, Celakovsk~, (1878) counted them among carpelloid stamens. In Argemone, individual "carpels" rotated by 90 ~ were observed by the present author (Fig. 33.1). A perfectly developed, open "carpel "ofPapaver (Figs. 32.2B & 33.6) has a convex lamina and a long, acute stigmatic lobe of triangular shape that is bent toward the center of the flower. The margins of the stigmatic lobe are covered by a strip of receptive papillate tissue; their lower edges are somewhat protruding. The bulky placentae bear numerous ovules and extend to the base of the organ. They appear to be in a submarginal position, but because of the considerable thickness of the organ edges ("lateral faces" after Puri, 1961) the exact position of its margins cannot be determined (see also Gu6d~s, ! 969). Three main vascular bundles run in a parallel course through the blade; the lateral ones serve as ovular supply. A likewise perfect carpelloid organ of Argemone (Figs. 32.2A & 33.1) has a flat to slightly convex lamina; its back is densely covered with bristles like a normal carpel. A few bristles may occasionally be formed on the otherwise smooth inner surface. The placentae are narrow marginal bulges beating many anatropous ovules. The broad stigmatic region is wavy to plicate and consists of several confluent lobes that extend downward along the margins of the organ outside the placentae (Fig. 33.2, cf. Sachar, 1955). Their inner surface is lined with receptive papillate tissue. The largest lobes are formed over the placentae, whereas the median part of the stigma is somewhat bifid (see also Lewis, 1912). The main venation consists of the pinnately branching dorsal bundle and the two lateral ones with diverging ovular traces; between the main bundles is a comparatively dense network of horizontally anastomosing secondary veins. The higher the degree of feminization of the organ, the denser is this network. ii. BRASSICACEAE A complete transformation series has not been found in a single flower, probably because of the small number of six stamens. As Polowick and Sawhney (1987) have shown within a line ofBrassica napus characterized by cytoplasmic male sterility, an experimental induction of hybrid organs in the position of the four longer stamens is possible through gradually lowered temperature regimes. The mosaic organs had naked ovules at the anther base and stigmatic tissue at its distal end. More transformed structures were devoid of filaments, fused in pairs or groups of all four, and made up of stigma, short style, and ovary with distinct zones of dehiscence; several of the marginally attached ovules were fertile. A similar transformation series was produced, through gradually increased temperatures, in the homozygous, tem-

254

THE BOTANICAL REVIEW

Fig. 33. Carpelloid stamens in Papaveraceae (1-4, Argemoneochroleuca; 5-7, Papavernudicaule). 1. Largely transformed androecium (black arrow points to a normally developed stamen); carpelloid stamens with pinnately branched dorsal vein (white arrow). 2. The undulate stigmatic area (arrow) extends downward along the carpelloid stamens. 3. Partially carpelloid stamen with three veins subbasally united (arrow) running in the filament; its anther is transformed, showing apical stigmatic tissue and bearing anatropous ovules at its margins. 4. Unilateral carpellization of an anther viewed from the bristly dorsal side; fertile theca at the right, ovuliferous placenta at the left. The feminized region is darker, with reticulate venation; male part has pinnate venation and a submarginal vein. 5. Longisection of opened bud, showing largely transformed androecium; the central gynoecium (asterisk) is surrounded by carpelloid stamens, with some normal stamens interspersed (arrow). 6. Carpelloid stamens with elongated, triangular stigmatic lobes bent toward the center. 7. Partially transformed anther with ovules between the pollen sacs ofa theca (arrow). Scale bars = 1 mm. Source: 1,4, 6, and 7, Br0ckner, 1996a.

CARPEL NUMBER

255

perature-sensitive mutant apetala3-1 ofArabidopsis thaliana (Bowman et al., 1989). Again, the two shorter stamens were less affected. R6bbelen (1965) describes another sterile mutant, in which the terminal flower had not only six carpelloid stamens but also an octomerous carpelloid perianth, all transformed organs being free and open (for genetic control of this phenotype, see Bowman et al., 1991; Coen & Meyerowitz, 1991). Spontaneous carpellody in Cheiranthus cheiri has rarely been confined to the sites of the two shorter stamens with the four longer ones transformed into reduced, sterile organs (Duchartre, 1870-1871); a flower with only one of the short stamens feminized has also been found (Gay in Brongniart, 1861). In the majority of cases, however, all six stamens are affected, with either two carpelloid organs or only one replacing a pair of the longer stamens. The carpelloid structures may be free and expose the ovules (Fig. 32.5). More commonly they develop in a marginally united state around the central regular pistil, the stigma-bearing style of which projects through the apical opening of the secondary"pistil" (Figs. 32.3 & 32.6). The two pistils may also arise fused, thus producing either more or less regular, six- to eight-chambered structures (Fig. 32.7) or, more frequently, various irregular structures varying in degree of septation (see Ducbartre, 1870-1871; Worsdell, 1915/1916; Sirks, 1924). A total lack of the central normal gynoecium has been observed by Sirks (1924). Both Sirks and Nelson (1929) studied the recessive mode of inheritance of the transformation. Holland (1882) and Spinner (1904) found irregular compound structures, the closed central unit of which was surrounded by several carpelloid organs, partly open, partly fused with each other and with the central unit; furthermore, curved filamentoid structures without anthers joined the complex. Because of the presence of apparently reduced stamens, Holland questioned the androecial nature of the additional carpelloid organs. A strikingly similar gynoecial complex combined with filamentous organs is, however, displayed by homozygous pistillata mutants of Arabidopsis. As was shown for these phenotypes, the stamen primordia fuse soon after inception with the gynoecial meristem, thereby causing a more or less perfect integration of their products into a compound gynoecial structure (see Pruitt et al., 1987; Bowman et al., 1989; Hill & Lord, 1989; Coen & Meyerowitz, 1991; Goto & Meyerowitz, 1994; Krizek & Meyerowitz, 1996); apetala3 mutants have a rather similar phenotype (Irish and Sussex, 1990; Bowman et al., 1991; Jack et al., 1992, 1994). Goto and Meyerowitz (1994) demonstrated that the PISTILLATA and APETALA3 proteins may act together in regulating P! and other genes. A comparable, probably homologous, genetic factor may have been responsible for the gynoecial abnormalities in the Cheiranthus flowers described by Holland (1882) and Spinner (1904), and the multichambered gynoecial organs consisting of several whorls in a plant of Brassica oleracea var. botrytis depicted by Yen (1957). A free carpelloid stamen of a completely transformed Cheiranthus androecium is a sessile, elongated organ with a convex lamina (Fig. 32.2C). The submarginally inserted ovules are mostly well developed. The "carpel" edge proper is a bulge of transmitting tissue that corresponds to a halved half-septum (see Gu6d~s, 1964) and is capped by a stigmatic protuberance; thus, the stigma is bifid (emarginate). The lamina has three main vascular bundles; the lateral ones, supplying the ovular traces, may be more or less inverted. In fused margins of neighboring organs, concentric or radially arranged bundles with the xylem poles facing each other may occur (see Gerber, 1904a; Gu6d~s, 1964). Corresponding organs of Arabidopsis and Brassica strongly resemble those of Cheiranthus but possess capitate stigmas, as do the normally developed pistils of these taxa. Are such carpelloid organs models of normal gynoecium elements (commonly called "carpels" or, more neutrally, "pistil-forming units" by Okada et al., 1989) of Papaveraceae and Brassicaceae? There is reason to suppose that the formation of a n-valvate gynoecium could be possible with not more than n of such organs present.

256

THE BOTANICAL REVIEW

1. Structure of the stigma: The sectors of the flat stigmatic disc of Papaver apparently consist of segments corresponding to the centripetally bent, acute stigmatic lobes of the carpelloid stamens found in the genus. The marginal strips of papillate tissue of neighboring lobes would give rise to the double stigmatic rays over the commissural areas. Also, the spreading parts of the disc margin over the placental regions are derivable from a combination of two protruding lower edges of the free stigmatic lobes. The normal stigma of Argemone has its largest areas over the placental regions; the same structure is displayed by the stigma of the carpelloid stamen. The large lobes over the placental regions are thus not part of additional carpels. The same is true for the convoluted stigmatic area in some Berberidaceae taxa, as carpelloid stamens of Podophyllumpeltatum have similar stigmas (Sawyer, 1926). The stigmatic lobes over the placental regions in the Brassicaceae are not unusual when compared with the emarginate stigma of a single carpelloid stamen of Cheiranthus. A marginal fusion of two such stigmatic structures would result in a typical commissural stigma, which Brown (in Horsfield, 1839-1852) supposed to originate from two bifid carpels that looked like carpelloid stamens known to him. Once again, the morphology of carpelloid stamens renders hypothetical solid carpels superfluous. 2. Placental regions and septum: The bulky, multiovulate placentae of Papaver are recognizable in the free carpelloid stamens. The less prominent placental strips of the carpelloid Argemone stamens correspond to the slightly intruding placentae of the normal gynoecium. In Cheiranthus and Arabidopsis, carpelloid stamens bear one row of ovules along each submarginal zone. In the fused state normal placentae would be found, particularly because the combined marginal bulges of transmitting tissue would produce a well-recognizable septum. Its occasional incompleteness does not argue against this derivation, for septum halves that do not meet in the ovary center are also seen in the wild type of some taxa as well as in several generated mutants. A median placentation as prerequisite for the nil theory can clearly be refuted. 3. Venation pattern: Even the most perfect papaveraceous "carpel" stages are not totally homeotic, because a median vascular bundle ("dorsal vein") is present. In normal gynoecia of Argemone, rudiments of these bundles are seen at the extreme base, whereas in Papaver there is no trace of them. Instead, precursors of, and well-developed, pseudodorsal veins supply the areas between the placental regions. However, partial homeosis may be accepted as total at the family level, because the lack of valve dorsal veins is a clearly advanced character state: the more primitive Chelidonieae and Eschscholzieae possess true dorsal veins that extend to the stigma. The marginal bundles of the carpelloid stamens produce, in fused state, a configuration fully similar to that in normal placental regions. The occurrence of concentric bundles and bundle systems with inner inverted fractions points to functional aspects of this arrangement of vascular tissue. Hypotheses concerning involute carpel margins, the existence of additional and even of addorsed carpels, and so forth, are superfluous for explanation of the vascular pattern. 4. Zones of dehiscence: Whereas in carpelloid stamens of Papaveraceae zones of dehiscence are not distinct, in the corresponding organs of Brassicaceae valves are more or less clearly demarcated by a continuous zone of separation tissue running back in itself within the lamina. Although final "dehiscence" of earpelloid stamens has not been described, this structure is undoubtedly homologous to the zones of dehiscence bordering normal valves. Thus, the opercular nature of the valves is confirmed. Because of the correspondence between carpelloid stamens and "true" gynoecial units, many morphologists support the nl theory (Allman, 1852; Le Sourd-Dussiples & Bergeron, 1861; Godron, 1871-1872; Braun, 1874; Schilberszky, 1893; Marchand, 1896; Gu6d~s,

CARPEL NUMBER

257

1964, 1969). Favorably, the complex of female characters is transferred here to organs, the primordia of which are initiated separately and do not merge, by meristematic fusion, into a ring stage. Thus, free organs arise that display a carpel structure very specific for the taxon concerned. Regarding the Brassicaceae gynoecium, Arber (193 la: 39) argued: "It may never, in the whole course of its history, have passed through a stage in which it could be said to consist of distinguishable leaf members. To ask how many carpels are involved in such a gynaeceum is a purely scholastic question which can never receive an answer, because no answer exists." She strongly tends, however, to accept the nl theory, and the structure of the carpelloid stamens can well be used to corroborate this theory. Additional carpels are not necessary to explain the structure ofvalvate gynoecia; nor is there much probability that syncarpous gynoecia are acarpellate intercalary tubes, for the occasional appearance of separate carpel-like organs with ovuliferous margins demonstrates that, in normal gynoecia, carpel growth is obviously not inhibited but concerted by meristematic fusion. d. Carpelloid Sepals A spontaneously occurring combination of gynoecium features in sepals (marginal placentation, stigmatic tissue, zones of dehiscence in the distal part, more or less complete marginal fusion) has been observed in several varieties of Brassica oleracea (var. capitata, gongylodes, italica; Detjen & McCue, 1933; Haskell, 1954; Sampson & MacArthur, 1959). A more or less total homeotic replacement of sepals with free carpelloid organs is also displayed by several A rabidopsis mutants (see Pruitt et al., 1987; Komaki et al., 1988; Bowman et al., 1989, 1991; Meyerowitz et al., 1989; Okada et al., 1989; Coen & Meyerowitz, 1991; Mizukami & Ma, 1992; Jack et al., 1992, 1994; Bowman, 1994; Clark & Meyerowitz, 1994; Sieburth et al., 1995; cf. R6bbelen, 1965). Again, transformation series are seen (in some cases due to temperature sensibility of the mutated alleles) that start with the development of stigmatic papillae at the tips of leafy sepals and continue with ovule initiation, the back of the transformed organ still bearing hairs typical of green leaves or sepals but lacking on carpels. The morphology of free, perfect "carpels" on sepal sites is identical with that ofcarpelloid stamens described above. Correspondingly, they have been used to confirm the nI theory. Okada et al. (1989) point out that, under the traditional 2n theory, a carpelloid sepal would consist of a sterile median carpel flanked by two halves of fertile carpels, which seems to be unnecessarily complicated. In the context of total homeosis a transformed sepal can be considered to have the value of a single carpel and not to be a combination of 89+ l + 89 organs. Indeterminate flowers have been observed in which the separated carpels of the first-order pistil, bearing more or less reduced ovules along their free margins, become the sepals of the second-order flower. The spontaneous occurrence of this phenomenon, often combined with virescenee, has been described several times (Godron, 1846, 1877, 1878: Cardaminepratensis L., Hesperis matronalis L.; Lignier, 1897: C. pratensis; Guyot and Gavaudan, 1961: Brassica napus, drabis caucasica Willd.). Again, comparable phenotypes are known in several artificial Arabidopsis mutants (Clark & Meyerowitz, 1994; Mizukami & Ma, 1995). The firstorder gynoecium commonly splits along the placentae into two carpels, which can also serve as proof of the nI theory. e. Carpelloid Ovules Observations on several Brassicaceae describe replacement of an ovule by reduced but distinct carpelloid or pistilloid structures, the neighboring ovules being more or less normally

258

THE BOTANICAL REVIEW

developed (de Candolle & de Candolle, 1841: Cheiranthus cheiri; Wydler, 1861: A lliaria petiolata; Masters, 1886: C. cheiri, Brassica sp.; Prain, 1898 tit. after Solms-Laubach, 1900: B. campestris; Diedicke, 1899: C. cheiri; Solms-Laubach, 1900: Capsella bursa-pastoris var. heegeri; Polowick and Sawhney, 1988: B. napus). The best-developed diminutive pistils displayed the regular organization ofa siliqua and even produced ovules. This homeotic replacement of a lower-order organ (ovule) by a higher-order organ (carpel, gynoecium) is termed "entropic homeosis" by Leavitt (1909). The diminutive pistils do not contribute much to carpel theories; however, one in Cheiranthus cheiri shows two stigmatic lobes over the valve backs (Masters, 1886), which may indicate a more primitive stigma than normal with lobes over the placental regions.

2. Increase in Gynoecial Components a. Additional Carpel Whorls The differentiation of more than one carpel whorl is a normal phenomenon in the flowers of the navel orange (Citrus sinensis [L.] Osbeck of the Rutaceae) and the pomegranate (Punica granatum L. of the Punicaceae). In other taxa it is counted among the teratologies, occurring occasionally in the families under discussion. Spontaneous cases as well as experimentally changed phenotypes of Arabidopsis mutants occasionally display strongly aberrant inflorescences as a whole---for example, with virescent and indeterminate flowers in which the axis produces distinct internodes between the phyllome whorls and extends beyond the carpel whorl. On this axial extension additional floral members, complete higher-order flowers, and secondary inflorescences may arise; lateral axes may then be subtended by carpels (see Schimper, 1829; Baillon, 1862-1863; Fermond, 1864; Celakovsk~,, 1884; Massalongo, 1908; Murty, 1953; Mizukami & Ma, 1995). If there is an additional carpel whorl in such aberrant flowers, its members are often separate (Eichler, 1865; Baccarini, 1918). More frequently, additional carpel whorls are formed within apparently normal, inconspicuous gynoecia (see Appendix 4). Because there is no external indication of their existence, they can be detected only by chance. In most cases they form a more or less perfect gynoecium diminished in size; more rarely, further carpel whorls are nested within them (see Gonnermann, 1979, 1980: papaveraceous Dicranostigma erectum with three whorls; Narayana, 1965: capparaceous Cadaba indica Lam. with four or five whorls). If additional carpel whorls are formed in normally bivalvate taxa, the first-order gynoecium is often composed of more than two elements, having three to four valves. Examples of this characteristic occurring spontaneously are specimens of Capparaceae (Cleome spinosa [Eichler, ! 865]) and Brassicaceae (Tropidocarpum eapparideum [Watson, 1882; Robinson, 1896], Lepidium villarsii Gren. & Godr. [Gerber, 1904b, 1904c, 1904d, 1905d], Capsella bursa-pastoris f. viguieri [Saunders, 1928a; Arber, 1931 a], Brassiea oleraeea [Yen, 1957], Sinapis alba [Fox Maule, 1970]); furthermore, it is shown by Arabidopsis, especially elavata mutants (Clark & Meyerowitz, 1994; Sieburth et al., 1995). Many cases of encaptic gynoecia have been described in Papaver somniferum, possibly due to the worldwide cultivation of the species for its capsules. For plants with an indeterminate gynoecium, Braun created the variety"endocephalum" (Godron, 1877).The carpels constituting the additional whorl are frequently more or less separate (see Schimper, 1829; Lankester, 1848, 1849; Scheffer, 1869; Wittmack, 1887) and look, for example, like totally carpelloid stamens in having a centripetally bent stigmatic lobe with marginal strips of papillate tissue and a lamina with (sub)marginal placentation. The second- and higher-order gy-

CARPEL NUMBER

259

noecia ofDicranostigma erectum and Cadaba indica (see above) also show split parts. Thus, organ initiation at the abnormally extended axis becomes increasingly irregular; however, the morphology of the free carpels supports the nl theory. b. Increase in Member Number within the Regular Carpel Whorl Increased numbers ofgynoecial members are easily noticed only in normally bivalvate gynoecia. The abundance of observations concerning these cases are so numerous that one can not include them all. Much more difficult to detect are the deviating gynoecia of Berberidaceae taxa that have more than one placenta. Appendix 5 summarizes the literature on the phenomenon in the families under discussion. In particular, reports of polyvalvate fruits of Brassicaceae have ranged from mere mention of their occurrence to detailed anatomical analysis and theoretical explanation. A certain number of aberrant gynoecia have developed through fasciation, as is the case for Eschscholzia crocea Benth. (Papaveraceae: Jepson, 1894), Brassica oleracea (Buchenau, 1871; Wittmack, 1886; Monnet, 1913), and Podophyllum peltatum (Berberidaceae: Trimble, 1882). Capsella bursa-pastoris f. viguieri is another taxon characterized by pronounced fasciation of the inflorescence (Blaringhem, 1910; Shull, 1929). Fasciated fruits may display a considerably increased valve number: in Brassica taxa, 14 (Buchenau, 1871), 10 (Wittmack, 1886), or 7 (Monnet, 1913); and in Capsella viguieri, 8 (Blaringhem, 1910). Godron (1864) described fruits ofBrassica oleracea with six valves that were again referred to by Chadefaud (1953) in a theoretical discussion of cruciferous floral construction. Fasciation may also have played a certain role in this case; on the other hand, the structure---two transverse valves, two pairs of median valves, also found in a Raphanus fruit (Schnizlein, 1843-1870)--resembles a perfect secondary pistil of staminal origin with the regular gynoecium aborted. (Moreover, a normal Raphanus fruit has extremely reduced valves. It is improbable that a mere increase in components would cause the differentiation of expanded valves.) Because both authors described fruits, there are no statements on the existence or lack of stamens in the flower. Eichler (1872) observed gynoecia with median valve pairs in Brassica napus and Cleome gynandra and considered this to be part of the trend toward organ doubling in the median plane, already manifest in the androecium (of. Endress, 1987). Theoretical interest has been ascribed to the (tri- or) tetravalvate forms of normally bivalvate gynoecia as well as Berberidaceae gynoecia with two placental regions. Tri- and tetravalvate gynoecia are often fully regular and symmetrical (Figs. 34.1, 35.1, & 35.4). In such Brassicaceae gynoecia, septation is either complete, with the septa uniting the three or four placental regions being Y- or X-shaped, respectively (Fig. 35.2), or incomplete, with short septa not touching at the center. Frequently, however, the additional valves in the median plane are much narrower than the transverse ones (Figs. 34.5 & 34.6), and their bases end considerably above the gynoecium base, the bases of the corresponding locules located at a higher level than the transverse ones (Fig. 34.2; see Eichler, 1865; Solms-Laubach, 1900; Puri, 1950; Murty, 1953). Also, in the Berberidaceae the additional loculus appears distinctly higher than the normal one (Chapman, 1936; Kaute, 1963). Acropetal analysis of serial sections of tetravalvate gynoecia reveals the following anatomical features (Fig. 36; of. Puri, 1950; Gonnermann, 1980): 1. In the gynoecium base, the dorsal veins of the transverse valves leave the stele. The remaining vascular tissue reorganizes to form two complexes in the median plane. They may or may not be more voluminous than normal placental bundle systems (Fig. 36A).

260

THE BOTANICAL REVIEW

Fig. 34. Increased number of members in the carpel whorl (1-4, Papaveraceae; 5-7, Fumariaceae). 1. Chelidonium majus, transection of young gynoecium showing three placental regions. 2-4. Dicranostigma ereetum. 2. Transection of the base ofa tetravalvate fruit; the bases of the two regular Iocules are visible, whereas the two additional ones, positioned in the median plane, end above the level of section. 3. Transection of an additional carpel in the median plane. 4. Narrow, fertile placental region of a tetravalvate fruit (note the diminished venation compared with the normal placental region of Fig. 19.2); zones of dehiscence are partly ruptured. 5. Dactylicapnos torulosa; young trimerous gynoecium in bud. 6. Corydalis lineariloba Siebold et Zucc., fruit transection showing three placental regions, the additional valve being conspicuously narrower. 7. C. flexuosa Franch., regularly trilobed stigma (arrows) topping a fruit with only two valves. Scale bars on the light micrographs = 1 mm (2, 3) and 0.1 mm (1,4), and on the stereomicrographs = 1 mm.

CARPEL NUMBER

261

Fig. 35. Increased number of members in the carpel whorl (1, Capparaceae; 2-4, Brassicaceae; 5 and 6, Rutaceae). 1. Polanisia dodecandra, dehisced trivalvate fruit (arrows point to the three replum tips). 2, Cheiranthus cheiri, transection oftrivalvate fruit with perfect septation. 3, 4. Capsella bursa-paswris, trivalvate, tetravalvate and bivalvate fruits of the same iofrutescence, all placentae fertile. 5, 6. Ptelea trifoliata. 5. Gynoecium with an additional carpel extending from the stigma to the middle of the ovary (recognizable by shortened wing, indicated by the arrow). 6. Gynoecium with two additional carpels, the wings of which end shortly below the middle of the ovary; arrows point to the wing bases of the two regular carpels. Scale bars = 1 ram.

THE BOTANICAL REVIEW

262

i I

t

E

D

(~

(3

O

9

00 QO (~(bO

O

I "*-- o

O ----~ |

oJUl

A Fig. 36. Schematic transection series of a teratologically tetravalvate gynoecium, starting from its base (A), in aeropetal direction. (The additional carpel pair may, depending on the case given, become apparent shortly above the gynoeeium base or much higher; thus, stages C-D cannot be ascribed to a certain constant level; of. the text.)

CARPEL NUMBER

263

2. The two transverse locules become discernible (Fig. 36B); in septless gynoecia, the placental regions become separated. They may be broader than normal. The placental bundle systems are already formed. 3. Within tissue of the placental regions, a cavity opens each (Fig. 36C), giving rise to the locules of the additional median valves. The placental regions thus split in half longitudinally. The vascular tissue of the placental bundle systems expands transversely. A small median portion becomes the dorsal vein of the additional valve, whereas the lateral portions, in the four"half-placentae," again reorganize in arrangements typical of placental regions (e.g., amphicribral bundles). All four "half-placentae" are fertile (Fig. 36D). If more than the normal amount of vascular tissue is present in the gynoecium base, the venation of the four secondary placental regions will not significantly differ from that in normally developed placental regions of a bivaivate gynoecium (see Murty, 1953). If there is, however, no basal increase of vascular tissue, splitting of the primary placental regions will result in a considerable decrease of venation supplying the secondary ones (Chodat & Lendner, 1897; Gonnermann, 1979, 1980). 4. The newly developed valves dilate acropetally, thus displacing the "half-placentae" from each other. Zones of dehiscence are more or less distinct along the flanks of all four secondary placental regions. The four placental bundle systems and the dorsal veins of the additional valves branch in a manner producing a valve-specific venation in the new structures (Fig. 36E). 5. Apically, all four valves end at the same level. The style is vascularized by portions of the four placental bundle systems and possibly the four valve dorsal veins. If the stigma is composed of distinct lobes, four of them are present. The surprisingly different conclusions that have been drawn from these facts will be discussed, first in the context of the n! theory: 1. Gynoecia with supernumerary valves contain (one or) two additional carpels, the result of a spontaneous meristic variation that occurs frequently in angiosperms. The insertion of the additional valves at a higher level caused Eichler (1865, 1872) to postulate a newly added carpel whorl that alternates with the normal one. Since the locules of the individual carpels arise at somewhat differing levels, Purl (1950) assumed a spiral arrangement of the carpels. The genetic background of increased carpel numbers has partly been elucidated. Shull (1929) demonstrated that bivalvate and tetravalvate forms of Capsella bursa-pastoris differ in but one gene determining this character. Also the clavata mutant ofArabidopsis thaliana displays supernumerary carpels (Pruitt et al., 1987; Okada et al., 1989; Bowman, 1994; Clark & Meyerowitz, 1994), the tetravalvate gynoecium being regular and fertile. The deviating phenotype is caused by a recessive mutation of a single gene positioned on chromosome I (Koornneef et al., 1983, cit. after Okada et al., 1989). The clavata mutants tend to produce fasciations (compare viguieri form of Capsella bursa-pastoris!). Variation in carpel number is not a specific feature for the families under discussion and has no theoretical value. 2. The occurrence of tri- and, especially, tetracarpellate gynoecia in the Brassicaceae is considered an atavism. It is thought to be derived from an ancestor with a tetramerous gynoecium consisting of either a four-membered whorl or two dimerous whorls. The median whorl that has become lost during evolution is thus supposed to reoccur spontaneously now and then, an interpretation widespread in the older literature. 3. The additional structures are not carpels and the aberrant gynoecia are bicarpellate. Raghavan (1939) and Raghavan and Venkatasubban (1941a) studied tri- and tetravalvate gynoecia of the Capparaceae taxa Cleome gynandra and Crateva religiosa. According to these authors, in these cases an incomplete fusion of the carpel margins causes a separation of the

264

THE BOTANICAL REVIEW

placental components; the space in between is then filled by parenchyma that imitates an additional valve. Each of the four placental regions of a tetravalvate gynoecium would thus belong to one carpel margin, being a mere half of a normal placental region. A similar idea had already been expressed by Wille (1886) regarding several tetravalvate siliculae in an infrutescence of Capsella bursa-pastoris. He postulated that, due to "exceptional pressure," each of the two parts forming a normal septum could transversely expand, resulting in splitting of the placental bundle system ("commissural vein"). The septum would thus double, and ovules could arise along both sides of the parts, which would lead to the formation of four ovular rows. Parenchymatous tissue would form a bridge between the septum halves, giving the impression of a valve but being an appendage of the carpel margins. ("Hierauf w~ichst das zwischen dem gespaltenen commissuralen Nerven liegende Stiick aus und ahmt die... Klappen der Karpelle nach, ware also als ein Blattzipfel aufzufassen" [Wille, 1886: 123].) If both septa were involved in this hypothetical doubling process, a tetravalvate gynoecium would be the result. The similarity of the vascular pattern in both "real" valves and expanded structures is, according to Wille, explained by the phyllomic nature of the latter, though they represent only parts ofphyllomes. The concept of diverging placentae or septa seems improbable at the first. A comparable ontogenetic process has, however, been revealed in normally developing bicarpellate taxa of Orobanche (Orobanchaceae: Hecht, 1990). The primordia of the parietal placentae of neighboring carpel margins are initiated in closest contact but, in the course of ontogeny of the gynoecial ridge, become more and more separated. Finally, four submarginal placentae are diagonally arranged. The fused marginal areas that have permanently overtopped the carpel tips form globular, commissural stigmatic lobes. The structure of such a regular gynoecium seems comparable to the Capparaceae gynoecium teratisms studied by Raghavan (1939) and Raghavan and Venkatasubban (1941a). Understandably, Tiagi and Sankhla (1963) apply the 2n theory to the Orobanche gynoecium. However, in the families under discussion, additional valves are more probably products of additional carpels. For example, in the papaveraceous Dicranostigma erectum these regions bear well-developed stigmatic lobes that are normally carpel tips. Although Purl (1950) and Murty (1953) criticized Raghavan's concept, more recently Ronse Decraene and Smets (1997b) refer to it to explain the tetramerous gynoecium of Capparis micracantha DC. According to them, this species demonstrates a doubled carpel number by splitting of the placentae, whereas in Capparis spinosa carpel number is increased through secondary addition of primordia to the original whorl (but see below). Additional valves are of great importance for some advocates of the 2n theory as well as for Saunders's support of the concept of carpel polymorphism. These structures are considered atavisms that reveal the reverse direction of the process of carpel "solidification." The reduced carpels--in extant taxa integrated more or less unrecognizably in the placental regions --would, in such occasional cases, show their true carpel nature. Saunders (1923: 465) stated emphatically that "there can be no doubt that in all these cases we are witnessing the reappearance of an ancestral character." The successive appearance of the locules proves, according to Eames (1961), a spiral carpel arrangement in Berbcridaceae. The number and behavior of the vascular bundles have painstakingly been analyzed to detect the fate of the bundles belonging to the exceptionally well developed carpels in the course of solidification (e.g., Chodat & Lendner, 1897; Chapman, 1936; Kaute, 1963; Gonnermann, 1979, 1980). Thus, knowingly or unknowingly, conclusions about the phylogeny of the vascular pattern in normal placental regions have been drawn that are founded on the concept of vascular conservatism (see VII.D.6). Among the 2n theoreticians there have also been different opinions about how to interpret additional valves:

CARPEL NUMBER

265

1. The occurrence of the "Tetrapoma type" (including all regular and symmetrical tetravalvate forms) is considered to corroborate the tetramerosity of the Brassicaceae gynoecia (e.g., Klein, 1894; Saunders, 1923, 1926). However, organization and vascular pattern of the tetravalvate "Tetrapoma" gynoecia are identical with those of the bivalvate "normal" gynoecia; there is only a doubling of components. As early as 1838, Bemhardi (1838a: 129, 131-132), an nI theoretician, felt the necessity to revise his idea of a tetravalvate prestage without septa; he wrote: Es liisst sich n~imlich nicht behaupten, dass die Frucht der Cruciferen urspriinglich eine vierklappige Kapsel ohne Scheidew~nde darstelle, indem mir seitdem zwei Arten aus dieser Familie bekannt geworden sind, welche anomalisch drei- und vierklappige Frilchte ansetzen, worin die Scheidew~inde nicht vermisst werden . . . . Nach diesen Beobachtungen l~isst sich daher nicht behaupten, dass die Scheidew~inde der Schoten der Cruciferen sich bloss auf Kosten zweier unvollkommen entwickelter Klappen bilden, indem man in der v611ig ausgebildeten Frucht ausser vier Kiappen auch Scheidewande findet . . . . Wenn aber die Scheidewand der Schote nicht auf Kosten zweier unausgebildeter Klappen entsteht, so fr~igt sich, wie man sich ilberhaupt die Entwicklung derselben sowohl in zwei- als mehrklappigen Friachten vorzustellen habe. If regular septa were produced by transformed valves, an atavism with only untransformed valves should have no septa--which is not, however, the case. Eichler (1872: 333) argues in a similar way; the 2n theory has to be rejected, "weil im Falle der Vierz/ihligkeit die alsdann vorhandenen 4 Placenten und 4 Scheidew~inde ganz denselben Bau besitzen, wie die entsprechenden Theile beim zweigliedrigen Pistill." Indeed, the "Tetrapoma type" cannot serve as a starting point for a gynoecium consisting of two valve carpels and two solid carpels but, as Celakovsk~' (1894: 84) rightly underlines, "nach jenem Fehlschluss [= 2n theory, C. B.] miasste das Pistill daselbst aus 8 Carpiden bestehen!"; that is, four valve carpels and four solid carpels had to be expected. However, Eichler (1872) and Eyde (1990) also were mistaken in believing that the identical organization of"Tetrapoma type" and normal bivalvate type sufficiently disproves the 2n theory. This mere fact cannot answer Lid6n's question (1992: 308): "But why could this anomalous species not be hetero-octocarpellate?" Gerber (1899c, 1905d), first maintaining 3n, later 2n carpels, never considered the "Tetrapoma type" to be atavistic but, instead, believed that it possesses double carpel numbers of either twelve or eight, respectively. 2. Finally, the present author did not ascribe importance to gynoecia with identical normal and additional valves, for there are bivalvate and polyvalvate taxa in both Papaveraceae and Capparaceae, even within the same genus (Capparis; Stylophorum of the Papaveraceae: two bivalvate species and S. diphyllum with three to five valves). Thus, to her, the Brassicaceae taxa producing tetravalvate gynoecia were mere variants in gynoecial merosity that may, in the future, become fixed at the higher number. Greater interest arose in the tetravalvate fruits ofDicranostigma erectum: in the fertile placental regions there was half as much vascular tissue as in those ofbivalvate fruits (Figs. 34.3 & 34.4; Gonnermann, 1979, 1980). Similar cases had already been described in the cruciferous Cheiranthus cheiri (Chodat & Lendner, 1897) and several Berberidaceae (Chapman, 1936; Eckardt, 1963; Kaute, 1963), all interpreted according to the 2n theory. These terata apparently showed that the vascular bundles of the more or less contracted additional carpels were integrated constituents of the complex placental bundle systems ofbivalvate taxa. It would therefore appear that in the (sometimes massive) placental regions of bivalvate gynoecia strongly contracted carpels would be included, thus rendering the gynoecia heterocarpellate; that is, consisting of valve carpels as well as solid

266

THE BOTANICAL REVIEW

carpels. Both carpel forms were, in the sense of Dickson (1935), considered modifications of the same type and not to be equated with Saunders's "valve" and "solid" carpels; moreover, they all were fertile. Thus, according to the former concept of the present author, there were two variants of tetravalvate gynoecia: A. The meristic variant (octo-heterocarpellate type; Fig. 37A): Four valve carpels plus four contracted carpels in the placental regions; each of the four placental regions identical with those ofa bivalvate morph concerning size, number and course of vascular bundles, and ovule number ("Tetrapoma type"). B. The atavistic variant (tetra-homocarpellate type, Fig. 37B): Two valve carpels plus two atavistically decontracted carpels (morphology therefore identical with that of valve carpels, valve forming, stigma bearing), primary placental regions hence expanded, each of the four secondary placental regions narrower than primary ones and with diminished vascular tissue and fewer rows of ovules. The same concept would also explain mono- and polyvalvate gynoecia. Extending her studies to further taxa and analyzing the background of the various carpel theories, however, the present author became uncertain whether there were really two qualitatively different variants. The following facts challenged the assumption: 1. There are series of gradual transition from extremely contracted additional carpels to expanded additional carpels, identical with the regular ones (see Raghavan, 1939; Puri, 1950; Murty, 1953). The amount and arrangement of the vascular tissue in the placental regions correspond to this behavior. The vascular pattern must not be considered a conservative feature. 2. Additional carpels are not specific for the families under consideration but, now and then, occur throughout the angiosperms without the general question of more than n carpels in these groups having been raised (except by Saunders). The following example illustrates such a case beyond the valvate taxa. Certain individuals ofPtelea trifoliata L. (Rutaceae), normally having a bicarpellate syncarpous gynoecium, often produce tri- to tetracarpellate fruits. Because the dorsal region of the discoid indehiscent fruit forms a broad wing, the additional wings are easily recognized. Perfectly developed additional carpels extend from the style to the receptacle; hence all fruit wings, regular as well as additional ones, have identical dimensions. However, shortened additional wings are frequent. Although they are usually nearly as wide as the regular ones, they extend from the style only to the base of the regular locules, to the locule middle, or, being minute, end above the top of the regular locules (Figs. 35.5 & 35.6); correspondingly, the accompanying locules are more or less diminished and the ovules aborted (Briaekner, 1991b). These additional carpels thus are inserted high above the site of insertion of the regular ones; following Eichler (1865, 1872), they would form a new upper whorl. Gynoecia consisting of more than one whorl are not uncommon in the family (Citrus); however this does not sufficiently explain the teratological Ptelea cases. Serial transections analyzed in acropetal direction strongly recall the teratological tetravalvate gynoeeia: Two locules, the regular ones, arise in the gynoecium base--much higher up, two additional ones are formed in the eommissural regions--the new locules enlarge, and the originally commissural regions grow out to form a wing typical for a carpel back--regular plus additional carpel pairs form the style and stigma. The morphology of these gynoecia should be interpreted by the same theory that fits the valvate gynoecia. The 2n theory should not be applied to Rutaeeae and all other taxa showing similar terata; another, more neutral, approach has to be taken. Following is an interpretation of additional carpels from the genetic background. Meristic variations within the floral whorls as well as all other malformations result from a more or less disturbed floral ontogeny. Although research on gene-directed meristem behavior has been progressing rapidly over the past several years, the statement "The development

CARPEL NUMBER

267

Fig. 37. Qualitatively different types of tetravalvate gynoecia as formerly postulated by the present author. A. Octo-heterocarpellate type (doubling of both solid and valve carpels = "Tetrapomatype"). B. Tetra-homocarpellate type (atavistic decontractionof the two solid carpels normally integrated in the placental regions).

of flowers is a mystery" (Meyerowitz et al., 1989: 209) remains accurate. However, the studies on mutants of Arabidopsis and other taxa have already accumulated essential knowledge. A sequential action of several genes and their (still widely unknown) products, respectively, is responsible for determination of the cells of the floral meristem in such a way that they may, in adequate spatial and temporal order, produce organ primordia of specific numbers and qualities. This determination takes place in concentric fields of the floral apex and begins long before any visible differentiation of the cells in the primordia. However, cell determination is not irreversible as long as the cells of the primordia have not started to differentiate (for detailed discussions, see Bowman et al., 1989; Meyerowitz et al., 1989; Bowman, ! 994). As has

268

THE BOTANICAL REVIEW

been shown, mutations of even single genes can cause serious deviations from normal flower development in terms of organ number and position. As they have a recessive character, the alleles concerned must be homozygous to affect the developmental process. Descriptions of phenotypes very similar to the experimentally generated homozygous Arabidopsis mutants have frequently been given in the literature on teratologies; homologous mutations may well be expected in these cases. To regard the floral characters transformed thereby as reversions to an ancestral condition is an entirely unfounded hypothesis. With reference to the "Tetrapoma type" Arber (1931b: 197-198) wrote: "Those who treat abnormalities as reversionary seem, as a rule, to do so simply on a priori grounds . . . . Now since numerical variation generally has a limited range on either side of the normal, it may well happen.., that among these variations some will coincide with those postulated for ancestral stages. But this is no justification for interpreting meristic anomalies as atavistic. Indeed,... they might quite as logically be called futuristic as atavistic." In wildflowers, most cases of single-allele mutations do not strikingly alter the phenotype; however, partial deficiencies in function may be expected that may cause a certain instability in character differentiation. So in individual plants, and often only in individual flowers, the number of carpels, though stable in the taxon, may "erroneously" be altered now and then. If three or four carpel primordia are initiated instead of two, a symmetrical three- to fourmembered gynoecium will consequently develop unless regulatory mechanisms immediately become active to quickly detect the "error" and return gynoecium ontogeny to the "normal program" of dimery. Although the existence of such regulatory mechanisms is only postulated here, it is not improbable (Okada et al. [1989: 36]: "It is considered that many genes are responsible for the regulation of pistil development and morphogenesis'). In dependence on the time of the regulatory mechanism becoming effective, the development of the additional carpels is inhibited sooner or later; if the wrong number is recognized immediately, they will remain rather underdeveloped and end high above the receptacle. Due to the basiplastic development of the gynoecium, its top (giving rise to style and stigma) will be formed of all primordia initiated; in its apical part the more or less well developed products of extra primordia are thus clearly visible (Fig. 34.7). Through developmental inhibition ofprimordia which are not part of the "normal program" of reduction, more or less solid carpels are formed, the growth of which stops before reaching the full length of the gynoecium. For functional reasons, they contain only a diminished amount of vascular tissue; moreover, their fertility may be affected. Transections of the base of the gynoecium (which differentiates last) show only two regular carpels; in their united placentae, however, some vascular tissue may be present joining the inhibited, higher-ending carpels with the stele. This interpretation, based on ontogeny, is able to explain gynoecia with differing carpel numbers not only in the taxa concerned but generally in angiosperms. It is in accord with the classical nl theory. The fact that, in the past, the 2n theory had readily been applied to supernumerary carpels is probably due to the common method of acropetal analysis and description of serial sections from receptacle to stigma; the same approach has deliberately been demonstrated here. Thereby the impression is given that additional carpels (of the same or a higher whorl) arise by expansion of the placental regions. The basiplastic differentiation of the gynoecium through intercalary growth is insufficiently reflected by such directed description (see VII.D.6); the fact that the apex is the oldest and first differentiated region of the pistil is overlooked this way. The difference in frequency of supernumerary carpels regarding the taxa concerned is striking. It is certainly due to a varying degree of genetic instability, possibly in combination with varying effective regulatory mechanisms. Whereas occasional deviations in carpel number are found in all angiosperms, increased numbers are a trend in Dieranostigma erec-

CARPEL NUMBER

269

turn, although there is stable dimery in the Chelidonieae, the tribe within which the genus has commonly been placed, and in the most closely related genus Glaueium as well as in the remaining members of Dicranostigma. However, tetramery has not yet been definitely fixed in the D. erectum genome. On the other hand, another taxon of otherwise dimerous Chelidonieae, Stylophorum diphyllum, regularly has carpel numbers higher than two (three to five), the gynoecia of the remaining species of the genus being dimerous throughout. Assuming a homologous genetic background, in S. diphyllum, unlike D. ereetum, the fixation of the meristic variation is already perfect. Similar examples are found in Capparaceae and Brassicaceae, although perhaps restricted to certain populations, such as Crateva adansonii and Cleome gynandra of Capparaceae and taxa ofRorippa, Draba, Brassica, and Tropidocarpum of Brassicaceae. The tetramerous gynoecium of Capparis micraeantha discussed by Ronse Decraene and Smets (1997b) can be compared to that of Stylophorum diphyllum, if the carpel number of four has been proved stable. The two median carpels, ending higher than the transverse members of the whorl, were once additional. The frequency ofmeristic variation in the gynoecia of Berberidaceae taxa has not been studied sufficiently, although identical genetic causes can be expected. Endress (1990, 1994a, 1995) postulates for the Magnoliidae a phylogenetic multiplication of carpel primordia, starting from monocarpellate forms, as one possible way of evolution leading to syncarpous gynoecia; this process may in some cases be more probable than "congenital fusion" of several, originally free, carpels. Thus, the bi- to polycarpellate Papaveraceae may have been derived from the monocarpellate berberidaceous alliance. Regarding the Papaveraeeae, Drinnan et al. (1994:100) state that the occurrence of"numerous 'carpels' (by increasing the number of placentae in a single ovary) is almost certainly secondary within the group." The concepts of increasing numbers of carpels or placentae are supported by earlier ontogenetic observations of subdividing carpel primordia in Monodora Dunal (Annonaceae: Leins & Erbar, 1982) and Kitaibelia Willd. (Malvaceae: Endress, 1981) that give rise to polycarpellate complexes. However, the interpretations of both cases have been criticized, the first by Deroin (1985, 1997), who suggests a more traditional explanation, andthe second by van Heel (1995), who questions the correctness of the observation. Be that as it may, an ontogenetic process of polymerization within the ovary has not been found in the taxa under discussion and is hence highly speculative. A possible way to paracarpous gynoecia may have led via a genetic fixation of meristic variants of monocarpellate gynoeci a in local populations without any need for an intermediate stage with axile placentation. In this concept, all polycarpellate Papaveraceae taxa would be advanced compared with the bicarpellate group. An increase in carpel number may have occurred independently several times (e.g., Stylophorum diphyllum fits well into Chelidonieae and should not be counted among the likewise polycarpellate Papavereae). Around the Capparales there are no monocarpellate taxa that could be considered possible relics of monocarpellate ancestors of the group. Without being too speculative, however, a tendency toward increase in carpel numbers may also be assumed. In particular, the polycarpellate gynoecia of several Capparoideae taxa could be considered more derived than the bicarpellate ones found in the family. This would then be the result of an evolutionary process counterdirected to the more common one of oligomerization.

VIII. Summarizing Discussion A broad comparison of the character set involved in structure and ontogeny of typical and teratological gynoecia permits the conclusion that, regarding the five families concerned, the existence of more than n carpels in gynoecia with n placental regions is neither provable nor

270

THE BOTANICAL REVIEW

necessary to interpret gynoecial organization. Theories that deviate from the nI theory are weakened by overemphasizing certain characters without adequately considering their range of variability (summed up in Table II). Considering the whole range of developmental processes known in gynoecium ontogeny, the nI theory is able to explain all aspects of the structure of mono- as well as bi- and polyvalvate gynoecia (and homologous indehiscent forms). The same will probably be true of paracarpous gynoecia in general. The 2n theory, now and then applied to a taxon, is as unsound in these cases as applied to the siliqua s.1. Within the scope of the nl theory, however, all hypotheses that are not well founded should be avoided. (For example, Puri's postulate [from 1945] that the "Rhoeadales" have an ancestor with axile placentation is based on the vascular pattern in the placental regions, thus referring to the now widely rejected concept of vascular conservatism.) Even if it may not be regarded as convincing that the nI theory is the right one to explain the structure of the discussed valvate gynoecia, it must be emphasized that one should apply the same theory to explain the gynoecia of the five families. This is especially directed to the authors of compiling works, floras, and general textbooks. Some examples of unequal application of the theories are: In Engler's "Syllabus der Pflanzenfamilien" (Melchior, 1964), Berberidaceae are accepted as pseudomonomerous (2n theory), whereas Papaverales and Brassicaceae are considered to have n carpels (nI theory). The same is found in "Sistema Magnoliofitov" (Takhtajan, 1987) and in Schmeil and Fitschen's "Flora of Germany" (Senghas & Seybold, 1993). In Rothmaler's "Exkursionsflora von Deutschland" (Schubert et al., 1994) the nI theory is used to explain the Papaverales; however, differing variants of the 2n theory are applied to Berberidaceae and Brassicaceae. In Strasburger's "Lehrbuch der Botanik" (Sitte et al., 1991), Ehrendorfer assumes pseudomonomery ofBerberidaceae gynoecia (but accepts true monomery in the 1998 edition), interprets Papaveraceae under the nI theory, and speculates on a certain probability that two sterile carpels and two fertile ones are present in the Brassicaceae gynoecium. G. Dahlgren (1987) expresses the same opinion' for Papaveraceae and Brassicaceae in her textbook "Systematische Botanik." The question arises as to whether the opinions of the proponents of the 2n and 3n theories have remained stable during their active period of scientific studies. The authors concerned may be classified into three groups: 1. Persons who dealt with the subject in only one paper (a doctoral dissertation, for example) and studied members of only one of the families; representatives are Klein (1894), Dickson (1935), Chapman (1936), Stoudt (1941), Bersillon (1955), Kuusk (1960), Kaute (1963), and Eigner (1973). Often their supervisors tended toward or maintained, respectively, the 2n theory themselves, Afterward the students changed the field of research, so their concept has not been subjected to critical reanalysis. 2. Authors who deal with preparation of textbooks or floras or who focus on other features of the families concerned; they adopt a concept without having an opportunity to check its degree of probability. Because this is a large group, it can be expected that the phrase "Brassicaceae have two fertile and two sterile carpels" or, at least, "carpels two (or four?)" will continue to appear in the descriptive literature, because it has been uncritically adopted again and again. 3. Authors who started like the members of group 1 but continued to study gynoecial morphology. Gained knowledge led to refinement of the concepts, which is reflected by a series of publications that, in some cases, show a gradual change in opinion and eventually culminate in final dropping of the theory. It should be noted that many supporters of the nl theory may be categorized into groups 1 and 2 in a similar manner. There are also many single papers, the arguments in which are too

CARPEL NUMBER

Fr',

E ~ ~

271

"r.

~•

L C

0 E C

0

;>

o

0

[...

0

r.-O C

I,)

m

~

0 i

Eoo

>

<

0 9- -

0~,-.

<

~" ~'~

o~

~

,

'~

"~ ~

~

,

'~

"c~ ~

~

t) ~

o

~o~

~m O,w

~

o~

p,

9

o

0

~

1

o0

o

~.~ N.-i

~.~

a o

>

~ ~ o..-

~ o

0

272

THE BOTANICAL REVIEW

poor to provide factual disproof of the 2n theory. Moreover, many authors adopted the nI theory only by intuition. Thus the camp of the nl theoreticians contains many adherents that are on the right side rather by chance. It is worthwhile to deal briefly with the work of some gynoecium specialists in group 3. On one hand are authors whose views did not change in the course of their scientific life. Wellknown representatives are Edith R. Saunders and Arthur J. Eames. Saunders published her first ideas on "carpel polymorphism in angiosperms," based on teratological Brassicaceae fruits, in 1923 at the age of fifty-five. Notwithstanding harsh criticism of the experts, she developed her concept steadily and painstakingly and apparently persisted in it until death in 1945 (see Schmid, 1977). Her outstanding work as a flower morphologist and geneticist was thus regrettably clouded, because the bizarre polymorphism concept projected a negative image onto her personality as a scientist. Eames uncompromisingly polemicized against the concept of carpel polymorphism; his own 2n theory, however, had likewise disputable weak spots (e.g., the postulate of the phylogenetic crossing of the ovules through the carpel wall from inside to outside). Although he admitted to that, he basically insisted on the 2n interpretation of the gynoecia in the families concerned that rested mainly on the assumption of vascular conservatism. On the other hand, several 2n and 3n theoreticians whose early fundamental works have been cited again and again by their followers to substantiate the correctness of that idea had their doubts about their own hypotheses in time. John Lindley, who actually initiated the debate on double carpel numbers in his 1828 work, wrote as early as 1830 (pp. 16-17): I am aware that it is possible to explain the peculiar economy of the replum of Cruciferae by that of Carmichaelia, and that the line of dehiscence in fruits is no evidence of the plan upon which it has been constructed. I also know that a less paradoxical way of understanding the structure of the siliqua, is to take two confluent carpella, each of which has a 2-lobed or 2-horned stigma, for the type of such a fruit;.., and moreover I have been shewn by Brown some instances of monstrous formation, which seem to confirm such an opinion. Nevertheless, I wish to record.., my view of the subject, whether it shall be ultimately found to be accurate or inaccurate, for the following reasons. In the first place, it will shew young botanists how narrowly it is necessary for them to observe the structure of the plants, and how indispensable it is to bear constantly in mind the analogies that exist between the formation of one plant and another; in the second place, by pursuing the discussion, I hope to induce someone to set the question at rest, by means of such demonstration as it is capable of receiving; and thirdly, I still retain my opinion, notwithstanding what I have seen and heard since it was formed; relying chiefly upon the peculiarities of Eschscholtzia, which seems.., so obviously formed upon the same plan as Cruciferae, whatever that plan may be, that what can be shewn to be true of one must be true of the other. Strictly speaking, Lindley summarized in these few sentences the whole line of argument against the 2n theory and revealed that only a little more evidence from comparative morphology of Brassicaceae and Papaveraceae would convince him of the general validity of the nl theory. But nearly 170 years were to pass before the present analysis removed--it is to be hoped!--the last doubts about the nI theory. Philippe van Tieghem, whose "anatomical method," the analysis of the floral vascular pattern, had caused a revolution in floral morphology, certainly supported the 2n theory regarding the families concerned here in his fundamental comparative works on pistil structure (1868, 1875), but even in the first edition of his "l~16ments de botanique" (1886/1888), as well as in the subsequent editions, he did not men-

CARPEL NUMBER

273

tion that concept but, instead, applied the nI theory. This fact seems to have been overlooked by many subsequent 2n theoreticians who refer to van Tieghem. Applying just that "anatomical method," Charles Gerber and Eduardo Martel developed rather simultaneously their variants of the 3n theory shortly before the beginning of the twentieth century. Additionally, Gerber studied teratological fruits having an increased amount of vascular tissue that at first seemed to confirm his assumptions. Whether Martel, after his 1902 publication, dealt further with the problem and possibly revised his extremely complex theory referring exclusively to vascular conservatism is not known to the present author. Gerber, however, while expanding his studies, began to question the correctness of his hypothesis. The first to describe the inverted bundles in the placental regions of the Brassicaceae, he found similar inversions in other plant organs and consequently dropped the postulate of inverted placental bundles as proof of addorsed carpels in the septa. So he joins the camp of the 2n theoreticians from 1904/1905. Heeding the warning by Lignier (1904) not to overestimate teratologies and realizing that three separate vascular bundles cannot be unambiguously assigned to each of his "carpels" (i.e., valves and placental regions), Gerber refrained from making any statements about carpel numbers in his later work (1907a, 1907b), which is rather strange behavior, considering his earlier publications. Gerber therefore must not be considered a persistent supporter of the 3n theory. Of the more recent authors, Vishvambhar Puri revised the 2n theory by Eames with the aim of avoiding a phylogenetic change of ovule position. Also, his new variant of the 2n theory rested on stability and validity of the vascular pattern. The finding of inverted bundles in the placentae of further taxa that were surely not furnished with solid carpels caused him to drop the 2n theory and to apply the nI theory from 1945 on. Peter Leins postulated 2n carpels in Brassicaceae but not in Papaveraceae according to ontogeny and course of vascular bundles in the gynoecia (Merxmiiller & Leins, 1966, 1967). Extensive studies of floral morphology and ontogeny, Capparaceae taxa included, caused him, however, to express noticeable doubts about the 2n theory (Leins & Metzenauer, 1979); he unambiguously belongs to the camp of the nI theoreticians (Leins, pers. comm.). The present author is also among those who first favored the 2n theory but, by comparative studies, obtained an increasing amount of counterarguments. Herewith she expresses her serious hope that she may be the last to have invested time and effort in refuting the 2n and 3n theories, at least for Berberidaceae, Papaverales, and Capparales. IX. Acknowledgments The author is greatly indebted to Prof. Dr. Peter Leins of Heidelberg, Germany, for enabling her to carry out ontogenetic studies in his laboratory, and to the members of his staff, especially to Mrs. Birgit Volz for manifold technical assistance. She is very grateful to Dr. Ursula Hofmann of GOttingen, Germany, for valuable suggestions on the manuscript. Further thanks go to Dr. Magnus Lid~n of Uppsala, Sweden, and Prof. Dr. Herwig Teppner of Graz, Austria, for providing material and literature, as well as to Dr. Elliot M. Meyerowitz of Pasadena, California, and Dr. Thierry Deroin of Paris, France, for literature. Sincere thanks are due to Prof. em. Dr. Rolf Sattler of Kingston, Ontario, for literature and encouraging discussions. Dr. Vladimir Choob of Moscow, Russia, kindly supplied a stylistic check of the Russian summary, and Mrs. Margitta Hielscher helped with typing parts of the manuscript. Special thanks to the editor of The Botanical Review, Dr. Dennis Wm. Stevenson, and to Lisa M. Campbell for putting the finishing touches on the text.

274

THE BOTANICAL REVIEW

X. Literature Cited Adachi, J., K. Kosuge, T. Denda & K. Watanabe. 1995. Phylogenetic relationships of the Berberidaceae based on partial sequences of the gapA gene. Pp. 351-353 in U. Jensen & J. W. Kadereit (eds.), Systematics and evolution of the Ranunculiflorae. P1. Syst. Evol., Suppl. 9. SpringerVerlag, Vienna, New York. Agardh, C. A. 1836. Ueber die Deutung der Bltithenteile und die Verwandtschaft der Balsaminen. Flora 19: 193-205,209-221. AI-Shehbaz, I. A. 1984. The tribes of Cruciferae (Brassicaceae) in the southeastern United States. J. Arnold Arbor. 65: 343-373. Alexander, I. 1952. Entwicklungsstudien an Bltiten von Cruciferen und Papaveraceen. Planta 14: 125-144. Aleykutty, K. M. & J. A. Inamdar. 1978. Structure, ontogeny and taxonomic significance oftrichomes and stomata in some Capparidaceae. Feddes Repert. 89: 19-30. AIlman, G. J. 1851. On the monstrosity of the common wallflower (Cheiranthus cheiri). Gard. Chron. 1851: 564. --- 9 1852. On the morphology of the fruit in the Cruciferae, as illustrated by a monstrosity in the wallflower. Rep. Brit. Assoc. Advancem. Sci. 21: Notices/Abstracts, 70. Alston, R. E. & B. L. Turner. 1963. Biochemical systematics. Prentice Hall, Englewood Cliffs, N.J. Alyavdina, A. A. 1931. Znachenie anatomii plodov i semyan dlya sistematiki semeistva Cruciferae. (The importance of fruit and seed anatomy for the systematics of the family Cruciferae.) Zhum. Russk. Bot. Obshch. 16: 85-100. (In Russian; summary in German.) Angiosperm Phylogeny Group9 1998. An ordinal classification for the families of flowering plants. Ann. Missouri Bot. Gard. 85: 531-553. Anuradha, S. M. J., M. Radhakrishnaiah & L. L. Narayana. 1988. Chemosystematics of Capparaceae. Feddes Repert. 99: 391-394. Appel, O. 1999. The so-called 'beak,' a character in the systematics of Brassicaceae? Bot. Jahrb. Syst. 121: 85-98. Arber, A. 1930. Studies in the Gramineae, IX: 1. The nodal plexus; 2. Amphivasal bundles. Ann. Bot. (London) 44: 593-620. 9 193 l a. Studies in floral morphology, I: On some structural features of the cruciferous flower. New Phytol. 30:11-41. 9 193 lb. Studies in floral morphology, II: On some normal and abnormal crucifers, with a discussion on teratology and atavism. New Phytol. 30: 172-203. 9 1931 c. Studies in floral morphology, III: On the Fumarioideae, with special reference to the androecium. New Phytol. 30:317-354. 9 1933. Floral anatomy and its morphological interpretation. New Phytol. 32: 231-242. 9 1938. Studies in flower structure, IV: On the gynoecium of Papaver and related genera. Ann. Bot. (London), n.s., 2: 649-664. Areangeli, G. 1894. Sopra alcuni casi teratologici osservati di recente. Boll. Soc. Bot. Ital. 1894: 305-308. Baeearini, P. 1918. Notule teratologiche. Nuovo Giorn. Bot. Ital., n.s., 25: 225-247. Bailey, L. H. & E. Z. Bailey. 1976. Hortus Third: A concise dictionary of plants cultivated in the United States and Canada. Revised and expanded by the staffofthe Liberty Hyde Bailey Hortorium. Macmillan, New York; Collier Macmillan, London. Baiilon, H. 1861-1862. Remarques sur l'organisation des Berb6rid6es. Adansonia 2: 268-291. - - . 1862-1863. Sur des fleurs monstrueuses de Sinapis arvensis. Adansonia 3: 351-353, plate 12. - - . 1872. Histoire des plantes. Vol. 3. Librairie Hachette, Paris. Barnsby, D. 1868. Note sur une an9malie pr6sent6e par le Raphanus caudatus. Bull. Soc. Bot. France 15: 151-154. Bateson, W. 1894. Materials for the study of variation. Macmillan, London. Baueh, R. 1940-1941. Das Sattelgelenk der Brassiceen-Frfichte, seine Abwandlung und seine systematische Bedeutung. Z. Bot. 37: 193-238.

CARPEL NUMBER

275

Beck, C, R. 1977. Anatomisch-embryologische Untersuchungen in den Chelidonioideae und ihre Beziehungen zu den iibrigen Papaveraceae. Diss., Humboldt University, Berlin. Becquerel, P. 1911. Apropos de la nouvelle esp~ce de Bourse-a-Pasteur, le Capsella viguieri Blaringhem. Bull, Soc. Bot. France 58: 376-378. Behnke, H.-D. 1977. Dilatierte ER-Zistemen, ein mikromorphologisches Merkmal der Capparales? Bet. Deutseh. Bot. Ges. 90: 241-251. - & W. Barthlott. 1983. New evidence from the ultrastructural and micromorphological fields in angiosperm classification. Nordic J. Bot. 3: 43-66. Beneeke, F. 1882. Zur Kenntnis des Diagramms der Papaveraceae und Rhoeadinae. Bot. Jahrb. Syst. 2: 373-390. Bennett, S. R. M., d. Alvarez, G. Bossinger & D. R. Smyth. 1995. Morphogenesis inpinoid mutants of Arabidopsis thaliana. P1. J. 8: 505-520. Benson, L. D. 1979. Plant classification. Ed. 2. D. C. Heath, Lexington, Mass. Bentham, G. 1860. Botanical memoranda. J. Linn. Soc., Bot. 5: 20-25. Reprinted in G. Bentham, 1970, Supplemental papers to Bentham & Hookers Genera Plantarum. J. Cramer, Lehre. Berg, R. Y. 1966. Seed dispersal of Dendromecon: Its ecology, evolutionary and taxonomic significance. Amer. J. Bot. 53: 61-73. 9 1969. Adaptation and evolution in Dicentra (Fumariaceae), with special reference to seed, fruit, and dispersal mechanism. Nytt. Mag. Bot. 16: 49-75. ~ . 1972. Dispersal ecology of Vancouveria (Berberidaceae). Amer. J. Bot. 59: 109-122. Bernhardi, d. 1833. Ueber den Charakter und die Verwandtschaft der Papaveraceae und Fumariaceae. Linnaea 8: 401-477. - - . 1838a. Ueber den Blfithen- und Fruehtbau der Cruciferen. Flora 21:129-138. - - . 1838b. Naehtrligliche Bemerkungen iiber Papaveraceae und Fumariaceae. Linnaea 12: 651668. - - , 1843. Ueber die Metamorphose der Pfianzen. Flora 26 (n.s. 1): 37-51, 53-67. Bersillon, G. 1955. Recherches sur les Papav6rac6es: Contribution ~ l'6tude des dicotyl6dons herbac6es. Ann. Sci. Nat., Bot., s6r. 11: 225-448. Blagowestsehenski, A. W. 1955. Die biochemischen Grundlagen des Evolutionsprozesses der Pflanzen. Akademie-Verlag, Berlin. Blaringhem, L. 1910. Les mutations de la Bourse/t Pasteur (Capsella heegeri Solms, C. viguieri n. sp. ). Bull. Sci. France Belgique, s6r. 7, 44: 275-307, plate VI. - & P. Viguier. 1910. Une nouvelle esp~ce de Bourse-~t-Pasteur: Capsella viguieri Blar., n~e par mutation. Compt. Rend. Hebd. S6ances Aead. Sci. 150: 988-991. Blattner, F. R. & J. W. Kadereit. 1995. Three intercontinental disjunctions in Papaveraceae subfamily Chelidonioideae: Evidence from chloroplast DNA. Pp. 147-157 in U. Jensen & J. W. Kadereit (eds.), Systematics and evolution of the Ranunculiflorae. P1. Syst. Evol., Suppl. 9. SpringerVerlag, Vienna, New York. Boeke, J. H. 1971. Location of the postgenital fusion in the gynoecium ofCapsella bursa-pastoris (L.). Med. Acta Bot. Neerl. 20: 570--576. Bokenfohr, S. M. & D. D. Cass. 1988. An example of an included gynoecium in Dicranostigma (Papaveraceae). Canad. J. Bot. 66: 2347-2351. Borb~is, V. v. 1879. Eine ungarische Crucifere mit vierP,ichriger Frucht. Oesterr. Bot. Z. 29: 246-247. Borekhausen, M. B. 1797. Linn6's Fumaria-Gattung. Pp. 43-47 in J. J. RSmer (ed.), Archiv f'tlr die Botanik. Vol. 1, pt. 2. Scbtiferische Buchhandlung, Leipzig. Bowman, J. L. (ed.). 1994. Arabidopsis: An atlas of morphology and development. Springer-Verlag, New York, Berlin. --., D. R. Smyth & E. M. Meyerowitz. 1989. Genes directing flower development in Arabidopsis. Plant Cell 1: 37-52. & .1991. Genetic interactions among floral homeotic genes ofArabidopsis. De"9 velopment 112: 1-20. Brassai, S. 1838. Botanische Miscellen aus Siebenbtlrgen. Flora 21:305--316. Braun, A. 1851. Betrachtungen ~lber die Erscheinung der VerjQngung in der Natur. Wilhelm Engelmann, Leipzig.

276

- - .

THE BOTANICAL REVIEW

1874. Ueber die Entwickelung der Placenten, im Anschlul3 an Franz Huisgen's Dissertation. Verh. Bot. Vereins Prov. Brandenburg 16: VIII. Sitzung vom 30. Januar 1874, 45-49. Bravais L. & A. Bravais. 1839. Essai sur la disposition gtntrale des feuilles rectistrites. Ann. Sci. Nat., Bot., str. 2, 12: 5--41, 65-77. Brett, J. F. & U. Posluszny. 1982. Floral development in Caulophyllum thalictroides (Berberidaceae). Canad. J. Bot. 60: 2133-2141. Brieehle-M~ek, M. 1993. BlOtenentwicklung und Anatomic von Gynoeceum und Frucht in den Gattungen Annona und Rollinia (Annonaceae). Abstract No. 5, p. 63 in D. F0mkranz & H. Schantl (eds.), Kurzfassungen XI Symposium Morphologie, Anatomie, Systematik, Salzburg 1993. Briquet, J. 1914. Sur l'organisation et les affinitts des Capparidactes ~ fruits vtsiculeux. Bot. Jahrb. Syst. 50 (Suppl., festschrift for A. Engler): 435-448. - - . 1915. La ddhiscence en Y dans la silique des Crucif~res. Compt. Rend. Soc. Phys. et Hist. Nat. G~neve 31 : 51-52. Brongniart, A. 1844. Examen de quelques cas de monstruositts vtgttales propres a 6clairer la structure du pistil et l'origine des ovules. Ann. Sci. Nat., Bot., str. 3, 2: 20-32. --.. 1861. Sur quelques cas de transformation des 6tamines en carpelles. Bull. Soc. Bot. France 8: 453-456. Brown, R. 1817. Observations on the natural family of plants called Compositae. Trans. Linn. Soc. London 12: 76-142. ---. 1818. On some remarkable deviations from the usual structure of seeds and fruits. Trans. Linn. Soc. London 12: 143-151. ---. 1821. An account of a new genus of plants, named Rafflesia. Trans. Linn. Soc. London 13: 201-234. 91826. Observations on the structure and affinities of the more remarkable plants collected by the late Walter Oudney, M.D., and Major Denham, and Captain Clapperton, in the years 1822, 1823, and 1824, during their expedition to explore Central Africa. Appendix, 208-246. Reprint: Pp. 259-303 in R. Brown (ed.), 1866, The miscellaneous botanical works. Vol. 1. Ray Society, Robert Hardwicke, London. Briiekner, C. 1982. Zur Kenntnis der Fruchtmorphologie der Papaveraceae Juss. s.str, und der Hypecoaceae (Prantl et K0ndig)Nak. Feddes Repert. 93: 153-212. - - . 1984. Zur Narbenform und zur karpomorphologischen Stellung der Fumariaceae DC. in den Papaverales. Gleditschia 11: 5-16. - - . 1985. Frucht- und Samenanatomie von Pteridophyllum racemosum Sieb. et Zucc. und die Position der monotypischen Gattung in den Papaverales. Feddes Repert. 96:199-213. - - . 1991a. Zur Interpretation des Karpells--Eine Obersicht. Gleditschia 19:3-14. - - . 1991b. Fruchtanatomische Studien an Dictamnus albus L., Zanthoxylum simulans Hance, Ptelea trifoliata L. und Ruta graveolens L. (Rutaceae). Feddes Repert. 102: 541-570. - - . 1992a. Gynoecium morphology and fruit anatomy in Pseudofumaria Medik. (Fumariaceae), with a discussion of carpellary composition. Bot. Jahrb. Syst. 114: 251-274. - - . 1992b. Gynoecium ontogeny and carpology in Corydalis DC. section Cheilanthifoliae Lidtn (Fumariaceae). Flora 187:299-316. - - . 1993. Comparative carpology in the tuber-bearing sections and sect: Capnogorium of Corydalis DC. (Fumariaceae). Bot. Jahrb. Syst. 115: 367-420. - - . 1996a. Carpelloid stamens in Papaveraceae Juss. and Brassicaceae Bumett (Cruciferae Juss.) and their bearing on theories of gynoecium organization. Feddes Repert. 107:321-337. --.. 1996b. Zur Kl~trung der Karpellzahl in Papaverales, Capparales und Berberidaceae. Postdoctoral thesis (Habilitationsschrift), Humboldt University, Berlin. Buchenau, F. 1866. Bemerkungen fiber den Blt~tenbau der Fumariaceen und Cruciferen. Flora 49: 39-46. --.. 1871. lnteressantere Bildungsabweichungen. Abh. Naturwiss. Ver. Bremen 2: 469-480, plates IV, V. Buchet, S. 1911. Apropos du Capsella viguieri Blaringhem. Bull. Soc. Bot. France 58: 378-380. Bugnon, P. 1926. Remarques sur la fleur des Eschscholtzia (Papavtracte). Bull. Soc. Bot. France 73: 970-974.

CARPEL NUMBER

277

Bush, N. A. 1904-1910. Rhoeadales i Sarraceniales flory Kavkaza. (Rhoeadales and Sarraceniales in the flora of the Caucasus.) K. Mattisen, lurev. ---. 1959. Sistematika vysshich rastenii. (Systematics of higher plants.) Ed. 3. Uchebno-pedagogicheskoe izdatel'stvo ministerstva prosveshcheniya R.S.F.S.R., Moscow. (In Russian.) Calestani, V. 1917. Evoluzione e classificazione delle Crocifere. Nuovo Giorn. Bot. Ital., n.s., Mem. Soc. Bot. Ital. 24: 245-290. Camus, J. 1888. Alcune nuove osservazioni teratologiche sulla flora del Modenese. Atti Soc. Nat. Modena, Mem., ser. 3, 7: 212-216. Capus, G. 1878. Anatomie du tissu conducteur. Ann. Sci. Nat., Bot., srr. 6, 7: 209-291, plates 1824. Carano, E. 1911. Sulla struttura di stami anomali nel Papaver rhoeas L. Ann. Bot. (Rome) 9: 389-392. Carlquist, S. 1969. Toward acceptable evolutionary interpretations of floral anatomy. Phytomorphology 19: 332-362. Cass, D. D. & G. C. Fabi. 1990. Early ovule development in Papaver rhoeas. Canad. J. Bot. 68: 258-265. Castroviejo, S., M9 Lainz, G9 L6pez Gonz~lez, P. Montserrat, F. Mufioz Garmendia, d. Paiva & L. Villar. 1986. Flora Iberica. Vol. 1. Real Jardin Bot,'inico, C.S.I.C., Madrid. Cejp, K. 1925. Beitrag zur vergleichenden Morphologie der dimerischen Bliiten. Beih. Bot. Centralbl. 41, sect. 1: 128-164. Celakovsk~,, L. J. 1875. Vergrfinungsgeschichte der Eichen von Alliaria officinalis Andrz. Bot. Zeitung 33: 129-138, 145-152, 161-171,177-182. ~-. 1876. Vergleichende Darstellung der Placenten in den Fruchtknoten der Phanerogamen. Abh. Krnigl. Brhm. Ges. Wiss. Math.-Naturwiss. CI., ser. VI, 8: 1-74. 9 1878. Teratologische Beitr/ige zur morphologischen Deutung des StaubgePasses. Jahrb. Wiss. Bot. 11: 124-174, plates V-VII. ~-. 1880. Uber vergrilnte Bltithen einer Hesperis matronalis. Sitzungsber. Krnigl. Brhm. Ges. Wiss. Prag 1879: 88-92. 9 1884. Neue Beitrlige zur Foliolartheorie des Ovulums. I. Anamorphosen der Eichen yon Raphanus sativus L. Abh. Krnigl. Brhm. Ges. Wiss., Math.-Naturwiss. CI., ser. VI, 12: 1-20, plate 1. 9 1894. Das Reductionsgesetz der Bl~ithen, das Drdoublement und die Obdiplostemonie, ein Beitrag zur Morphologie der Bliithen. Sitzungsber. Krnigl. Brhm. Ges. Wiss., Math.-Naturwiss. CI, ser. III: 1-140, plates 1-5. - - . 1899. Epilog zu meiner Schrift ~iberdie Placenten der Angiospermen nebst einer Theorie des antithetischen Generationswechsels der Pflanzen. Sitzungsber. Krnigl. Brhm. Ges. Wiss., Math.Naturwiss. CI., ser. III: 1-35. - - . 1902. l]ber die inversen Placentarb0ndel der Cruciferen. Oesterr. Bot. Z. 52: 89-92. Chadefaud, M. 1953. Anomalies et constitution florale chez les Crucifrres. Bull. Soc. Bot. France 100: 55-59. - - . 1954. Anatomie florale et pirces triples chez les Crucifrres. Bull. Soc. Bot. France 101: 217-222. 9 1956. La fleur et les pirces florales des Crucifrres d'aprrs quelques structures trratologiques. Bull. Soc. Bot. France 103: 454-460. Chapman, M. 1936. Carpel anatomy of the Berberidaceae. Amer. J. Bot. 23: 340-348. Chase, M. W., D. E. Soltis, R. G. Olmstead, D. Morgan, D. H. Les, B. D. Mishler, M. R. Duvall, R. A. Price, H. G. Hills, Y.-L. Qiu, K. A. Kron, J. H. Rettig, E. Conti, J. D. Palmer, J. R. Manhart, K. J. Sytsma, H. J. Michaels, W. J. Kress, K. G. Karol, W. O. Clark, M. Hedrrn, B. S. Gaut, R. K. Jansen, K.-J. Kim, C. F. Wimpee, J. F. Smith, G. F. Furnier, S. H. Strauss, Q.-Y. Xiang, G. M. PlunkeR, P. S. Soltis, S. M. Swensen, S. E. Williams, P. A. Gadek, C. J. Quinn, L. E. Eguiarte, E. Golenberg, G. H. Learn Jr., S. W. Graham, S. C. H. Barrett, S. Oayanandan & V. A. Albert9 1993. Phylogenetics of seed plants: An analysis of nucleotide sequences from the plastid gene rbcL. Ann. Missouri Bot. Gard. 80: 528-580. Chatin, A. 1861. Sur l'androcre des Crucifrres. Bull. Soc. Bot. France 8: 370-374, 471--475. Chlttenden, F. J. 1914. The rogue wallflower. J. Bot. 52: 265-269. Chodat, R. 1888. Neue Beitr,'ige zum Diagramm der Cruciferenbltlthe. Flora 71 : 145-149.

278

THE BOTANICAL REVIEW

& A. Lendner. 1897. Remarques sur le diagramme des Crucif~res. Bull. Herb. Boissier 5: 925-938. Citerne, P.-E. 1892. Berb6rid6es et l~rythrosperm6es. Th~se, Pads. Clark, S. E. & E. M. Meyerowitz. 1994. Arabidopsis flower development. Pp. 435--466 in E. M. Meyerowitz & C. R. Somerville (eds.), Arabidopsis. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York. CIos, D. 1862. Deuxic~me fascicule d'observations t6ratologiques. M6m. Acad. Sci. Toulouse, s6r. 5, 4: 51-70. ---. 1865. Observations sur le pistil ou le fruit des genres Papaver et Citrus. Ann. Sci. Nat., Bot., s6r. 5, 3: 312-320. Coen, E. S. & E. M. Meyerowitz. 1991. The war of the whorls: Genetic interactions controlling flower development. Nature 353:31-37. Corner, E. J. It. 1958. Transference of function. J. Linn. Sot., Bot. 56: 33-40. ---. 1963. A criticism of the gonophyll theory of the flower. Phytomorphology 13: 290-292. ---. 1976. The seeds of dicotyledons. Vols. 1,2. Cambridge University Press, Cambridge, London. Cramer, C. 1879. In Botanische Sektion, Sitzung den 12. August 1878. Verh. Schweiz. Naturforsch. Ges. Bern, 61. Jahresversamml., Jahresber. 1877/1878:106-108. Crane, P. R. 1985. Phylogenetic analysis of seed plants and the origin of angiosperms. Ann. Missouri Bot. Gard. 72: 716-793. 9 1986. The morphology and relationships of Bennetitales. Pp. 163-175 in R. A. Spicer & B. A. Thomas (eds.), Systematic and taxonomic approaches in paleobotany. Systematics Association Special Vol. 31. Clarendon Press, Oxford. Crepin, F. 1866. Recueil de faits t6ratologiques. Suite (1). Bull. Soc. Roy. Bot. Belgique 5:108-112. Cresens, E. M. & E. F. Smets. 1989. The carpel--A problem child of floral morphology and evolution. Bull. Jar& Bot. Belg. 59: 377-409. - &- - . 1992. On the character "carpel-form": Trends in the development of the Magnoliatae pistil. Candollea 47: 373-390. Croizat, L. 1960. Principia Botanica. Vols. 1, 2. Published by the author, Caracas. - - . 1964. Thoughts on high systematics, phylogeny and floral morphogeny, with a note on the origin of the Angiospermae. Candollea 19:17-96. Cronqulst, A. 198 I. An integrated system of classification of flowering plants. Columbia University Press, New York. - - . 1988. The evolution and classification of flowering plants. Ed. 2. The New York Botanical Garden, New York. Cuboni, G. 1889. Le forme teratologiche nei fiori di Diplotaxis erucoides DC. e loro causa. Nuovo Giom. Bot. Ital. 21: 507-511. D'Arey, W. G. 1979. Flora of Panama, Family 73A: Capparaceae-Tovarioideae. Ann. Missouri Bot. Gard. 66:117-121. Dahl, ]~. E. 1989. Taxonomic and morphological studies in Hypecoum sect. Hypecoum (Papaveraceae). PI. Syst. Evol. 163: 227-280. - - . 1990. Infrageneric division of the genus Hypecoum (Papaveraceae). Nordic J. Bot. 10:129-140. Dahigren, G. (ed.). 1987. Systematische Botanik. Springer-Verlag, Berlin, Heidelberg. Dahlgren, R. M. T. 1977. A commentary on a diagrammatic presentation of the angiosperms in relation to the distribution of character states. Pp. 253-283 in K. Kubitzki (ed.), Flowering plants: Evolution and classification of higher categories. PI. Syst. Evol., Suppl. 1. Springer-Verlag, Vienna, New York. - - . 1980. A revised system of classification of the angiosperms. Bot. J. Linn. Soc. 80: 91-124. - - . 1983. General aspects of angiosperm evolution and macrosystematics. Nordic J. Bot. 3: 119-149. Davis, P. H. (ed.). 1965. Flora of Turkey and the East Aegean Islands. Vol. 1. University Press, Edinburgh. De Candolle, A. P. 1813. Th6orie 616mentaire de la Botanique, ou exposition des principes de la classification naturelle et de l'art de d6crire et d'~tudier les v6g6taux. Deterville, Paris. - - . 1821a. M6moire sur la famille des Crucif/~res. M6m. Mus. Hist. Nat. 7: 169-252.

-

-

CARPEL NUMBER

279

1821b [1819, volume dated 1821]. M~moire sur les affinit~s naturelles de la famille des Nymphaeac~es. M6m. Soc. Phys. Hist. Nat. Gen~ve 1: 209-244. ~.. 1824. Prodromus systematis naturalis regni vegetabilis. Vol 1. Treuttel et Wflrtz, Paris. --.. 1827. Organographie v6g~tale. Vol, 2. Deterville, Pads. ~.. 1833. Pflanzen-Physiologie. Translated by J. ROper. Vol. 1. J. G. Cotta'sche Buchhandlung, Stuttgart, TQbingen. & A. de Candolle. 1841. Monstruosit(~s v6g6tales, premier fascicule. Neue Denkschr. Allg. Schweiz. Ges. Naturwiss. 5: 1-25, plates 1-7. De Maggie, A. E. & C. L. Wilson. 1986. Floral structure and organogenesis in Podophyllumpeltatum (Berbaddaceae). Amer. J. Bet. 73: 21-32. Delplno, F. 1894. Eterocarpia ed eteromericarpia helle Angiosperme. Gamberini e Parmeggiani, Bologna. Deroin, T. 1985. Contribution ~ la morphologie compar6e du gyn6c6e des Annonaceae-Monodoroideae. Bull. Mus. Nat. Hist. Nat., Paris, s6r. 4, 7, sect. B, Adansonia: 167-176. ~-. 1997. Confirmation and origin of the paracarpy in Annonaceae, with comments on some morphological aspects. Candollea 52: 45-58. Detjen, L. R. & C. A. McCue. 1933. Cabbage characters and their heredity. Univ. Delaware Agric. Exp. Sta. Bull. 180: 1-149. Dickie, G. 1867. Notice of some vegetable monstrosities. J. Linn. See., Bet. 9: 182-184. Diekson, J. 1935. Studies in floral anatomy, II: The floral anatomy of Glauciumflavum with reference to other members of the Papaveraceae. J. Linn. See., Bet. 50: 175-224. Diedieke, H. 1899. In Bericht fiber die Herbst-Hauptversammlung in Weimar am 8. Oktobcr 1899. Mitth. Thllring. Bet. Vereins, N. F. 13/14: 121. Doell, I. C. 1843. Rheinische Flora. (Author's report.) Flora 26: 598-602. Dormer, K. J. 1954. The acacian type of vascular system and some of its derivaties, I: Introduction, Menispermaceae and Lardizabalaceae, Berberidaceae. New Phytol. 53:301-311. Doweld, A. B. 1996. On the origin of the carpel as evidenced by its vascular skeleton. Phytomorphology 46: 387-394. Doyle, J. A. 1978. Origin of angiosperms. Ann. Rev. Ecol. Syst. 9: 365-392. & M. J. Donoghue. 1986. Seed plant phylogeny and the origin of angiosperms: An experimental cladistic approach. Bet. Rev. (Lancaster) 52: 321-431. - & .1987. The origin of angiosperms: A cladistic approach. Pp. 17-49 in E. M. Friis, W. (3. Chaloner & P. R. Crane (eds.), The origins of angiosperms and their biological consequences. Cambridge University Press, Cambridge, England. Drinnan, A. N., P. R Crane & S. B. Hoot. 1994. Patterns of floral evolution in the early diversification of non-magnoliid dicotyledons (eudicots). Pp. 93-122 in P. K. Endress & E. M. Friis (eds.), Early evolution of flowers. PI. Syst. Evol., Suppl. 8. Springer-Verlag, Vienna, New York. Du Petit-Thouars, A. 1820. Notice sur une fleur de pavot oriental, dent toutes les ~tamines 6taient chang~es r pistil. Bull. Sci. Soc. Philom. Paris 1820: 187-190. Duehartre, P. 1867. l~l~ments de botanique, comprenant l'anatomie, l'organographie, la physiologic des plantes, les familles naturelles et la g6ographie botanique. J. B. Bailli/,'re et ills, Paris. ~ . 1870-1871. Note sur une monstruosit6 de la fleur du violier (Cheiranthus cheiri). Ann. Sci. Nat., Bet., s6r. 5, 13: 315-339, plate 1. Dunal, M. F. 1817. Monographie de la famille des Anonac6es. Treuttel et Wiirtz, Paris, London. Dopuy, P. & M. Gu6d~s. 1964. Transformation de l'ovule des Angiospermes au cours des virescences: Comparaison avec diff6rentes formes foliaires normales ou t~ratologiques. Compt. Rend. Hebd. S6ances Acad. Sci. 258: 1616-1619. Duthie, J. F. 1882. A fruit within a fruit. Gard. Chron., n.s., 17: 601. Eames, A. J. 1931. The vascular anatomy of the flower with refutation of the theory of carpel polymorphism. Amer. J. Bet. 18: 147-188. ~ . 1961. Morphology of the angiosperms. McGraw-Hill, New York, Toronto. & C. L. Wilson. 1928. Carpel morphology in the Cruciferae. Amer. J. Bet. 15: 251-270. & .1930. Crucifer carpels. Amer. J. Bet. 17: 638-656. Eekardt, T. 1937. Untersuchungen fiber Morphologie, Entwicklungsgeschichte und systematische Bedeutung des pseudomonomeren Gynoeceums. Nova Acta Leop., N. F. 5:1-112. ~..

280

--..

THE BOTANICAL REVIEW

1963. Zum Blfitenbau der Angiospermen im Zusammenhang mit ihrer Systematik. Ber. Deutsch. Bot. Ges. 76: [38]-[49]. Eggers, O. 1935. Ober die morphologische Bedeutung des Leitbfindelverlaufes in den Blfiten der Rhoeadalen und fiber das Diagramm der Cruciferae und Capparida'ceen. Planta 24: 14-58. Eiehholz, G. 1886. Untersuchungen fiber den Mechanismus einiger zur Verbreitung von Samen und Friichten dienender Bewegungserscheinungen. Jahrb. Wiss. Bot. 17: 543-590. Eiehler, A. W. 1865.0ber den Blfithenban der Fumariaceen, Cruciferen, und einiger Capparideen. Flora 48: 433-445, 449-460, 497-508, 513-521,529-536, 545-559. --.. 1872. Abermals einige Bemerkungen fiber die Cruciferenblfithe. Flora 55: 328-334. --.. 1875, 1878. Blfithendiagramme. 2 vols. Wilhelm Engelmann, Leipzig. Eigner, J. 1973. Zur Stempel- und Fruchtentwicklung ausgewahlter Brassieaeeae (Cruciferae) unter neueren Gesichtspunkten der Blfitenmorphologie und der Systematik. Beitr. Biol. Pflanzen 49: 359-427. Emberger, L. 1960. Les v6g6taux vasculaires. In M. Chadefaud & L. Emberger, Trait6 de botanique syst~matique. Vol. 2, pts. 1, 2. Masson, Paris. Eadress, P. IC 1972. Zur vergleichenden Entwicklungsmorphologie, Embryologie und Systematik bei Laurales. Bot. Jahrb. Syst. 92: 331-428. --.. 1977. Blfitenmorphologie---Rfickblick und aktuelle Probleme. Ber. Deutsch. Bot. Ges. 90: 1-13. --.. 1981. Gynoecia with ontogenetical carpel multiplication and their evolutionary significance. Abstracts XIII Int. Bot. Congr. Sydney: 137. --.. 1987. Floral phyllotaxis and floral evolution. Bot. Jahrb. Syst. 108: 417-438. --.. 1989. Chaotic floral phyllotaxis and reduced perianth in Achlys (Berberidaceae). Bot. Aeta 102: 159-163. ---. 1990. Evolution of reproductive structures and functions in primitive angiosperms (Magnoliidae). Mem. New York Bot. Gard. 55: 5-34. --.. 1992. Evolution and floral diversity: The phylogenetic surroundings of Arabidopsis and Antirrhinum. Int. J. Plant Sci. 153: S106-S122. ---. 1994a. Diversity and evolutionary biology of tropical flowers. Cambridge Tropical Biology Seties. Cambridge University Press, Cambridge, England. ---. 1994b. Floral structure and evolution of primitive angiosperms: Recent advances. PI. Syst. Evol. 192: 79-97. ---. 1995. Floral structure and evolution in Ranunculanae. Pp. 47-61 in U. Jensen & J. W. Kadereit (eds.), Systematies and evolution of the Ranunculiflorae. PI. Syst. Evol., Suppl. 9. SpringerVerlag, Vienna, New York. ---. 1996. Evolutionary aspects of fruits in basal flowering plants. Norske Vidensk.-Akad., Mat.Naturvidensk. KI., Avh., n.s., 18: 21-32. Engelmann, G. 1832. De antholysi prodromus. Dissertatio inauguralis phytomorphologica. Frankfurt am Main. Engler, A. 1872. Uber monstr6se Blfithen von Barbarea vulgaris R. Br., ein Beitrag zur Bes~tigung des D6doublements in der Crueiferenblilthe. Flora 55: 449--456. Erbar, C. & P. Leins. 1997. Studies on the early floral development in Cleomoideae (Capparaceae) with emphasis on the androecial development. PI. Syst. Evol. 206:119-132. Erdtman, G. 1952. Pollen morphology and plant taxonomy: Angiosperms. Almqvist & Wiksell, Stockholm. Ernst, W. R. 1961. On the family status of the Fumariaeeae (Abstract). Amer. J. Bot. 48: 54. --.. 1962a. A comparative morphology of the Papaveraeeae. Ph.D. diss., Stanford University. ---. 1962b. The genera of Papaveraceae and Fumariaceae in the southeastern United States. J. Arnold Arbor. 43:315-343. ---. 1963. The genera of Capparaeeae and Moringaeeae in the southeastern United States. J. Arnold Arbor. 44: 81-95. ---. 1964. The genera of Berberidaceae, Lardizabalaceae, and Menispermaceae in the southeastern United States. J. Arnold Arbor. 45: 1-35. ---. 1967. Floral morphology and systematics of Platystemon and its allies Hesperomecon and Meconella (Papaveraceae: Platystemonoideae). Univ. Kansas Sci. Bull. 47: 25-70.

CARPEL NUMBER

281

Everett, T. H, 1981. The New York Botanical Garden illustrated encyclopedia of horticulture. Vol. 3. Garland Publishing, New York, London. Eyde, R. H. 1975. The bases of angiosperrn phylogeny: Floral anatomy. Ann. Missouri Bot. Gard. 62: 521-537. ~-. 1990. Four-merous gynoecia, real and unreal, in Cruciferae. Amer. J. Bot. 77 (Suppl.): 131. Ezelarab, G. E. & K. J. Dormer. 1963. The organization of the primary vascular system in Ranunculaceae. Ann. Bot. (London), n.s., 27: 23-38. & .1966. The organization of the primary vascular system in the Rhoeadales. Ann. Bot. (London), n.s., 30: 123-132. Fagerlind, F. 1946. Strobilus und Bltlte yon Gnetum und die MOglichkeit, aus ihrer Struktur den BRRenbau der Angiospermen zu deuten. Ark. Bot. 33A, no. 8: 1-57. ~ . 1958. Is the gynoecium of the angiosperms built up in accordance with the phyllosporous or the stachyosporous scheme? Svensk Bot. Tidskr. 52: 421-425. Fahn, A. & M. Zohary. 1955. On the pericarpial structure of the legumen, its evolution and relation to dehiscence. Phytomorphology 5:99-111. Fanta, A. 1894. Die Unregelmassigkeiten der Samenkapsel beim Gartenmohn. Bot. Centralbl. 57: 39. Farr, E. R., J. A. Leussink & F. A. Stafleu (eds.). 1979. Index nominum genedcorum. Vols. 1-3. Bohn, Scheltema & Holkema, Utrecht; W. Junk, The Hague. Fedde, F. 1909. Papaveraceae-Hypecoideae et Papaveraceae-Papaveroideae. In A. Engler (ed.), Das Pflanzenreich IV, 104. Wilhelm Engelmann, Leipzig. 9 1936. Papaveraceae. Pp. 5-145 in A. Engler & H. Harms (eds.), Die nattirlichen Pflanzenfamilien. Ed. 2. Vol. 17b. Wilhelm Engelmann, Leipzig. Feng, M. & A.-M. Lu. 1998. Floral organogenesis and its systematic significance in the genus Nandina (Berberidaceae). Acta Bot. Sin. 40: 102-108, plates I-III. Fermond, C. 1855. Recherehes sur le nombre des parties de la fleur des dicotyl~dones. Bull. Soc. Bot. France 2: 466-471. - - . 1864. Essai de phytomorphie. Vol. 1. Libraide Mtdicale Germer Bailli~re, Pads. Fisher, J. B. 1971. Inverted vascular bundles in the leaf of Cladium (Cyperaceae). Bot. J. Linn. Soc. 64: 277-293. Fisher, J. E. 1972. The transformation of stamens to ovaries and of ovaries to inflorescences in Triticum aestivum L. under short-day treatment. Bot. Gaz. 133: 78-85. Fleischer, ?. 1862. Ueber Missbildungen verschiedener Culturpflanzen und einiger anderer landwirthschaftlichen Gew[tchse. Conrad Weychardt, Esslingen. Flora of North America Editorial Committee (eds.). 1997. Flora of North America north of Mexico. Vol. 3. Oxford University Press, New York, Oxford. Fournier, E. 1856a. Note sur un d~veloppement anormal de I'androcte dans le Cheiranthus cheiri L. Bull. Soc. Bot. France 3: 352-354. ~ . 1856b. Note sur la transformation de l'androcte en ovaire dans le Cheiranthus cheiri L., examinte au point de vue morphologique. Compt.-Rend. Stances Mtm. Soc. Biol., str. 2, 3: 170171. - - . 1864. Sur les caracttres histologiques du fruit des Cruciftres. Bull. Soc. Bot. France 11: 237-246, 288-293. - - . 1865a. Recherches anatomiques et taxonomiques sur la famille des Crucif~res et sur le genre Sisymbrium en particulier. Thtse, Pads. - - . 1865b. Sur la constitution du fruit des Cruciftres. Compt. Rend. Hebd. Stances Acad. Sci. 61: 404--408. Fox Manic, A. 1970. An anomalous form ofSinapis alba. Bot. Tidskr. 65: 209-224. Friedel, J. 1929. Filiation des Papavtrac~es (2e note). (2) Essai sur I'orthogtnie du fruit chez les Papavtfacies/t siliques. Bull. Soc. Bot. France 76: 285-291. Friis, E. M. & P. K. Endress. 1990. Origin and evolution of angiosperm flowers. Advances Bot. Res. 17: 100-162. Froebe, H. A. & R. Classen-Bockhoff. 1994. Das trialektische Typuskonzept der botanischen Morphologie. Senckenberg-Buch 70: 143-167. Waldemar Kramer, Frankfurt am Main.

282

THE BOTANICAL REVIEW

Frohne, D. 1962. Das Verh~Utnis von vergleichender Serobotanik zu vergleichender Phytochemie, dargestent an serologischen Untersuehungen in Bereich der"Rhoeadales." PI. Med. (Stuttgart) 10: 283-297. & U. Jenseu. 1992. Systematik des Pflanzenreichs. F_~I.4. Gustav Fischer Verlag, Stuttgart, Jena. Fukuhara, T. & M. Lid~n. 1995. Pericarp anatomy in Fumariaeeae-Fumarioideae. Bot. Jahrh. gyst. 117: 499-530. Gagnepain, F. 1900. Nouvelles notes de t~ratologie v~gdtale. Bull. Soc. Hist. Nat. Autun 13: Compt. rend. s~anc, excurs., 37-55. Gallaud, M. 1926. Anomalies exp~rimentales provoqu~es a l'aide d'un puceron sur "Arabis sagittata" DC. Ann. Sei. Nat., Bot., s&. 10, 8: 213-219. Gay, J. 1842. Fumariae officinalis adumhratio, quit suam de Fumariacearum structurfi florali opinionem, in apparato stamineo interpretando novam. Ann. Sei. Nat., Bot., s~r. 2, 18: 214-218. Gazet du Chatelier, G. 1946. Le diagramme de la fleur des Crueif&es. Recueil Tray. Inst. Bot. Montpellier 2: 5-9. Gerber, C. 1899a. Essai d'interpretation du fruit des Crucif~res par l'anatomie t~ratologique. Compt.Rend. S~anees M~m. Soc. Biol., sdr. 10, 6: 291-294. ~.. 1899b. Le pistil des Crucif~res. Compt.-Rend. Sdances M~m. Soc. Biol., sdr. 11, 1: 662-665. ~.. 1899c Le genre Tetrapoma: Sa signification. Compt.-Rend. S~ances M~m. Soc. Biol., s~r. 11,1: 665-667. --.. 1899d. Les fruits tri et quadriloculaires de Crucif~res: Leur valeur thdorique. Bull. Soc. Bot. France 46: IX-XXX. ~-. 1900a. Observations au sujet de la communication de M. Martel: Sur les analogies anatomiques qui relient la fleur de l'Hypecoum ~ celle des Fumariaedes et des Crucif~res. Actes I Congr. Int. Bot. Paris: 176-181. ~-. 1900b. Recherches sur le hombre des feuilles carpellaires qui entrent dans la constitution du gyndc~e des Crucif~res. Bull. Sei. France Belgique 33: 493-521. 9 1904a. l~tamines carpellisdes de la Girofl6e. Compt. Rend. Hebd. Sdances Acad. Sci. 139: 219-221. 9 1904b. Siliques emboR6es du Lepidium villarsii G.G. Leur signification. Compt. Rend. Hebd. Sdances Acad. Sci. 139: 302-304. 9 1904e. Thdorie earpellaire de la fausse eloison des Crucif~res. Compt. Rend. Soc. Biol. 56: 1109-1111. 9 1904d. Faisceaux inverses et destruction du parenchyme des cloisons correspondantes dans la silique des Crucifdres. Compt. Rend. Soe. Biol. 56:1111-1113. 9 1904e. Communication sur le fruit des Crucif~res. Bull. Soc. Bot. France 51, Session jubilaire a Paris (Aofit 1904): CLXVII-CLXVIII. - 1905a. Petales inversds du Cheiranthus cheiri L. var. l-gynantherus DC. et fausse cloison des Crucifdres. Compt. Rend. Hebd. Sdances Aead. Sei. 140:1109-1111. 1905b. Le diagramme floral des Crueifdres. Compt. Rend. Hebd. Sdanees Acad. Sci. 140: 1143-1146.

1905c. Interprdtation anatomique de la fleur des Crueif~res. Compt. Rend. Soe. Biol. 58: 624--626. - 1905d. Interpr6tation anatomique des ovaires bi, tri, quadriloculaires des Crueifdres. Compt.Rend. Sdances Mdm. Soc. Biol. 58: 626--628. - 1907a. Thdorie de Celakowsky sur la cloison des Crucif~res. Compt.-Rend. Sdances Mdm. Soc. Biol. 62: 974-976. - 1907b. L'arc renvers6 de Aubrietia deltoidea DC. Compt.-Rend. Sdances Mdm. Soc. Biol. 62: 976-978. Gershenzon, d. & T. J. Mabry. 1983. Secondary metabolites and the higher classification of angiosperms. Nordic J. Bot. 3: 5-34. Gladis, T. & K. ltammer. 1992. Die Gaterslebener Brassica-Kollektion--Brassicajuncea, B. napus, B. nigra und B. rapa. Feddes Repert. 103: 469-507. G o d r o n , D.-A. 1846. Description d'une monstruosit6 observ~e sur la fleur de plusieurs Crucif~res. Mdm. Soc. Roy. Sci. Nancy 1845: 39-50.

CARPEL NUMBER

~..

283

1864. Mrmoiro sur l'inflorescenco ot des flours dos Crucifh,es. Ann. Sci. Nat., Bet., srr. S, 2: 281-305, plate 18. ~.. 1871-1872. Mrlangos do trratologio vegetale, I: De l'organisation dos carpelles dans Io fruit dos Pavots. M(~m. See. Sci. Nat. Cherbourg 16: 81-94. ~.. 1874. Nouvoaux mrlanges do trratologie vrgrtale. Mrm. Soc. Sci. Nat. Cherbourg 18:318-352. ~.. 1877. Observations sur un genre particulier de proliflcations m(.~dianes des flours. Bull. Soc. Bet. France 24: 192-198. ~-. 1878. Etudes sur los prolifications. Mrm. Acad. Stanislas, s&'. 4, 10: 274-342. Goebol, K. 1933. Organographio der Pflanzen. III/2, Samenpflanzen. Gustav Fischer Verlag, Jena. Goethe, I. W. v. 1790. Versuch die Metamorphose der Pflanzen zu orklaren. Ettinger, Gotha. Guldborg, A. 1986. Classification, evolution, and phylogeny of the families of dicotyledons. Smithsonian Contributions to Botany, 58. Smithsonian Institution Press, Washington, D.C. Gunnermann, C. 1979. Karpomorphologischo Untersuehungen an ausgew~thlten PapaveraceenGattungen als Beitrag zur Kenntnis der Sippenstruktur dieser Familie. Diss., Humboldt University, Berlin. ~-. 1980. Beitrage zur Kenntnis der Gynooceumsstruktur der Papaveracoao Juss. s.str. Feddes Rep. on. 91: 593-613. ~.. 1982. Karpomorphologischo Untersuchungen an ausgewahlten Papaveracoen-Gattungen als Boitrag zur Kenntnis der Sippenstruktur dieser Familie. (Author's report.) Gloditschia 9: 387-388. Gt;ppert, H. R. 1832. III. Botanische Notizen. Flora 15: 252-253. ~.. 1850a. Ueber metamorphosierte Mohnkrpfo. Bet. Zoitung 8: 514. ~-. 1850b. Kurze Notizen. Bet. Zeitung 8: 664. Goto, IC & g. M. Moyerowitz. 1994. Function and regulation of the Arabidopsis floral homeotic gene PISTILLATA. Genes & Development 8: 1548-1560. Goyal, S. C. & A. Pillai. 1985. Morpho-histological studies of the shoot apex in Papaver rhoeas Linn. Phytomorphology 35: 257-263. Gr(~goire, V. 1931. La valour morphologique des carpellos dans los angiospermes. Bull. Acad. Roy. Sci. Belg., CI. Sci., srr. 17, 5: 1286-1302. ---. 1938. La morphogrnrso ot l'autonomio morphologique do l'appareil floral, I: Le carpelle. Collule 47: 287-452. Grifflth, W. 1847. Notulae ad plantas asiatieas / leones plantarum asiaticarum. Ft. 1, Development of organs in phanerogamous plants. Bishop's College Press, Calcutta. Gris, A. 1858. Note sur quelques cas de monstruosit6 et sprcialement sur la Rose verte. Ann. Sci. Nat., Bet., srr. 4, 9: 76-83. Grnenland, J. 1860. Sur deux monstruositrs observres dans le genre Papaver. Rev. Hurt. 1860: 292-296. Gurd/~s, M. 1964. Le gynrcre des Crucif~res: Septum et stigmates. Compt. Rend. Hebd. Srances Acad. Sci. 259: 2292-2295. 9 1965. La notion de replum ot la morphologie du earpello. Compt. Rend. Hobd. Srances Acad. Sci. 260:6671-6674. 9 1966a. Le carpelle du Prunuspaniculata Thunb. (P. sorrulata Lindl.): Ses modifications morphologiquos dans les flours doubles et sa signification. Flora 156: 464-499. 9 1966b. Stamen, carpel and ovule. The teratologieal approach to their interpretation. Advancing Frontiers PI. Sei. 14: 43-108. - - . 1967. The cruciferous flower. Advancing Frontiers P1. Sci. 18: 169-218. 9 1969. Homologies de l'rtamino ot du carpello chez Papaver orientale L. Compt. Rend. Hebd. Srancos Acad. Sci. srr. D, 268: 926-929. ~ . 1972. Stamen-carpel homologies. Flora 161: 184-208. 9 1973. Carpel morphology and axis-sharing in syncarpy in some Rutaceae, with further comments on "New Morphology." Bot. J. Linn. Soc. 66: 55-74. ~ . 1977. Lo gyn~cre de Podophyllum (Berbeddacoae): Monomrrio waie et placentation suturalo do la portion congrnitalement close du carpelle. Compt. Rend. Hobd. Sranees Acad. Sei. srr. D, 285: 755-758. ~ . 1979. Morphology of seed-plants. J. Cramer, Vaduz.

284

- -

THE BOTANICAL REVIEW

& P. Dupuy. 1964. Transformation, au cours des virescences, d'ovules, uni- ou bitegmin6s, en phyllomes v6g6tatifs ou sexualis6s. Compt. Rend. Hebd. S6ances Acad. Sci. 258: 2168-2171. Gu6rin, P. 1924. Anomalie florale du Cheiranthus cheiri L. Bull. Soc. Bot. France, s6r. 4, 24: 676-678, plate 7. Guillard, A. 1857. Sur deux chlorantbies. Bull. Soc. Bot. France 4: 760-762. Gunn, C. R. & M. I. Seldin. 1976. Seeds and fruits of North American Papaveraceae. Techn. Bull. U.S.D.A. No. 1517: 1-96. Giinther, K.-F. 1975. Beitrage zur Morphologie und Verbreitung tier Papaveraceae, pt.l: Infloreszenzmorphologie der Papaveraceae; Wuchsformen der Chelidonieae. Flora 164:185-234. Guyot, M. 1962. Virescence et organisation de la fleur des Crucif~res. Bull. Soc. Bot. France 109: 170-176. - & P. Gavaudan. 1961. T&atologie florale, virescence et r6alisation sexuelle chez les Crucif~res. Compt. Rend. Soc. Biol. 155: 2006-2009. Hagemann, W. 1975. Eine mOglicbe Strategie der vergleicbenden Morphologie zur phylogenetischen Rekonstruktion. Bot. Jahrb. Syst. 96: 107-124. Hagen, C. 1873. Untersuchungen fiber die Entwickelung und Anatomie der Mesembryanthemeen. Inaugural-Diss., C. Georgi, Bonn. Hagerup, O. 1934. Zur Abstammung einiger Angiospermen durch Gnetales und Coniferae. Kongel. Danske Vidensk. Selsk. Biol. Meddel. 11: 1-83. Halsted, B. D. 1894. Botanical notes. Bull. Torrey Bot. Club 21: 296. Hamburger, E. 1842. Symbolae quaedam ad doctrinam de plantarum metamorphosi. Inaugural-Diss., Bratislavia. Hanausek, T. F. 1882. Ueber eine Vergrtinung yon Sinapis arvensis L. (form. dasycarpa Neilr.). Ocsterr. Bot. Z. 32: 315-316. Hanelt, P. 1986. Cruciferae. Pp. 272-332 in J. Schultze-Motel (ed.), Rudolf Mansfelds Verzeichnis landwirtschaftlicher und gartnerischer Kulturpflanzen (ohnc Zierpflanzen). Ed. 2. Vol. 1. Akademie-Verlag, Berlin. Hannan, G. L. 1980. Heteromericarpy and dual seed germination modes in Platystemon californicus (Papaveraceae). Madroflo 27:163-170. Hannig, E. 1901. Untersuchungen fiber die Scheidew[[nde der CruciferenfrQchte. Bot. Zeitung 59: 207-245, plates 8-10. Hanstein, J. 1873. Bericht Qber einige entwicklungsgeschichtliche Arbeiten aus dem Bonner Botanischen Institut. Sitzungsber. Niederrhein. Ges. Natur- Heilk. Bonn, 4 August 1873 in Verh. Naturhist. Vereines Preuss. Rheinlande Westphalens 30 (3. Folge 10): 192-202. Harms, H. 1936. Reihe Rhoeadales. Pp. 1-4 in A. Engler & H. Harms (eds.), Die Natfirlichen Pflanzenfamilien. Ed. 2. Vol. 17b. Wilhelm Engelmann, Leipzig. Haskeil, G. 1954. Pleiocotyly and differentiation within angiosperms. Phytomorphology 4: 140-151. Hecht, U. 1990. Vergleichende BlfRenontogenie bei Orobanche. Diploma paper (unpubl.), Heidelberg University. Hegi, G. 1986. Illustrierte Flora von Mitteleuropa. Vol. 4, pt. 1. Ed. 3. Verlag Paul Parey, Berlin, Hamburg. Hegnauer, R. 1961. Die Gliederung der Rhoeadales sensu Wettstein im Lichte der Inhaltsstoffe. P1. Med. (Stuttgart) 9: 37-46. --.. 1964, 1969, 1973. Chemotaxonomie dcr Pflanzcn. Vols. 3, 5, 6. Birkhauser Verlag, Basel u. Stuttgart. Heinrieher, E. 1891. Uber einen eigenthiimlichen Fall yon Umgestaltung eincr Oberhaut und dessen biologische Dcutung. Sitzungsbcr. Kaiserl. Akad. Wiss. Wir Math.-Naturwiss. CI. 99, sec. 1: 25-39. --.. 1925. Zwei in histologischer und biologiscber Hinsicht bemerkenswerte Frilchte yon Fumariaceen. Flora 118/119: 184-197. Henslow, G. 1880. On the origin of floral aestivations, with notes on the structure of the cruciferous flower, on that of Adoxa, and on the corolla of Primula. Trans. Linn. Soc. London Bot., ser. 2, 1: 177-245, plate 25. ---. 1888. The origin of floral structures through insect and other agencies. Kegan Paul, Trench, London.

CARPEL NUMBER

--..

285

1891. On the vascular systems of floral organs, and their importance in the interpretation of the morphology of flowers. J. Linn. Soc. Bot. 28: 151-197, plates 23-32. Hergt, B, 1913. In Bericht fiber die Herbsthauptversarnmlung in Frankenhausen am 6. nnd 7. Oktober 1912. Mirth. Thfiring. Bot. Vereins, n.s., 30: 129. Heslop-Harrison, 3. 1952. A reconsideration of plant teratology. Phyton (Horn) 4: 19-34. Heslop-Harrison, Y. & IC R. Shivanna. 1977. The receptive surface of the angiosperm stigma. Ann. Bot. (London), n.s., 41: 1233-1258, plates 1-4. Hilger, H. H. 1992. Morphology of Heliotropium (Boraginaceae) dispersal units. Bot. Acta 105: 387-393. & 3. R. Hoppe. 1995. Die rnorphologische Vielfalt der generativen Diasporen--Pr~sentation eines Lehr- und Lernschernas. Feddes Repert. 106:503-513. Hill, J. P, & E. M. Lord9 1989. Floral development in Arabidopsis thaliana: A comparison of the wild type and the horneotic pistillata mutant. Canad. J. Bot. 67: 2922-2936. Himrnelbaur, W. 1914. Die Berberidaceen und ihre Stellung irn System. Denkschr. Kaiserl. Akad. Wiss. Wien. Math.-Natunviss. KI. 89: 733-795. Hoehstetter, C. F. 1847. Aufbau der Graspflanze, ihres Halms, Blfithenstands, ihrer Blfithe und Frucht, nebst rnorphologischen und taxonornischen Andeutungen, andere Pflanzenfarnilien betreffend. Jahresh. Vereins Vaterl. Naturk. W•rtternberg 3: 1-84. ---. 1848. Nachtraglicher Cornmentar zu der Abhandlung "Auibau der Graspflanze." Flora 31: 105-118, 121-135, 140-152, 154-167, 171-188. Hoffmann, H. 1877. Culturversuche. Bot. Zeitung 35: 281-295, plate III. 9 1878. Culturversuehe. Bot. Zeitung 37:193-207, plate IIIB. 9 1881. Rfickblick auf rneine Variations-Versuche yon 1855-1880. Bot. Zeitung 39:393-399. Hofrneister, W. F. B. 1851. Vergleichende Untersuchungen der Keirnnng, Entfaltung und Fruchtbildnng h6herer Kryptogarnen (Moose, Farm, Equisetaceen, Rhizokarpeen und Lycopodiaceen) und tier Sarnenbildung der Coniferen. F. Hofrneister, Leipzig. Reprint: 1979, Hist. Nat. Classica 105. Crarner, Vaduz. Holland, R. 1882. Monstrous development of Cheiranthus cheiri. J. Bot. 20: 282-283. Hoot, S. B., 3. W. Kadereit, F. R. Blattner, K. B. Jork, A. E. Sehwarzbaeh & P. R. Crane9 1997. Data congruence and phylogeny of the Papaveraceae s.l. based on four data sets: atpB and rbeL sequences, trnK restriction sites, and morphological characters. Syst. Bot. 22: 575-590. Horsfield, T. 1838-1852. Plantae javanicae rariores, descriptae iconibusque illustratae, quas in insula Java, annis 1802-1818, legit et investigavit Thomas Horsfield, M.D. e siccis descriptiones et characteres plurimarurn elaboravit Joannes J. Bennett; observationes structurarn et affinitates praesertim respicientes passim adjecit Robertus Brown. G. H. Allen, London. Howell, J. W. 1842. On some hitherto unnoticed peculiarities in the structure of the capsule of Papaveraceae; and on the nature of the stigma of Cruciferae. Ann. Mag. Nat. Hist. 10: 248-255. Huber, H. 1982. Die zweikeirnblattrigen GehOlze irn System der Angiospermen. Mitt. Bot. Staatssamml. Mfinchen 18: 59-78. 9 1991. Angiosperrnen: Leitfaden durch die Ordnnngen und Farnilien der Bedecktsarner. Gustav Fischer Verlag, Stuttgart, New York. Huisgen, F. 1873. Untersuchungen fiber die Entwicklung tier Placenten. Inaugural-Diss., C. Georgi, Bonn. Hutchinson, J. 1959. The families of flowering plants. Vol. 1, Dicotyledons. Ed. 2. Clarendon Press, Oxford. ~ . 1967. The genera of flowering plants (Angiosperrnae), Dicotyledones. Vol. 2. Clarendon Press, Oxford. - - . 1969. Evolution and phylogeny of flowering plants; Dicotyledons: Facts and theory. Academic Press, London, New York. 9 1973. The families of flowering plants. Ed.3. Clarendon Press, Oxford. ll'ina, G. M. 1968. K rnorfologii ginetseya ploskotychinochnika kalifornskogo (Platystemon californicus Benth.). (On gynoecium morphology ofPlatystemon californicus Benth.) Pp. 142-156 in L. V. Kudryashev (ed.), Morfologiya vysshich rastenii. Nauka, Moscow. (In Russian.) lltis, H. H. 1957. Studies in the Capparidaceae, III: Evolution and phylogeny of the western North American Cleomoideae. Ann. Missouri Bot. Gard. 44:77-119.

286

THE BOTANICAL REVIEW

& T. Ruiz Zapata. 1997. Studies in the Capparaceae, XIX: Cleome torticarpa n. sp., a Venezuelan endemic. Novon 7: 367-372. - - , L. J. Cumana C., R. E. Delgado M. & G. C. Aymard. 1996. Studies in the Capparidaeeae, XVIII: A new giant-fruited Capparis (C. muco) from eastern Venezuela. Novon 6: 375-384. Irish, V. F. & I. M. Sussex. 1990. Function of the apetala-1 gene during Arabidopsis floral development. PI. Cell 2: 741-753. Jack, T., L. L. Brockman & E. M. Meyerowitz. 1992. The homeotic gene APETALA3 of Arabidopsis thaliana encodes a MADS box and is expressed in petals and stamens. Cell 68: 683-697. - - , G. L. Fox & E. M. Meyerowitz. 1994. Arabidopsis homeotic gene APETALA3 ectopic expression: Transcriptional and posttranscriptional regulation determine floral organ identity. Cell 76: 703-716. Jacob, F., E. J. Jliger & E. Ohmann. 1994. Botanik. Ed. 4. UTB fi~rWissenschaft: Uni-Taschenb~cher 1431. Gustav Fischer Verlag, Jena. Jaeobs, M. 1964. The genus Crateva (Capparaceae). Blumea 12: 177-208. J a n c h e n , E. 1942. Das System der Cruciferen. Oesterr. Bot. Z. 91: 1-28. - - . 1950. Die Herkunft der Angiospermen-Bliite und die systematischen Stellung der Apetalen. Oesterr. Bot. Z. 97: 129-167. Janse, J. M. 1929. A natural system of classification of monstrosities. Ann. Jard. Bot. Buitenzorg 40: 87-138. J e n s e n , U. 1967. Serologische Beitr/ige zur Frage der Verwandtschaft zwischen Ranunculaceen und Papaveraceen. Ber. Deutsch. Bot. Ges. 8: 621-624. --.. 1974. The interpretation of comparative serological results. Pp. 217-227 in G. Benz & J. Santesson (eds.), Chemistry in botanical classification. Proceedings of the 25th Nobel Symposium. Academic Press, New York, London. --.. 1995. Secondary compounds of the Ranunculiflorae. Pp. 85-97 in U. Jensen & J. W. Kadereit (eds.), Systematics and evolution of the Ranunculiflorae. PI. Syst. Evol., Suppl. 9. SpringerVerlag, Vienna, New York. - & B. Greven. 1984. Serological aspects and phylogenetic relationships of the Magnoliidae. Taxon 33: 563-577. - & J. W. Kadereit (eds.). 1995. Systematics and evolution of the Ranunculiflorae. P1. Syst. Evol., Suppl. 9. Springer-Verlag, Vienna, New York. J e p s o n , W. L. 1894. Teratological notes. Erythea 2:14-15. J e r n s t e d t , J. A. & C. Clark. 1979. Stomata on the fruits and seeds ofEschscholzia (Papaveraceae). Amer. J. Bot. 66: 586-590. J o h r i , B. M., K. B. Ambegaokar & P. S. Srivastava. 1992. Comparative embryology of Angiosperms. Vol. 1. Springer-Verlag, Berlin, Heidelberg. Joshi, A. C. 1933. Some abnormal flowers ofArgemone mexicana, Linn. J. Indian Bot. Soc. 12: 255-271. ---. 1939. Some abnormal flowers ofArgemone mexicana and their bearing on the morphology of the gynoecium of Papaveraceae. Ann. Bot. (London), n.s., 3: 503-504. K a d e r e i t , J. W. 1987. The taxonomy, distribution, and variability of the genus Roemeria Medic. (Papaveraceae). Flora 1 7 9 : 1 3 5 - 1 5 3 . ---. 1993. Papaveraceae. Pp. 494-506 in K. Kubitzki, J. G. Rohwer & V. Bittrich (eds.), The families and genera of vascular plants. Vol. 2. Flowering Plants. Dicotyledons: Magnoliid, Hamamelid and Caryophyllid Families. Springer-Verlag, Berlin, Heidelberg. ---, F. R. Blattner, K. B. Jork & A. Sehwarzbaeh. 1994. Phylogenetic analysis of the Papaveraceae s.l. (incl. Fumariaceae, Hypecoaceae, and Pteridophyllura) based on morphological charaeters. Bot. Jahrb. Syst. 116: 361-390. ---, , & .1995. The phylogeny of the Papaveraceae sensu lato: Morphological, geographical and ecologic implications. Pp. 133-145 in U. Jensen & J. W. Kadereit (eds.), Systematies and evolution of the Ranunculiflorae. PI. Syst. Evol., Suppl. 9. Springer-Verlag, Vienna, New York. ---, A. E. Schwarzbaeh & K. B. Jork. 1997. The pylogeny ofPapaver s.l. (Papaveraceae): Polyphyly or monophyly? P1. Syst. Evol. 204: 75-98. - -

CARPEL NUMBER

287

Kadkol, G. P., V. C. Beilharz, G. M. I-Ialloran & R. H. MacMillan. 1986. Anatomical basis of shatterresistance in the oilseed brassicas. Austral. J. Bot. 34: 595--601. Kajanus, B. 1919. Genetische Studien fiber die Bltiten yon Papaver somniferum L. Ark. Bot. 15: 1-87. Kapoor, L. D. & B. M. Sharma. 1963. drgemone mexieana L.---Organograpby and floral anatomy with reference to the laticiferous system. Phytomorpbology 13: 465--473. Karrer, A. B. 1991. Bltitenentwicklung und systematische Stellung der Papaveraceae und Capparaceae. Inaugural-Diss., Zurich. Kato, M. 1990. Ophioglossaceae: A hypothetical archetype for the angiosperm carpel. Bot. J. Linn. Soc. 102:303-311. ~-. 1991. Further comments on an opbioglossoid archetype for the angiosperm carpel: Ovular paedomorpbosis. Taxon 40: 189-194. Kaute, U. 1963. Beitrage zur Morphologie des Gynoeceums der Berberidaceen mit einem Anhang tiber die Rhizomknospe von Plagiorhegma dubium. InauguraI-Diss., Free University, Berlin. Keighery, G. J. 1981. The breeding system ofEmblingia (Emblingiaceae). PI. Syst. Evol. 137: 63-65. Kerner v. Marilaun, A. 1898. Pflanzenleben. Vol. 2. Ed. 2. Bibliographisches Institut, Leipzig, Vienna. Kh~inh, T. C. 1973. Beitrag zur Kenntnis der Sippenstruktur des Genus Dicentra subgenus Dactylicapnos (Wall.) K. R. Stern (Fumariaceae). Feddes Repert. 83:511-575. Klaanna, L. P. 1931. Abnormal flowers of mustard (Brassica alba). New Phytol. 30: 73-75. Klm, Y.-D. & R. K. Jansen. 1995. Phylogenetic implications of chloroplast DNA variation in the Berberidaceae. Pp. 341-349 in U. Jensen & J. W. Kadereit (eds.), Systematics and evolution of the Ranunculiflorae. PI. Syst. Evol., Suppl. 9. Springer-Verlag, Vienna, New York. Klrschleger, F. 1854. Ueber eine Vergrtinung (Virescenz) derDiclytra spectabilis. Flora 37: 545-546. Klein, d. 1894. Der Bau tier Cruciferenbltithe auf anatomischer Grundlage. Bet. Deutsch. Bot. Ges. 12: 18-24. Klinsmann, ?. 1835-1836. Ueber zwei Pflanzen-Monstrositliten. Linnaea 10: 604-608. Klopfer, K. 1974. Systematik tier Pflanzen. Programmiertes Studienmaterial zur Wiederbolung und Ubung. VEB Gustav Fischer Verlag, .lena. Koch, K. 1869. In Tageblatt der 43. Versammlung Deutscber Naturforscher und ~rzte in Innsbruck: 132-133. Kolbe, K.-P. 1978. Serologischer Beitrag zur Systematik der Capparales. Bot. Jabrb. Syst. 99: 468-489. Komaki, M. K., K. Okada, E. Nishino & Y. Shimura. 1988. Isolation and characterization of novel mutants ofArabidopsis thaliana defective in flower development. Development 104:195-203. Komarov, V. L. (ed.). 1939. Flora SSSR. (Flora of the U.S.S.R.) Vol. 8. Edited by N. A. Bush. Izdatel'stvo Akademii Nauk, Moscow, Leningrad. (In Russian.) Kozo-Poljansky, B. M. 1937. Teratologija tsvetka i novye voprosy ego teorii, I. (Teratology of the flower and new problems in its theory, I.) Sovetsk. Bot. 6: 56-70. (In Russian.) ---. 1945. Proiskhozhdenie tsvetka Cruciferae v teratologicheskom osveshchenii. Teratologia tsvetka i novye voprosy ego teorii, III. (Development of flower in Cruciferae as viewed from the standpoint of teratology. Teratology of the flower and new problems in its theory, III.) Bot. Zhurn. (Moscow & Leningrad) 30: 14-30. (In Russian; brief summary in English.) Krause, E. H. L. 1900. Floristische Notizen. Bot. Centralbl. 21: 205-206. Krizek, B. A. & E. M. Meyerowitz. 1996. The Arabidopsis homeotic genes APETALA3 and PISTILLATA are sufficient to provide the B class organ identity function. Development 122:11-22. Kronfeld, M. 1886. Studien zur Teratologie der Gew~lchse, I. Verh. K. Kgl. Zool.-Bot. Ges. Wien 36: 103-122. Kubitzki, K., J. G. Rohwer & V. Bittrich. (eds.). 1993. The families and genera of vascular plants. Vol. 2 . Flowering Plants. Dicotyledons: Magnoliid, Hamamelid and Caryophyllid Families. Springer-Verlag, Berlin, Heidelberg. Kumazawa, M. 1938a. Systematic and phylogenetic consideration of the Ranunculaceae and Berberidaceae. Bot. Mag, (Tokyo) 52: 9-15. ~-. 1938b. On the ovular structure in the Ranunculaceae and Berberidaceae. J. Jap. Bot. 14: 10-25. Kunth, K. S. 183 I. Handbuch der Botanik. Duncker und Humblot, Berlin. --.. 1833. Zwei botanische Abhandlungen . . . . fiber die Bltithen- und Fruchtbildung der Cruciferen . . . . KOnigliche Akademie der Wissenschaften, Berlin.

288

THE BOTANICAL REVIEW

Kuusk, V. 1960. On the composition of the ovary of Cruciferae. Izv. Akad. Nauk Estonsk. S.S.R. Ser. Biol. 9, No. 4: 314-324. (In Estonian; summaries in Russian and English.) Lain, H. J. 1948. Classification and the New Morphology. Acta Biotheor. 16: 127-182. ---. 1950. Stachyospory and phyllospory as factors in the natural system of the cormophyta. Svensk Bot. Tidskr. 44: 517-534. ---. 1959. Taxonomy: General principles and angiosperms. Pp. 3-75 in W. B. Turrill (ed.), Vistas in Botany. Pergamon Press, London, New York. L a n k e s t e r , E. 1848. Gelehrte Gesellschaften. Bot. Zeitung 6: 887-888. --.. 1849. On some vegetable monstrosities illustrating the laws of morphology. Rep. Brit. Assoc. Advaneem. Sei. 18: Notices/Abstracts, 85-86. Layka, S. 1976. Les methodes modemes de la palynologie appliqu6es/t l'6tude des Papaverales. Diss., Montpellier, C.N.R.S.A.O. 12.535. Le Maout, E. 1844. Lefons elementaires de botanique fond6es sur l'analyse de 50 plantes vulgaires et formant un trait6 complet d'organographie et de physiologie v6g6table, a l'usage des 6tudiants et des gens du monde. Langlois et Leclerq, Paris. - & J. Decaisne. 1868. Trait6 g6n6ral de botanique descriptive et analytique. Firmin Didot, Paris. Le Sourd-Dussiples, E. & G. Bergeron. 1861. Note sur un cas de m6tamorphose ascendante: Transformation des 6tamines en feuilles carpellaires. Bull. Soc. Bot. France 8: 348-351. Leavitt, R. G. 1909. A vegetative mutant, and the principle of homeosis in plants. Bot. Gaz. (London) 47: 30~8. Leelerc, F. 1866. Quelques observations sur les fonetions du cadre placentaire et de la columelle darts les fruits des Crucif6res. M6m. Soc. Emul. Doubs, s6r. 4, 2: 349-358. L e c l e r c du Sablon, M. 1884. Recherches sur la dehiscence de fruits/t pericarpe sec. Ann. Sci. Nat., Bot., s6r. 6, 18: 31-34. Lehmann, N. L. & R. Sattler. 1993. Homeosis in floral development ofSanguinaria canadensis and S. canadensis 'Multiplex" (Papaveraceae). Amer. J. Bot. 80: 1323-1335. L e i n f e l l n e r , W. 1956. Zur Morphologie des Gyn6zeums yon Berberis. Oesterr. Bot. Z. 103: 600612. Leins, P. & C. Erbar. 1982. Das monokarpenate Gynoeceum von Monodora crispata (Annonaceae). Beitr. Biol. Pflanzen 57: 1-13. - & .1985a. Zur frtlhen Entwicklungsgeschichte des Apiaceen-Gynoeceums: Ein Beitrag zur Entmythologisierung des unterst/lndigen Fruchtknotens. Bot. Jahrb. Syst. 106: 53-60. - & .1985b. Ein Beitrag zur Bltitenentwicklung der Aristolochiaceen, einer Vermittlergruppe zu den Monokotylen. Bot. Jahrb. Syst. 107: 343-368. - & G. Metzenauer. 1979. Entwicklungsgeschichtliche Untersuchungen an Capparis-Blflten. Bot. Jahrb. Syst. 100: 542-554. L a s t i b o u d o i s , T. 1823a. Notice sur le fruit des Papav6rac6es. Soc. Amateurs Sci. Lille Recueil Trav. 1819-1822: 181-194. - - . 1823b. M6moire sur les fruits siliqueux. Soc. Amateurs Sci. Lille Reeueil Trav. 1819-1822: 195-207. Lewis, I. M. 1912. Pistillody in Argemoneplatyceras Link and Otto. Torreya 12: 85-88. Lid6n, M. 1986. Synopsis of Fumarioideae (Papaveraceae), with a monograph of the tribe Fumarieae. Opera Bot. 88: 1-133. - - . 1992. Evolution and classification of seed plants. Progr. Bot. 53:282-317. - - . 1993. Fumariaceae, Pteridophyllaceae. Pp. 310-318 and 556-557 in K. Kubitzki, J. G. Rohwer & V. Bittrich (eds.), The families and genera of vascular plants. Vol. 2. Flowering Plants. Dicotyledons: Magnoliid, Hamamelid and Caryophyllid Families. Springer-Verlag, Berlin, Heidelberg. --~ T. Fukuhara, 3. Rylander & B. Oxelman. 1997. Phylogeny and classification of Fumariaeeae, with emphasis on Dicentra s.l., based on the plastid gene rps 16 intron. P1. Syst. Evol. 206: 411-420. Lignier, O. 1896a. Anatomic compar6e de la fleur des Crucif6res et des Fumari6es. Compt. Rend. Congr. Assoc. Fran~. Avancem. Sci. 25, sess. 1: 176-177. - - . 1896b. La fleur des Crucif6res compar6e a celle des Fumariac6es. Compt. Rend. Congr. Assoc. Franq. Avancem. Sci. 25, sess. 2: 403-406.

CARPEL NUMBER

~ ,

289

1896e. Explication de la fleur des Fumari~es d'apr~s son anatomie. Compt. Rend. Hebd. S~antes Acad. Sei. 122: 630-632. ~.. 1896d. Explication de la fleur des Crucif~res d'apr~s son anatomie. Compt. Rend. Hebd. S~antes Acad. Sei. 122: 675-678. --.. 1897. Recherches sur les fleurs prolif&'es du Cardaminepratensis. Bull. Soc. Linn. Normandie, s~r. 4, I0 (1896): 21-25. ~.. 1904. In Bull. Soc. Bot. France 51: Session j ubilaire a Paris (Aofit 1904), CLXVII. ~.. 1911. Notes anatomiques sur l'ovaire de quelques Papav~rac~es. Bull. Soc. Bot. France s~r. 4, 11: 279-238, 337-344, 429-435. ~-. 1914. Nouvelles contributions A la connaissance de la fleur des Fumari~es et des Crucif~res. Compt. Rend. Hebd. S~anees Acad. Sei. 159: 202-205. ~-. 1915. Eschscholtzi6es: Explication anatomique de la fleur. Bull. Soc. Bot. France s~r. 4, 15: 298-342. Lindley, J. 1828. Text to plate 1168 (Eschscholtzia californica). Bot. Reg. 14. ~-. 1830. An introduction to the natural system of botany. Longman, Rees, Orme, Brown, and Green, London. ---. 1847. The vegetable kingdom. Ed.2. Bradbury & Evans, Whitefriars, London. ~.. 1859. Zur Morphologie des Papaver somniferum. (Translation of an editorial in Gardeners' Chronicle.) Bonplandia 7: 336-337. Loconte, H. 1993. Berberidaceae. Pp. 147-152 in K. Kubitzki, J. G. Rohwer & V. Bittrich (eds.), The families and genera of vascular plants. Vol. 2. Flowering Plants. Dicotyledons: Magnoliid, Hamamelid and Caryophyllid Families. Springer-Verlag, Berlin, Heidelberg. - & J, R. Estes. 1989a. Generic relationships within Leonticeae (Berberidaceae). Canad. J. Bot. 67: 2310-2316. - - & .1989b. Phylogenetic systematics of Berberidaceae and Ranunculales (Magnoliidae). Syst. Bot. 14: 565-579. ---, L. M. Campbell & D. Win. Stevenson. 1995. Ordinal and familial relationships of Ranunculid genera. Pp. 99-118 in U. Jensen & J. W. Kadereit (eds.), Systematics and evolution of the Ranunculiflorae. P1. Syst. Evol., Suppl. 9. Springer-Verlag, Vienna, New York. Long, A. G. 1977. Lower carboniferous pteridosperrn cupules and the origin of angiosperms. Trans. Roy. Soc. Edinburgh 70: 13-35. Loreh, d. 1963. The carpel--A case-history of an idea and a term. Centaurus 8: 269-291. Magnus, P. 1876. In XXXIV. Sitzung vom 26. Mai 1876. Verh. Bot. Vereins Prov. Brandenburg 18: 76-79. - - . 1877. In XLVII. Sitzung vom 27.Juli 1877. Verh. Bot. Vereins Prov. Brandenburg 19: 95-96. Mareello, L. 1903. Sopra alcuni casi di teratologia vegetale. Boll. Soe. Naturalisti Napoli, set. 1, 17: 41-44. Marehand, E. 1896. Note sur la fleur des Crucif~res ~.propos d'une anomalie florale chez le Cheiranthus cheiri L. Bull. Soc. Sci. Nat. Ouest France 6: 159-177, plate 3. Marehand, L. 1863-1864. Note sur des fleurs monstrueuses d'Epimedium. Adansonia 4:127-132. Martel, E. 1895. Di una nuova interpretazione dell'architettonica florale delle Crocifere e generi afflni. Atti Accad. Sci. Torino, CI. Sci. Fis. Mat. Nat. 31: 587-592. - - . 1898. Note sur le diagramme floral des Crucifies et des Fumariac6es. J. Bot. (Morot) 12: 29-31. ~ . 1899. Contribuzione all'anatomia dell'Hypecoumprocumbens. Mem. Reale Accad. Sci. Torino, ser. 2, 48: 209-220, plates 1, 2. - - . 1900. Observations sur les analogies anatomiques que relient la fleur de l'Hypecoum ~ celle des Fumariac6es et des Crucif~res. Aetes I. Congr. Int. Bot. Paris: 168-175. - - . 1902. Intorno all'unit~ anatomica e morfologica del fiore delle Crociflore. Mem. Reale Accad. Sci. Torino, ser. 2, 51: 241-266. Martin, P. G. & J. M. Dowd. 1991. Studies of angiosperm phylogeny using protein sequences. Ann. Missouri Bot. Gard. 78: 296-337. Massalongo, C. 1902. Nuove spigolature teratologiche; IIa nora. Boll. Soc. Bot. Ital. 1902:134-138. ~ 1908. Osservazioni fitologiche. Madonna Verona 2: 29--40.

290

THE BOTANICAL REVIEW

Masters, M. T. 1886. Pflanzen-Teratologie. Translated by U. Dammer. H. Haessel, Leipzig. Meaeham, C. A. 1980. Phylogeny of the Berberidaceae, with an evaluation of classification. Syst. Bot. 5: 149-172. Meakin, P. J. & J. A. Roberts. 1990a. Dehiscence of fruit in oilseed rape (Brassica napus L.), I: Anatomy of pod dehiscence. J. Exp. Bot. 41: 995-1002. - & .1990b. Dehiscence of fruit in oilseed rape (Brassica napus L.), II: The role of cell wall degrading enzymes and ethylene. J. Exp. Bot. 41:1003-1011. Medieus, F. K. 1789. Philosophische Botanik. Erstes Heft. Von den mannigfaltigen Umhfillungen der Saamen. Neue Hof- u. Akademische Buehhandlung, Mannheim. Meense, A. D. J. 1975. Taxonomic relationships of Salieaceae and Flacourtiaceae: Their bearing on interpretative floral morphology and Dilleniid phylogeny. Acta Bot. Need. 24: 437-457. ---. 1979. The Anthocorm theory--A summary. H. de Vries Laboratory, University of Amsterdam. ---. 1990. Flowers and fossils. Edburon, Delft. Mehta, I. J. & M. F. Moseley Jr. 1981. The floral anatomy ofKoeberlinia Zucc.: Systematic implications. Amer. J. Bot. 68: 482-497. Melchior, H. 1964. A. Engler's Syllabus der Pflanzenfamilien. Ed. 12. GebrOder Borntraeger, BerlinNikolassee. Mellink, J. F. A. 1895. Iets over vormverandering der cities van Sisymbrium alliaria Stop. Ned. Kruidk. Arch., set. 2 , 6 : 248-253, plate 2. Melville, R. 1960. A new theory of the angiosperm flower. Nature 188, No. 4744: 14-18. - - . 1962. A new theory of the angiosperm flower, I: The gynoecium. Kew Bull. 16: 1-50. - - . 1963. A new theory of the angiosperm flower, II: The androecium. Kew Bull. 17: 2-63. - - . 1983. Glossopteridae, Angiospermidae, and the evidence for angiosperm origin. Bot. J. Linn. Soc. 86: 279-323. Merxmiiller, H. & P. Leins. 1966. Zum BEitenbau der Brassicaceae. Ber. Deutsch. Bot. Ges. 79: 250-252. - & .1967. Die Verwandtsehaftsbeziehungen der Kreuzhltitler und Mohngewiichse. Bot. Jahrb. Syst. 86:113-128. Meyen, S. V. 1988. Origin of the angiosperm gynoecium by gamoheterotopy. Bot. J. Linn. Soc. 97: 171-178. Meyer, V. G. 1966. Flower abnormalities. Bot. Rev. (Lancaster) 32:165-218. Meyerowitz, E. M., D. R. Smyth & J. L. Bowman. 1989. Abnormal flowers and pattern formation in floral development. Development 106: 209-217. Mirbel, C. F. 1825a. Remarques sur l'affinit~ des Papav~rac~es avec les Crucif~res. Ann. Sci. Nat. 6: 266-273, plate 11. --.. 1825b. M~moire sur l'organisation du p~ricarpe. Ann. Sci. Nat. 6: 476-487. Mitchell, R. S. 1983. Berberidaceae through Fumariaceae of New York State. Bull. New York State Mus. 451: 1-64. Mizukami, Y. & H. Ma. 1992. Ectopic expression of the floral homeotic gene AGAMOUS in transgenic Arabidopsis plants alters floral organ identity. Cell 71:119-131. - & .1995. Separation of AG function in floral meristem determinacy from that in reproductive organ identity by expressing antisense AG RNA. Plant Molec. Biol. 28: 767-784. Moeliono, B. M. 1970. Cauline or carpellary placentation among dicotyledons. Vols. 1,2. Van Gorcum & Comp. N. V., Dr. H. J. Prakke & H. M. G. Prakke, Assen. Mohl, H. v. 1836. Beobachtungen 0ber die Umwandlung von Antheren in Carpelle. Flora 19: 513-526, 529-543,545-558. Monnet, P. 1913. Sur les fruits pluricarpellaires de Brassica oleracea. Rev. G6n. Bot. 25: 443-447. Moore, D. 1875. On a monstrous state ofMegacarpaea. J. Bot. 13: 56. Moquin-Tandon, A. 1842. Pflanzen-Teratologie. Translated by J. C. Schauer. Haude und Spenersche Buchhandlung, Berlin. - & A. Barker Webb. 1848. On the structure of eruciferous flowers. London J. Bot. 7: 1-17. - - & .1849. Considerations sur la fleur des Crucif~res. M~m. Acad. Sci. Toulouse, s~r. 3, 5: 364-387.

CARPEL NUMBER

291

Moritre, J. 1859. Correspondance. Bull. Soc. Linn. Normandie 4 (1858-1859): 254-255. ~ , 1862. Transformation des ~tamines en carpelles dans plusieurs esp~ces de Pavot. M6m. Soc. Linn. Normandie 12 (1860-1861): 1-15, plates 1, 2. Morini, F. 1891. Osservazioni intorno ad una mostruosi~ del fiore di Capparis spinosa L. Rendiconto Sess. Reale Accad. Sci. Ist. Bologna 1890-1891: 22-26. Morren, C. & J. Deeaisne. 1834. Observations sur la Flore du Japon, suivies de la monograpbie du genre Epimedium (second article). Ann. Sci. Nat., Bot., s~r. 2, 2: 347-361, plates 12, 13. Mory, B. 1979. Beitrage zur Kenntnis der Sippenstruktur der Gattung Glaucium Miller (Papaveraceae). Feddes Repert. 89: 499-594. Motte, J. 1946. Une nouvelle interpr6tation du diagramme des Crucif~res. Recueil Tray. Inst. Bot., Ann. Univ. Montpellier 2: 10-13. ~.. 1957. Le sequelette floral de Moricandia arvensis et ses incidences sur l'interpretation diagrammatique de la fleur des Crucif~res. Opusc. Bot. necnon alia (privately printed) 1: 17-28. Mflller, E. H. 1878. Bericht Uber die Verhandlungen der Section f'tirObst- und Gartenbau im Jahre 1877. Jahresber. Schles. Ges. Vaterl. Cult. 55: 330-346. Murbeek, S. 1912. Untersuchungen ~lberden Blfltenbau de,"Papaveraceen. Kongl. Svenska Vetenskapsakad. Handl. 50: 1-168. Murty, Y. S. 1953. A contribution to the anatomy and morphology of normal and some abnormal flowers of Gynandropsis gynandra (L.) Briq. J. Indian Bot. Soc. 32: 108-122. Narayana, H. S. 1962. Studies in the Capparidaceae, I: The embryology of Capparis decidua (Forsk.) Pax. Phytomorphology 12: 167-177. ~-. 1965. Studies in Capparidaceae, II: Floral morphology and embryology of Cadaba indica Lamk. and Crataeva nurvala Buch.-Ham. Pbytomorphology 15: 158-175. Nelson, A. 1929. The inheritance of sex in an abnormal (carpellodic)wall-flower. Papers & Proc. Roy. Soc. Tasmania 1928:119-122. Nelson, E. 1954. Gesetzmltl3igkcitender Gcstaltwandlung im BltRcnbcreich. lhrc Bcdcutung fllrdas Problem der Evolution. Published by the author, Chcrnex-Montrcux. Nenukov, S. S. 1939. On the relations between the Papavcraceae, Compositac and Campanulalcs. Chron. Bot. 5: 412-414. Neumayer, H. 1924. Die Geschichte dcr BltRe. Abh. Zool.-Bot. Gcs. Wicn 14: l-110. Niekol, M. G. 1995. Phylogeny and inflorescenccs of Berberidaccac---A morphological survey. Pp. 327-340 in U. Jcnsen & J. W. Kadereit (eds.),Systematics and evolution of the Ranunculiflorac. PI. Syst. Evol., Suppl. 9. Springer-Vcrlag, Vienna, New York. Nicotra, L. 1880. Ccnno intorno ad alcune anomalie vegctali.Nuovo Giorn. Bot. Ital.12: 48-51. Norman, J. M. 1857. Quelques observations de morphologie vdgdtale faites au jardin botanique de Christiania.Programme dc l'Universitepour Ic Icr Sere. 1857. Imprimerie de Broggcr & Christie, Oslo (Christiania). Norris, T. 1941. Torus anatomy and nectary characteristicsas phylogcnetic criteriain the Rhoeadales. Amer. J. Bot. 28: I01-I13. Nowicke, J. W. & J. J. Skvarla. 198 I. Pollen morphology and phylogcnetic relationshipsof the Bcrbcridaccae. Smithsonian Contr. Bot. 50: 1-83. Okada, K., M. K. Komaki & Y. Shimnra. 1989. Mutational analysis of pistilstructureand development ofArabidopsis thaliana. Cell Different.Developm. 28: 27-38. Overbeek, F. 1925.0ber den Mcchanismus dcr Samenausschlcuderung yon Cardamine impatiens L. Bcr. Deutsch. Bot. Ges. 43: 469-475. Ownbey, G. B. 1947. Monograph of the North American speciesof Corydalis. Ann. Missouri Bot. Gard. 34: 187-259. Parashar, G. & V. Singh. 1986. Development of the gynoecium in Solanaceac. Phytomorphology 36: 219-227. Parkin, J. 1926. Comments on the theory of the solidcarpel and carpel polymorphism. New Phytol. 25: 191-201. ~ , 1955. A plea for a simpler gynoecium. Phytomorphology 5: 46-57. Pauz6, F. & R. Sattler. 1979. La placentation axillairechez Ochna atropurpurea. Canad. J. Bot. 57: I00-I07.

292

THE BOTANICAL REVIEW

Pax, F. & K, Hoffmann. 1936. Capparidaceae. Pp. 146-223 in A. Engler & H. Harms (eds.), Die NatOrlichen Pflanzenfamilien. Ed 2. Vol. 17b. Wilhelm Engelmann, Leipzig. Payer, J.-B. 1857. Traite d'organog6nie compar6e de la fleur. Victor Masson, Paris. Penzig, O. 1921. Pflanzenteratologie. Vols. 1,2. Ed 2. Gebr~der Borntraeger, Berlin. Petri, ?. 1869. In Verhandlungen der Section fur Botanik und Pfianzenphysiologie der 43. Versammlung deutscher Naturforscher und Aerzte zu Innsbruck. Bot. Zeitung 27: 750. Peyritseh, J. 1872. Ueber Bildungsabweichungen bei Cruciferen. Jahrb. Wiss. Bot. 8:117-130, plates 7-9. --.. 1882. Zur Aetiologie der Chloranthien einiger Arabis-Arten. Jahrb. Wiss. Bot. 13: 1-22. Plantefol, L. 1949. Fondements d'une th6orie florale nouvelle: L'ontog6nie de la fleur. Masson, Paris. PInskal, F. S. 1849. Ueber Phytomorphosen. Flora 32: 529-539. - - . 1851. Beitr~ge zur Teratologie und Pathologie der Vegetation. Oesterr. Bot. Woehenbl. 1: 99-101,107-109. Podkolzina, V. E. 1974. Yavlenie mnogostvorchatosti plodov u vidov Brassica. (The phenomenon of polyvalvularity in fruit of Brassica species.) Bot. Zhurn. (Moscow & Leningrad) 59: 688-691. (In Russian.) Polowick, P. L. & V. K. Sawhney. 1986. A scanning electron microscopic study on the initiation and development of floral organs of Brassica napus (cv. Westar). Amer. J. But. 73: 254-263. - & .1987. A scanning electron microscopic study on the influence of temperature on the expression of cytoplasmic male sterility in Brassica napus. Canad. J. But. 65:807-814. - & .1988. High temperature induced male and female sterility in Canola (Brassica napus L.). Ann. But. (London), n.s., 62: 83-86. Potoni6, H. 1892. In Briefkasten. Naturwiss. Wochenschr. 7: 306. Prain, D. 1895. A revision of the genus Chelidonium. Bull. Herb. Boissier 3: 570-587. --.. 1896. Novitiae Indicae, X: Some additional Fumariaceae. J. Asiat. Soc. Bengal, n.s., 65: 10-41. P r a n t l , K. 1891. Berberidaceae; Cruciferae. Pp. 70-77 and 145-206 in A. Engler & K. Prantl (eds.), Die natiirlichen Pflanzenfamilien. Pt. III, sec. 2. Wilhelm Engelmann, Leipzig. Presl, C. B. 1831. Einige Metamorphosen der Blumentheile des Sisymbrium officinale. Linnaea 6: 599-601, plate 9. Proehaska, M. 1930. Beitrag zur Kapsel- und Samenbildung der Papaveraceen mit Ber0cksichtigung der Gartenmohne. Gartenbauwissenschaft 3: 277-284. Pruitt, R. E., C. Chang, P. P.-Y. Pang & E. M. Meyeruwitz. 1987. Molecular genetics and development of Arabidopsis. Pp. 327-338 in W. F. Loomis (ed.), Genetic regulation of development. 45th Symposium of the Society for Developmental Biology, 1986. Alan R. Liss, New York. Puri, V. 1941. Studies in floral anatomy, I: Gynoeceum constitution in the Cruciferae. Proc. Indian Acad. Sci., sec. B, 14: 166-187. ~-. 1945. Studies in floral anatomy, Ill: On the origin and orientation of placental strands. Proc. Natl. Acad. Sci. India, sec. B, 15: 74-91. ~-. 1950. Studies in floral anatomy, VI: Vascular anatomy of the flower of Crataeva religiosa Forst. with special reference to the nature of the carpels in the Capparidaceae. Amer. J. But. 37: 363-370. ~-. 1951. The role of floral anatomy in the solution of morphological problems. But. Rev. (Lancaster) 17: 471-553. ~-. 1952. Placentation in angiosperms. But. Rev. (Lancaster) 18: 603-651. ~-. 1961. The classical concept of angiosperm carpel: A reassessment. J. Indian But. Sue. 40: 511-524. Raghavan, T. S, 1937. Studies in the Capparidaceae, I: The life-history of Cleome chelidonii Linn. f. J. Linn. Sue., But. 51: 43-72. ---. 1939. Studies in the Capparidaceae, II: Floral anatomy and some structural features of the Capparidaceous flower. J. Linn. Sue., But. 52: 239-257. & K. R. Venkatasubban. 1941a. Studies in the Capparidaceae, V: The floral morphology of Crataeva religiosa Forst. Beih. But. Centralbl. 60(A): 388-396. - & .194 lb. Studies in the Capparidaceae, VII: Floral ontogeny and anatomy of Crataeva religiosa with special reference to the morphology of the carpel. Beih. But. Centralbl. 60(A): 397-416.

CARPEL NUMBER

293

Rainio, A. J. 1929. 0her die Intersexualit~t ben der Oattung Papaver. Ann. Soc. Zool. Bot. Fenn. Vanamo 9: 258-285. Rathore, R. K. S. & R. P. Singh. 1968. Embryological studies in Brassica campestris L. var. Yellow Sarson Prain. J. Indian Bot. Soc. 47: 341-349. Reehinger, K. H. 1967. Notizen zur Orient-Flora, No. 91: Cryptocapnos Rech. f., gen. nov. Fumariacearum. Ariz. Oesterr. Akad. Wiss., Math.-Naturwiss. Kl. 104: 418-420. Relssek, S. 1843. Ueber das Wesen der Keimknospe. Linnaea 17: 655-670, plate 20. Retallnek, G. & D. L. Dilcher. 1981. Arguments for a glossopterid ancestry of angiosperms. Paleobiology 7: 54-67. R6bbelen, G. 1965. Flower malformations in mutants as a means of partitioning the developmental process. Arabidopsis Inform. Serv. 2: 12-13. Robinson, B. L. 1896. The fruit of Tropidocarpum. Erythea 4:109-119, plate 3. ~.. 1897. A case of ecblastesis and axial prolification in Lepidium apetalum. Bot. Gaz. 24:209-212, plate 10. Rohweder, O. 1959-1960. fiber verlauhte BlUten von Barbarea vulgaris R. Br. und ihre morphologisehe Bedeutung. Flora 148: 255-282. & P. K. Endress. 1983. Samenpflanzen: Morphologie und Systematik der Angiospermen und Gymnospermen. Georg Thieme Verlag, Stuttgart, New York. Rollins, R. C. 1993. The Crueiferae of continental North America: Systematics of the mustard family from the Arctic to Panama. Stanford University Press, Stanford, Calif. Ronse Deeraene, L. P. & E. F. Smets. 1990. The systematic relationship between Begoniaceae and Papaveraeeae: A comparative study of their floral development. Bull. Jard. Bot. Belg. 60: 229-273. & .1992. An updated interpretation of the androecium of the Fumariaceae. Canad. J. Bot. 70: 1765-1776, & .1993a. D6douhlement revisited: Towards a renewed interpretation of the androeeium of the Magnoliophytina. Bot. J. Linn. Soe. 113: 103-124. & .1993b. The distribution and systematic relevance of the androecial character polymery. Bot. J. Linn. Soc. 113: 285-350. & .1994. Merosity in flowers: Definition, origin, and taxonomic significance. P1. Syst. Evol. 191: 83-104. & .1997a. A floral ontogenetie study of some species of Capparis and Boscia, with special emphasis on the androeeium. Bot. Jahrb. Syst. 119: 231-255. ~ & . .1997b. Evidence for carpel multiplications in the Capparaceae. Belg. Journ. Bot. 130: 59-67. Roth, I. 1957. Die Histogenese der lntegumente yon Capsella bursa-pastoris und ihre morphologische Deutung. Flora 145: 212-235. ~ . 1977. Fruits of angiosperms. In W. Zimmermann, S. Carlquist, P. Ozenda & H. D. Wulff(eds.), Handbuch der Pflanzenanatomie, Special Section, vol. 10, pt. 1. Gebrllder Borntraeger, Berlin, Stuttgart. Ruiz de Clavijo, E. 1994. Heterocarpy and seed polymorphism in Ceratocapnos heterocarpa (Fumariaceae). Int. J. Plant sen. 155: 196-202. Rntishnuser, R. & R. Sattler. 1985. Complementary and heuristic value of contrasting models in structural botany, I: General considerations. Bot. Jahrb. Syst. 107: 415-455. Ryberg, M. 1960. A morphological study of the Fumariaeeae and the taxonomic significance of the characters examined. Aeta Horti Berg. 19: 121-248. Rylski, I. 1986. Pepper (Capsicum). Pp. 341-354 in S. P. Monselise (ed.), CRC Handbook of fruit set and development, CRC Press, Boca Raton, Fla. Sachar, R. C. 1955. The embryology ofArgemone raexicana--A reinvestigation. Phytomorphology 5: 200-218. Sachs, J. 1870. Lehrbuch der Botanik. Wilhelm Engelmann, Leipzig. Saksenll, l-l. B. 1954. Floral morphology and embryology ofFumariaparviflora Lamk. Phytomorphology 4: 409--417. Sampson, D. R. & M. MaeArthnr. 1959. Floral abnormalities in Broccoli. Ann. Bot. (London), n.s., 23: 211-216, plates 1,2.

294

THE BOTANICAL REVIEW

Sands, M. J. S. 1973. New aspects of the floral vascular anatomy in some members of the order Rhoeadales sensu Hutch. Kew Bull. 28:211-256. Sastri, R. L. N. 1969. Floral morphology, embryology, and relationships of the Berberidaceae. Austral. J. Bot. 17: 69-79. Sattler, R. 1973. Organogenesis of flowers: A photographic text-atlas. University of Toronto Press, Toronto. ~.. 1974. A new approach to gynoecial morphology. Phytomorphology 24: 22-34. ---. 1977. KronrOhrenentstehung bei Solanum dulcamara L. und "kongenitale" Verwachsung. Ber. Deutsch. Bot. Ges. 90: 29-38. ~-. 1988. Homeosis in plants. Amer. J. Bot. 75: 1606--1617. --.. 1994. Homology, homeosis, and process morphology in plants. Pp. 423-475 in B. K. Hall (ed.), Homology: The hierarchical basis of comparative biology. Academic Press, New York. - & C. Laeroix. 1988. Development and evolution of basal cauline placentation: Basella rubra. Amer. J. Bot. 75: 918-927. & L. Perlin. 1982. Floral development of Bougainvillea spectabilis Willd., Boerhaavia diffusa L. and Mirabilisjalapa L. (Nyctaginaeeae). Bot. J. Linn. Soc. 84: 161-182. Saunders, E. R. 1923. A reversionary character in the stock (Matthiola incana) and its significance in regard to the structure and evolution of the gynoecium in the Rhoeadales, the Orchidaeeae, and other families. Ann. Bot. (London) 37: 451-482. ---. 1925. On carpel polymorphism, I. Ann. Bot. (London) 39: 123-167. 9 1926. A reply to comments on the theory of the solid carpel and carpel polymorphism. New Phytol. 25: 294-306. ---. 1927. On carpel polymorphism, II. Ann. Bot. (London) 41: 569-627. ~-. 1928a. Illustrations of carpel polymorphism, I. New Phytol. 27: 47-60. ---. 1928b. Illustrations of carpel polymorphism, II. New Phytol. 27: 175-192. ---. 1929a. On a new view of the nature of the median carpels in the Cruciferae. Amer. J. Bot. 16: 122-137. ---. 1929b. A chapter in floral evolution. Rep. Brit. Assoc. Advancem. Sci., South Africa: 389. 9 1930. Illustrations of carpel polymorphism, V. New Phytol. 29: 44-55. - - . 1931. Illustrations of carpel polymorphism, VII. New Phytol. 30:80-118. 9 1932. On some recent contributions and criticisms dealing with morphology in angiosperms. New Phytol. 31: 174-219. - - . 1937. Floral morphology. 2 vols. W. Heifer & Sons, Cambridge, England. Sawhney, V. K. & R. I. Greyson. 1973. Morphogenesis of the stamenless-2 mutant in tomato, II: Modifications of sex organs in the mutant and normal flowers by plant hormones. Canad. J. Bot. 51: 2473-2479. Sawyer, L. 1926. Carpeloid stamens ofPodophyllumpeltatum. Bot. Gaz. 82: 329-332. Scheffer, R. H. C. C. 1869. Over eene monstrositeit der vrucht van Papaver somniferum L. Verslagen Meded. Afd. Natuurk. Kon. Akad. Wetensch., ser. 2, 3: 85-92. Sehilberszky, K. 1893. Neuere Beitr/tge zur Kenntnis anormaler B10tenorgane. Math. Naturwiss. Ber. Ungarn 10: 189-192. Sehimper, C. 1829. I. Correspondenz [aus einem Schreiben des Hm. Carl Schimper in Miinchen an Dr. Hoppe in Regensburg]. Flora 12: 417--431,433-448. - - . 1843. Ueber den Bau der Cruciferenbliithe. Mtm. Congr. Sci. France, X (Strasbourg, 1843), vol. 2: 62-65. Sehlechtendal, D. F. L. v. 1845, Missbildungen und Frucht bei Papaver. Bot. Zeitung 3: 6-9. Sehleiden, M. J. 1839. Sur la signification morphologique du placentaire. Ann. Sci. Nat., Bot., s~. 2, 12: 373-376. - - . 1843. Grundztige der wissenschaftlichen Botanik, 2: Theil: Morphologie, Organologie. Verlag Wilhelm Engelmann, Leipzig. Sehlotterbeek, P. J. 1755. Schediasma botanicum de monstris plantarum. Acta Helv. Phys.-Math. 2:1-14. Schmid, R. 1972. Floral bundle fusion and vascular conservatism. Taxon 21: 429-446. - - . 1977. Edith R. Saunders and floral anatomy: Bibliography and index to families she studied. Bot. J. Linn. Soc. 74: 179-187.

CARPEL NUMBER

295

S. Carlquist, L. D. Hufford & G. L. Webster. 1984. Systematic anatomy of Oceanopapaver, a monotypic genus of the Capparaceae from New Caledonia. Bot. J. Linn. Soc. 89:119-152. Schmidt, H. 1911. Teratologische Beobachtungen an einigen einheimischen Pflanzen. Beih. Bot. Centralbl. 28: 301-328. Schmitz, F. 1878. Die Familiendiagramme der Rhoeadinen. Abh. Naturf. Ges. Halle 14: 1-140. Schneider, S. 1935. Untersuchungen fiber die Samenschleudermechanismen verschiedener Rhoeadales. Jahrb. Wiss. Bot. 81: 663-704. Schnizlein, A. 1843-1870. Iconographia familiarum naturalium regni vegetabilis. Vol. 3. Max Cohen & Sohn, Bonn. Schubert, R., K. Werner & H. Mensel (eds.). 1994. Wemer Rothmaler, Exkursionsflora von Deutschland. Vol. 2, GePa~pflanzen. Ed. 15. Volk und Wissen Verlag GmbH, Berlin. Schulz, O. E. 1936. Cruciferae. Pp. 227-658 in A. Engler & H. Harms (eds.), Die Natfirlichen Pflanzenfamilien. Ed. 2. Vol. 17b. Wilhelm Engelmann, Leipzig. Schumann, K. 1890. Neue Untersuchungen fiber den Blfithenanschluss. Wilhelm Engelmann, Leipzig. Sehwarzbach, A. E. & J. W. Kadereit. 1995. Rapid radiation of North American desert genera of the Papaveraceae: Evidence from restriction site mapping of PCR-amplified chloroplast DNA fragments. Pp. 159-170 in U. Jensen & J. W. Kadereit (eds.), Systematics and evolution of the Ranunculiflorae. PI. Syst. Evol., Suppl. 9. Springer-Verlag, Vienna, New York. Senghas, K. & S. Seybold (eds.). 1993. Schmeil/Fitschen, Flora von Deutschland und angrenzender L~inder. Ed. 89. Quelle & Meyer Verlag, Heidelberg, Wiesbaden. Seringe, N. C. 1830. Notice sur une monstruosit~ du Diplotaxis tenuifolia. Bull. Bot., Geneva 1: 5-10. Sessions, R. A. 1997. Arabidopsis (Brassicaceae) fower development and gynoecium patterning in wild type and ettin mutants. Amer. J. Bot. 84:1179-1191. - & P. C. Zambryski. 1995. Arabidopsis gynoecium structure in the wild type and in ettin mutants. Development 121: 1519-1532. Shneyer, V. S., N. G. Kutyavina & N. S. Morosova. 1995. Serotaxonomical investigation in the Papaverales. Pp. 181-182 in U. Jensen & J. W. Kadereit (eds.), Systematics and evolution of the Ranunculiflorae. PI. Syst. Evol., Suppl. 9. Springer-Verlag, Vienna, New York. Shnll, G. H. 1929. Species hybridizations among old and new species of Shepherd's Purse. Int. Cong. Plant Sci. 1: 837-888. Sieburth, L. E., M. P. Running & E. M. Meyerowitz. 1995. Genetic separation of third and fourth whorl functions ofAGAMOUS. PI. Cell 7: 1249-1258. Singer, W. 1867. Ein Fall yon rfickschreitender Metamorphose, beobachtet an Sisymbrium alliaria Scop. Flora 50: 313-314. Singh, D. & D. S. Negi. 1962. A contribution to the morphology and embryology ofDicentra scandens Walp. Agra Univ. J. Res., Sci. 11: 143-155. Sirks, M. J. 1924. Die gynanthere Form des Goldlacks und ihre Vererbung. Genetica 6: 537-548. Sitte, P., H. Ziegler, F. Ehrendorfer & A. Bresinsky. 1991. Strasburger, Lehrbuch der Botanik. Ed. 33. Gustav Fischer Verlag, Stuttgart, Jena. ---, ~-, - & .1998. Strasburger, Lehrbuch der Botanik. Ed. 34. Gustav Fischer Verlag, Stuttgart, Jena. Smith, A. C. 1972. An appraisal of the orders and families of primitive extant angiosperms. J. Indian Bot. Soc., Golden Jubilee Vol. 50a: 215-226. Snogerup, S. 1967. Studies in the Aegean Flora, VIII: Erysimum sect. Cheiranthus. A.: Taxonomy. Opera Bot. 13: 5-70. Sobick, U. 1983. Blfitenentwicklungsgeschichtliche Untersuchungen an Resedaceen unter besonderer Berficksichtigung von Androeceum und Gynoeceum. Bot. Jahrb. Syst. 104: 203"248. Solms-Laubaeh, H. Grafzn. 1900. Cruciferenstudien, I: Capsella Heegeri Solms, eine neu entstandene Form der deutschen Flora. Bot. Zeitung 58: 167-190. Soltis, D. E., P. S. Soltis, D. L. Niekrent, L. A. Johnson, W. J. Hahn, S. B. Hoot, J. A. Sweere, R. K. Kuzoff, K. A. Kron, M. W. Chase, S. M. Swensen, E. A. Zimmer, S.-M. Chaw, L. J. Gillespie, W. J. Kress & K. J. Sytsma. 1997. Angiosperm phylogeny inferred from 18S ribosomal DNA sequences. Ann. Missouri Bot. Gard. 84: 1-49.

--.,

296

THE BOTANICAL REVIEW

Spinner, H. 1904. Sur des fruits anormaux de Cheiranthus. Bull. Soc. Neuch~teloise Sci. Nat. 32: 140-146. Spjut, R. W. 1994. A systematic treatment of the fruit types. Mere. New York Bot. Gard. 70: 1-182. Spratt, E. R. 1932. The gynoecium of the family Cruciferae. J. Bot. 70: 308-314. Sprotte, K. 1940. Untcrsuchungen ~iber Wachstum und Nervatur der Fruchtbl,,ittcr. Bot. Arch. 40: 463-506. St.-Hilaire, A. de. 1841. Lc@onsde botanique comprenant principalement la morphologie vdgrtale, la tcrminologie, la botanique comparre, l'examen de la valeur des caracthres dans les diverses families naturelles, etc. P.-J. Loss, Paris. Stebbins, G. L. 1974. Flowering plants: Evolution above the species level. Belknap Press of Harvard University Press, Cambridge, Mass. Steinheil, A. 1839. Considdrations sur l'usage que 1'oll peut faire des rapports de position qui existent entre la bractde et les parties de chaque verticille floral, dans la ddtermination du plan normal sur lequel les diffdrentes fleurs sont construites. Ann. Sci. Nat., Bot., sdr. 2, 12:169-211,279-361. Stern, K. R. 1961. Revision ofDicentra (Fumariaceae). Brittonia 13: 1-57. Stevens, F. L. 1894. Botanical notes. Bull. Torrey Bot. Club 21: 489-490. Stopp, K. 1950. Karpologische Studien, I: Vergleicbend-morphologische Untcrsuchungen ~lberdie Dehiszenzformen der Kapselfrilehte; II: Uber "Fensterbildung" an Fruehtfl~lgeln und ~hnliche Erscheinungen. Akad. Wiss. Mainz, Abh. Math.-Naturwiss. KI. 7: 165-218. Stoudt, H. N. 1941. The floral morphology of some of the Cappaddaceac. Amer. J. Bot. 28: 664-675. Sun, G., D. L. Dileher, S. Zheng & Z. Zhou. 1998. In search of the first flower: A Jurassic angiosperm, Archaefruetus, from northeast China. Science 282: 1692-1695. Suringar, W.-F.-R. 1873a. Waarnemingen van eenige plantaardige monstruositeiten. Verslagen Meded. Afd. Natuurk. Kon. Akad. Wetensch., ser. 2, 7: 131-150, plates 1-6. ~.. 1873b. Quelques observations de monstruosit~s vrgrtales. Arch. Nrerl. Sci. Exact. Nat. 8: 436--455, plates 15-20. ~.. 1883. Observations sur une monslruosit~ de Sisymbrium alliaria avee phyllodie des carpelles et des ovules. Assoc. Frang. Avancem. Sci., Compt. Rend. Sess. 1882, 11: 444--449. ~-. 1886. Verslag van de zeven en dertigste vergadering der Nederlandsehe Botanische Vereeniging gehouden te Sittard den 27, Juli 1883. Ned. Kruidk. Arch., ser. 2, 4: 143-145. Takhtajan, A. L. 1959. Die Evolution der Angiospermen. VEB Gustav Fischer Verlag, Jena. 9 1969. Flowering plants, origin and dispersal. Smithsonian Institution Press, Washington, D.C. ~-. 1987. Sistema Magnoliofitov. Nauka, Leningrad. (In Russian.) ~-. 1994. Six new families of fiowering plants. Bot. Zhum. (Moscow & Leningrad) 79: 96--97. Tamura, M. 1974. Phylogeny and classification of the angiosperms. Sanseido, Tokyo. Taylor, D. W. 1991. Angiosperm ovules and carpels: Their characters and polarities, distribution in basal clades, and structural evolution. Postilla 208: 1--40. Terabayashi, S. 1977. Studies in the morphology and systematics of Berberidaceae, I: Floral anatomy of Ranzaniajaponica. Acta Phytotax. Geobot. 28: 45-57. - - . 1978. Studies in the morphology and systematics of Berberidaceae, II: Floral anatomy of Mahoniajaponica (Thunb.) DC. and Berberis thunbergii DC. Acta Phytotax. Geobot. 29:106-118. 9 1979. Studies in the morphology and systematics of Berberidaceae, III: Floral anatomy of Epimedium grandiflorum Morren subsp, sempervirens (Nakai) Kitamura and Vancouveria hexandra (Hooker) Morren et Deeaisne. Aeta Phytotax. Geobot. 30: 153-168. ~ . 1981. Studies in the morphology and systematics of Berberidaceae, IV: Floral anatomy of Plagiorhegma dubia Maxim., Jeffersonia diphylla (L.) Pers. and Achlys triphylla (Smith) DC. ssp. japonica (Maxim.) Kitamura. Bot. Mag. (Tokyo) 94: 141-157. ~ . 1983a. Studies in the morphology and systematics of Berberidaceae, V: Floral anatomy of Caulophyllum Miehx., Leontice L., GymnospermiumSpach and Bongardia Mey. Mem. Fac. Sci. Kyoto Univ., Ser. Biol. 8: 197-217. - - . 1983b. Studies in the morphology and systematics of Berberidaceae, VI: Floral anatomy of Diphylleia Michx., Podophyllum L. and Dysosma Woodson. Acta Phytotax. Geobot. 34: 27-47. ~ 1983c. Studies in the morphology and systematics of Berberidaceae, VII: Floral anatomy of Nandina domestica Thunb. J. Phytogeogr. Taxon. 31:1 6-21.

CARPEL NUMBER

-- 9

297

1985a. The comparative floral anatomy and systematics of the Berberidaceae, I: Morphology. Mem. Fac. Sci. Kyoto Univ., Set. Biol. 10: 73-90. - - . 1985b. The comparative floral anatomy and systematics of the Berberidaceae, II: Systematic considerations. Acta Phytotax. Geobot. 36: 1-13. Thellung, A. 1919.54. Fam. Cruciferae: KreuzbltRler. Pp. 51--482 in G. Hegi (ed.), Illustrierte Flora von Mittel-Europa. Vol. 4, pt. 1. J. F. Lehmanns Verlag, Munich. Thomas, B. A. & R. A. Spicer. 1987. The evolution and paleobiology of land plants. Croom Helm, London, Sydney. Thomas, H. Hamshaw. 1931. The early evolution of the angiosperms. Ann. Bot. (London) 45: 647-672. - - . 1934. The structure and origin of the stigma: A contribution towards a new morphological interpretation of the angiosperm flower. New Phytol. 33: 173-198. 9 1938. Pteridosperm evolution and the Angiospermae. Pp. 1311-1321 in W. J. Jongmans (ed.), Compte rendu II Congr. Avancem. l~tudes Stratigr. Carbonif. 3 (Heerlen 1935). Maestricht. Thompson, J. McLean9 1934. The state of flowering known as angiospermy. Publ. Hartley Bot. Lab. 12: 5-47. - - . 1937. On the place of ontogeny in floral enquiry. Publ. Hartley Bot. Lab. 17: 3-20. Thorne, R. F. 1974. A phylogenetic classification of the Annoniflorae. Aliso 8: 147-209. - - . 1983. Proposed new realignments in the angiosperms. Nordic J. Bot. 3:85-117. - - . 1992a. An updated phylogenetic classification of the flowering plants. Aliso 13: 365-389. - - . 1992b. Classification and geography of the flowering plants. Bot. Rev. (Lancaster) 58: 225348. Tiagl, B. & N. S. Sankhla. 1963. Studies in the family Orobanchaceae, V: A contribution to the embryology of Orobanche lucorum. Bot. Mag. (Tokyo) 76:81-88. TSren, J9 1950. Les caract~res morphologiques, anatomiques et cytologiques de Bongardia chrysogohum Boiss. Rev. Fac. Sci. Univ. Istanbul, s6r. B, 15: 239-263. Townsend, C. C. & E9 Guest (eds.). 1980. Flora oflraq. Vol. 1, pt. 2, Bignoniaceae to Resedaceae. Ministry of Agriculture & Agrarian Reform, Baghdad. Tr~cul, A. 1843. Observations sur les fruits de Prismatocarpus speculum et hybridus et sur celui des Crucif~res. Ann. Sci. Nat., Bot., s~r. 2, 20: 339-344, plate 17. - - . 1873. De la th~orie carpellaire d'apr~s les Papav~ac~es. Compt. Rend. Hebd. S6ances Acad. Sci. 76: 139-145, 181-189, 322-326. - & T. Paty. 1845. Cas rernarquable de t~ratologie v~g6tale. J. Pharm. Chim., s~r. 3, 7:158-159. Treviranus, L9 C. 1847. Einige Bemerkungen 0ber den Fruchtbau der Cruciferen. Bot. Zeitung 5: 409-416, 432-437. Triable, W9 1882. Teratological and other notes. Bull. Torrey Bot. Club 9:140-141. Troll, W. 1928. Zur Auffassung des parakarpen Gynaeceums und des coenokarpen Gynaeceums iiberhaupt. Planta 6: 255-276. - - . 1957. Praktische Einf'tihrung in die Pflanzenmorphologie, pt. 2: Die bl0hende Pfianze. VEB Gustav Fischer Veflag, Jena. Trzaski, L., H. Weiss & A. Szmidt. 1997. Xylem strands distribution in the receptacle of Arabidopsis thaliana (L.) Heynh. Acta Biol. Siles. 31: 17-29. Turpin, P.-J.-F. (ed.). 1837. Oevres d'histoire naturelle de Goethe: Atlas contenant deux planches d'anatomie compar~e, trois de botanique et deux de g~ologie. Abraham Cherbuliez, Paris. Tutin, T. G., N. A. Burges, A. O. Chater, J. R. Edmondson, V. H. Heywood, D. M. Moore, D. H. Valentine, S. M. Waiters & D. A. Webb (eds.). 1993. Flora Europaea. Ed. 2. Vol. 1, Psilotaceae to Platanaceae. Cambridge University Press, Cambridge, England. Van Heel, W. A. 1978. Morphology of the pistil in Malvaceae-Ureneae. Blumea 24: 123-137. 9 1995. Morphology of the gynoecium of Kitaibelia vitifolia Willd. and Malope trifida L. (Malvaceae-Malopeae). Bot. Jahrb. Syst. 117: 485-493. Van Tieghem, P. 1868. Recherches sur la structure du pistil. Ann. Sci. Nat., Bot., s~r. 5, 9: 127-226. 9 1875. Recherches sur la structure du pistil et sur l'anatomie compar~e de la fleur. M6m. Acad. Roy. Sci. Inst. France 21: 1-262. ---. 1886/1888. l~lfiments de botanique. Vols. 1, 2. Ed. 1. Librairie F. Savy, Paris. ---. 1906. l~l~ments de botanique. Vols. 1, 2. Ed. 4. Masson, Paris.

298

THE BOTANICAL REVIEW

Vandendries, R. 1910. Note sur des pistils t~ratologiques chez Cardaminepratensis. Bull. Soc. Roy. Bot. Belgique 47: 351-359. Velenovsk~;', J. 1881. Ueber die vergrfinten Eichen von Alliaria officinalis Andrz. Flora 64: 33-45, plate 1. --.. 1910. Vergleichende Morphologie der Pflanzen. Pt. 3. Vedagsbuchhandlung F. Rivn~c, Prague. Venema, H. J. 1930. Analyse cener monstrositeit van Sisymbrium alliaria Stop. Diss., Leiden. Villani, A. 1902a. Dello stimma e del preteso stilo delle Crocifere. (Nota prima). Malpighia 16: 261-279. --.. 1902b. Dello stimma e del preteso stilo delle Crocifere. (Nora seconda). Malpighia 17:512-527. Vogel, G. 1900. In Bericbt fiber die monaflichen Sitzungen des Preussischen Botanischen Vereins im Winter 1899/1900. Schriften K6nigl. Phys.-0kon. Ges. KOnigsberg 41: 83. Voytenko, V. F. 1968. Formy geterokarpii v semeistve Brassicaceae Burn. i ikh evolutsionnaya otsenka. (The forms ofheterocarpy in the Brassicaceae Burn. family and the evaluation of their evolutionary significance.) Bot. Zhum. (Moscow & Leningrad) 53: 1428-1439. (In Russian; brief summary in English.) Vries, H. de. 1901/1903. Die Mutationstheorie: Versuche und Beobachtungen fiber die Entstehung von Arten im Pflanzenreich. Vols. 1, 2. Veit, Leipzig. 9 1906. Arten und Varietaten und ihre Entstehung dutch Mutation. Translated by H. Klebahn. Gebrflder Borntraeger, Berlin. Vuillemin, P. 1916. Modifications de l'androc~e et extension du pistil dans le genre Papaver. Bull. Soc. Bot. France 63: 164-171. Wartenberg, R. 1886. Durcbwachsene MohnkOpfe. Deutsch. Gart.-Zeitung h 572. Watson, S. 1882. A fruit within a fruit. Gard. Chron., n.s., 17:11. Weber, A. & M. Biswas-Brenner. 1993. Infloreszenz, Blfitenbau und Blfitenentwicklung yon Thottea und Asiphonia piperiformis (Aristolochiaceae). Abstract No. 49 in D. Ffirnkranz & H. Scbantl (eds.), Kurzfassungen XI Symposium Morphologie, Anatomic, Systematik, Salzburg 1993. Weberling, F. 1981. Morphologie der Blfiten und der Blfitenstande. Eugen Ulmer, Stuttgart. Weiss, H. & L. Trzaski. 1997. Vascular system in mature siliques ofArabidopsis thaliana (L.) Heynh. Acta Biol. Siles. 31: 30--48. (In Polish; summaries in English and Russian.) Wendelbo, P. 1974. Fumariaceae. In K. H. Rechinger (ed.), Flora Iranica, No. 110. Akademische I)rucku. Verlagsanstalt, Graz. Werle, G. 1984. Zur Entwicklung des Gynoeceums bei Bergenia cordifolia (Saxifragaceae). P1. Syst. Evol. 147: 167-175. Werle-Sprengel, G. 1993a. Coenokarpie im Spiegel der Apokarpie. Abstract No. 84 in D. Ffirnkranz & H. Schantl (eds.), Kurzfassungen XI Symposium Morphologie, Anatomie, Systematik, Salzburg 1993. - - . 1993b. Apokarpie und Coenokarpie--Kritische Bemerkungen zur "kongenitalen Verwaehsung." Inaugural-Diss., Heidelberg. Wesmael, A. 1861. T~ratologie v~g6tale: Anomalie du Draba verna L. Bull. Soc. Bot. France 8: 393-394. Wettstein, R. 1907. Die Entwicklung der Blfite der angiospermen Pflanzen aus derjenigen der Gyrrmospermen. Wissen ffir alle 45: 705-708. --.. 1935. Handbuch der systematischen Botanik. Ed. 4. Franz Deuticke, Leipzig, Vienna. Weymar, H, 1988. Bueh der Kreuzblfitler. Neumann Verlag, Leipzig-Radebeul. Wiegmann, ?. 1836. Correspondenz. Flora 19: 21-29. Wille, N. 1886. Ueber missgebildete Frflchte bei Capsella bursa-pastoris L. Bot. Centralbl. 26: 121124. Winkler, H. 1940. Zur Einigung und Weiterffibrung in der Frage des Fruchtsystems. Beitr. Biol. Pflanzen 27: 92-130. --.. 1941. Verstehen wir das Gyn6zeum der Angiospermen schon? Beitr. Biol. Pflanzen 27: 242-267. Wittmack, L. 1886. Verbanderte Kohlschoten. Deutsch. Gart.-Zeitung 1: 294-295. 9 1887. Durchwachsener Mobnkopf. Gartenflora 36: 77. Worsdell, W. C. 1915/1916. The principles of plant-teratology. Vols. 1, 2. Ray Society, London.

CARPEL NUMBER

299

Wretschko, M. 1868. Beitrag zur Entwicklungsgeschichtc der Crucifercn-Bltlthe. Sitzungsbcr. Kaisefl. Akad. Wiss. Wien, Math.-Naturwiss. CI. 58:211-226. Wydler, H. 1859. Kleinerc Beitr~lge zur Kcnntnis einheimischer Gew~lchse. Flora 42: 274-287, 289318. ~.. 1861. Beschreibung einiger Blilthen-Antholysen yon ,4lliaria officinalis. Dcnkschr. KOnigl. Bayr. Bot. Ges. Regcnsburg 4: 77-83. Yen, C. 1957. On the structure of the siliques based upon the tcratological material. Acta. Bot. Sin. 6: 259-267. (In Chinese) ~.. 1959. On a new view of carpel morphology in Brassica. Acta Bot. Sin. 8:271-278. (In Chinese; summary in English.) Zimmerli, W. 1973. Neue Untersuchungen Qber die Verwandtschaftsbeziehungen zwischen Cruciferen und Papaveraceae. Inaugural-Diss., Zurich. Zodda, G. 1902. Di alcuni nuovi casi teratologici. Malpighia 17:492-511. Zohary, M. 1948a. Follicular dehiscence in Cruciferae. Lloydia 11: 226-228. ~-. 1948b. Carpological studies in Cruciferae. Palestine J. Bot., Jerusalem 4: 158-165.

300

THE BOTANICAL REVIEW

XI. Appendix 1: Chronological List of Essential Literature Supporting the nI Theory of Carpel Number Papaveraceae Brown, 1817 De Candolle,e 1821b, 1824 Schimper, 1829 Bernhardi, 1833, 1838b Agardh, c 1836 St.-Hilaire, c 1841 Howell, 1842 Le Maout, 1844 Payer, r 1857 Wydler, 1859 Bentham, 1860 Le Sourd-Dussiples & Bergeron, 1861 Mori6re, 1862 Duchartre, 1867 Scheffer, 1869 Godron, 1871-1872 Baillon, 1872 Capus, 1878 Eichler, 1878 Benecke, 1882 Van Tieghem, 1886/1888 Schilberszky, 1893 Prain, 1895 Martel, ~ 1899 Celakovslo), 1899 Fedde, 1909, 1936 Velenovsl~, 1910 Lignier, 1911, 1915 Murbeck, 1912 Bugnon, 1926 Troll, 1928, f 1957 Friedel, 1929 Rainio,g 1929 Eggers, 1935 Arber, 1938 Joshi, 1939 Puri, 1945, 1950 Alexander, 1952 Parkin, 1955 Ernst, 1962a, 1962b, 1 9 6 7 Kapoor & Sharma, 1963 Gu6d6s, 1966c, 1969 Merxmtiller & Leins, 1966, 1967 II'ina, 1968 Zimmerli, 1973 Beck, 1977 Mory, 1979 Jernstedt & Clark, 1979 Cronquist, 1981 Rohweder & Endress, 1 9 8 3 Takhtajan, 1987 Kadereit, 1987, 1993

Bokenfohr & Cass, 1988 Ronse Decraene & Smets, 1990 Karrer, 1991 Endress, 1995 Fumariaceae Bemhardi, 1833, 1838b Agardh, c 1836 Schleiden: 1839 Steinheil, 1839 Gay, 1842 Payer: 1857 Wydler, 1859 Bentham, 1860 Eichler, 1865, 1878 Buchenau, 1866 Duchartre, 1867 Baillon, 1872 Capus, 1878 Benecke, 1882 Van Tieghem, 1886/1888 Chodat, b 1888 Heinricher, 1891 Martel, ~1898, 1902 Murbeck, 1912 Lignier, 1914 Eggers, 1935 Fedde, 1936 Winkler, 1940 Puri, 1945, 1950 Ownbey, 1947 Stopp, 1950 Stern, 1961 Singh & Negi, 1962 Ernst, 1962b Berg, 1969 Zimmerli, 1973 Khfinh, 1973 Cronquist, 1981 Rohweder & Endress, 1 9 8 3 Takhtajan, 1987 Ronse Decraene & Smets, 1990 Capparaeeae De Candolle, 1824 Brown, 1826 Agardh: 1836 Steinheil, 1839 Payer: 1857 Eichler, 1865, 1878 Duchartre, 1867 Baillon, 1872 Benecke, 1882

Van Tieghem, 1886/1888 Chodat, b 1888 Martel, 1902 Murbeck, 1912 Troll, 1928 Eggers, 1935 Pax & Hoffmann, 1936 Raghavan, 1937, 1939 Winkler, 1940 Raghavan & Venkatasubban, 1941a, 1941b Puri, 1945, 1950 Stopp, 1950 Murty, 1953 Narayana, 1962, 1965 Ernst, 1963 Leins & Metzenauer, 1979 Cronquist, 1981 Mehta & Moseley, 1981 Schmid et al., 1984 Takhtajan, 1987 Karrer, 1991 Erbar & Leins, 1997 Ronse Decraene & Smets, 1997a, 1997b

Brassieaeeae De Candolle, b 1813) 1821a, 1821b) 1824 Brown, 1817, 1826 Schimper, b'c 1829, 1843 Seringe, 1830 Presl, 1831 Agardh: 1836 Bernhardi, b 1838a, 1843 Steinheil, 1839 Schleiden, c 1839, 1843 St.-Hilaire: 1841 Howell, 1842 Gay, b 1842 Brogniart, 1844 Treviranus: 1847 Moquin-Tandon & Webb, b 1848, 1849 Payer: 1857 Wydler, b 1859 Chatin, b 1861 Godron, b 1864 Eichler, 1865, 1872, 1878 Leclerc: 1866 Duchartre, 1867 Barnsby, b 1868 Wretschko, 1868 Baillon, 1872 Engler, 1872

CARPEL NUMBER

Xl. Appendix 1 Brassieaceae (continued) Peyritsch, 1872 Braun, 1874 Capus, 1878 Sehmitz,b 1878 Benecke, 1882 Suringar, 1883 Van Tieghem,b 1886/1888 Chodat, b 1888 Prantl, 1891 Celakovslo), b 1894, 1899, 1902 Solms-Laubaeh, 1900 Hannig, 1901 Bush, 1904-1910, 1959k Velenovsk3:', 1910 Murbeek, 1912 Lignier, 1914 Tbellung, 1919 Penzig, 1921 Cejp, 1925 Troll, 1928/1957 Arber, 1931a, 1931b Eggers, 1935 Winkler, 1940, 1941 Kozo-Poljansky,J 1937, 1945 Puri, b 1945, 1950, 1951 Motte,la 1946

301

(continued)

Zohary, 1948b Alexander, 1952 Chadefaud, d 1953, 1954, 1956 Nelson, 1954 Parkin, 1955 Rohweder, 1959-1960 Guyot & Gavaudan, 1961 Guyot,' 1962 Guides, 1964, 1965b, 1966c, 1967, 1979 Fox Maule, 1970 Zimmerli, 1973 Leins & Metzenauer, h 1 9 7 9 Cronquist, 1981 Rohweder & Endress, 1983 Al-Shehbaz, 1984 Takhtajan, 1987 Eyde, 1990 Karrer, 1991 Rollins, 1993 Sessions & Zambryski, 1 9 9 5

Berberidaceae De Candolle, 1824 Lindley, 1847 Payer, c 1857 Wydler, 1859 Baillon, 1861-1862, 1872 Duchartre, 1867 Van Tieghem, 1868, 1875, 1886/1888 Eichler, 1878 Prantl, 1891 Citerne, 1892 Celakovskb, 1899 Eekardt, h 1937 Winkler, 1940 Stopp, 1950 T6ren, 1950 Leinfellner, 1956 Ernst, 1964 Sastri, 1969 Guides, 1977, 1979 Cronquist, 1981 Brett & Posluszny, 1982 De Maggio & Wilson, 1986 Endress, 1989, 1995 Huber, 1991 Loconte, 1993 Feng& Lu, 1998

a The flower is considered a condensed inflorescence. b A tetramerous ancestor is supposed. c The placentae are considered axial structures. a Carpels are interpreted as appendages consisting of several parts. c A cupule-like axial structure is thought to envelop the ovary with only the carpel tips being free. f The paraearpous gynoecium is interpreted as a style that has become fertile with the syncarpous ovary s.str, being extremely reduced. g Carpels are explained as "sexual phyllomes" consisting of female lateral regions and a male median region (anthers are interpreted in an analogous manner). b The carpel number is said to be uncertain. i Two dimerous carpel whorls are assumed, with the aborting median pair belonging to the outer whorl. i The septum is considered a largly receptacular structure. k Variant A of the 2n theory would also be accepted (see section VI.C). I Independent vascular bundles of an outer organ whorl are said to be interspersed between the carpels. mThe carpel number is stated for Hypecoum only.

302

THE BOTANICAL REVIEW

XII. Appendix 2: Chronological List of Literature Dealing with Gynoecial Virescence Papaveraceae Norman, 1857 (Chelidonium majus) Jepson, 1894 (Dendromecon rigida) Joshi, 1933, 1939 (Argemone mexicana)

Celakovsk~, 1875 (Alliariapetiolata), 1876 (Brassicaeeae taxa), 1880 (Hesperis matronalis), 1884 (Raphanus sativus) Cramer, 1879 (Diplotaxis tenuifolia, Sinapis

arvensis) Fumariaceae Kirsehleger, 1854 (Dicentraspectabilis) Eichler, 1865 (Fumariaceae--referring to Kirschleger?)

Capparaceae Gay in Moquin-Tandon, 1842 (Cleomegynan-

dra) Brassicaceae Schimper, 1829 (several taxa) Seringe, 1830 (Diplotaxis tenuifolia) Presl, 1831 (Sisymbrium officinale) Engelmann, 1832 (several taxa) Klinsmann, 1835-1836 (Hesperis matronalis) Reissek, 1843 (Alliariapetiolata) Brogniart, 1844 (Brassica napus) Godron, 1846 ~Cardaminepratensis), 1877, 1878 (several taxa) Griffith, 1847 (Sinapis spp.) Pluskal, 1849 (Alliariapetiolata), 1851 (Bras-

sica oleracea)

Guillard, 1857 (Sinapis arvensis) Wydler, 1861 (Alliariapetiolata) Baillon, 1862-1863 (Sinapis arvensis) Fleischer, 1862 (Brassica napus) Fermond, 1864 (Brassica napus) Singer, 1867 (Alliariapetiolata) Engler, 1872 (Barbarea vulgaris) Peyritsch, 1872 (Arabis alpina and other taxa), 1877 (Alliariapetiolata), 1882 (species of

Arabis) Suringar, 1873a, 1873b (Matthiola incana), 1883 (Alliariapetiolata), 1886 (Barbarea

stricta) Braun, 1874 (Brassicaceae taxa)

Nicotra, 1880 (Biscutella lyrata) Velenovsk~, 1881 (Alliariapetiolata) Hanausek, 1882 (Sinapis arvensis) Chodat, 1888 (Capsella bursa-pastoris) Cuboni, 1889 (Diplotaxis erucoides) Arcangeli, 1894 (Lunaria annua) Mellink, 1895 (Alliariapetiolata) Robinson, 1897 (Lepidium densiflorum) Gerber, 1899a, 1900b (Sisymbrium orientale, Brassica napus var. oleifera) Gagnepain, 1900 (several taxa) Solms-Laubach, 1900 (Capsella heegeri-mutant of C. bursa-pastoris) Massalongo, 1902 (Alliariapetiolata), 1908

(Diplotaxis tenuifolia) Zodda, 1902 (Biscutellalyrata) Marcello, 1903 (Brassica napus) Schmidt, 1911 (Hesperis matronalis) Gallaud, 1926 (Arabis sagittata) Venema, 1930 (.4lliariapetiolata) Khanna, 1931 (Brassica alba) Kozo-Poljansky, 1945 (several taxa) Rohweder, 1959-1960 (Barbarea vulgaris) Yen, 1959 (Brassica napella) Guyot & Gavaudan, 1961 (several taxa) Guyot, 1962 (several taxa) Dupuy & Gu6d6s, 1964 (Raphanus sativus; ovules) Gu6d6s & Dupuy, 1964 (Brassica oleracea; ovules) Gu6d~s, 1964 (several taxa), 1966c (Brassicaceae) Coen & Meyerowitz, 1991 (Arabidopsisthaliana) Bowman, 1994 (Arabidopsis thaliana) Clark & Meyerowitz, 1994 (Arabidopsisthaliana) Sieburth et al., 1995 (Arabidopsis thaliana)

CARPEL NUMBER

303

XIII. Appendix 3: Chronological List of Literature Dealing with Carpelloid Stamens. Papers describing a transformation of the androecial members into diminutive pistils are marked (DP). Papaveraceae Du Petit Thouars, 1820 (Papaver orientale) Brown, 1821 (Papavernudicaule) De Candolle, 1827 (Papaversomniferum) (DP) Sehimper, 1829 (Papaversomniferum) GOppert, 1832, 1850a, 1850b (Papaversomniferura) (DP) Mohl, 1836 (Papaver orientale) Turpin, 1837 (Papaverbracteatum) (DP) Hamburger, 1842 (Papaversomniferum) (DP) Schlechtendal, 1845 (Papaversomniferurn) (DP) Trdcul & Paty, 1845 (Papaverorientale) Gris, 1858 (Macleaya cordata) Lindley, 1859 (Papaversomniferum) (DP) Mori&e, 1859 (Papaversomniferum), 1862 (species ofPapaver) (DP) Groenland, 1860 (Papaversomniferum) (DP) Le Sourd-Dussiples & Bergeron, 1861 (Pa-

paver orientale) Godron, 1871-1872 (species of Papaver) Magnus, 1876, 1877 (Papaversomniferum) Hoffmann, 1877, 1878 (Papaver rhoeas), 1881 (Papaver rhoeas, Papaver somniferum) Pfeiffer in M011er, 1878 (Papaversomniferum) (De) Schilberszky,1893 (Papaver orientale,Papaver rhoeas) (DP) Vogel, 1900 (Papaver braeteatum) (DP) Krausc, 1900 (Papaver somniferum) De 'Cries,1901/1903 (Papaver commutatum, Papaver somniferum), 1906 (Papaversomniferum) (DP) Carano, 1911 (Papaverrhoeas) Lewis, 1912 (Argeraoneplatyceras) Hergt, 1913 (Papaver rhoeas) (DP) Vuillemin, 1916 (Papaverorientale, Papaver rhoeas) (DP) Worsdell, 1916 (Papaver rhoeas) Kajanus, 1919 (Papaversomniferum) (DP) Rainio, 1929 (species of Papaver) Proehaska, 1930 (Papaversomniferum) Sachar, 1955 (Argemone mexicana) Guides, 1969, 1972 (Papaver orientale) Gonnermann, 1979 (Argeraoneochroleuca, Papaver nudicaule), 1980 (Papaver nudi-

caule) B~ekner, 1996a, 1996b (Argemone ochro-

leuca, Papaver nudicaule) Capparaceae Morini, 1891 (Capparis spinosa) Murty, 1953 (Cleome gynandra)

Brassicaeeae Brown, 1821 (Cheiranthus cheiri, Armoracia

rusticana) De Candolle, 1824 (Cheiranthus cheiri) Wichura in Moquin-Tandon, 1842 (Barbarea

vulgaris) Allman, 1851, 1852 (Cheiranthus cheiri) Fertnond, 1855 (Brassica oleracea) Braun, 1851, 1874 (Cheiranthus cheiri) Fournier, 1856a, 1856b (Cheiranthus cheiri) Brogniart, 1861 (Cheiranthus cheiri) Gay in Brogniart, 1861 (Cheiranthus cheiri) Dickie, 1867 (Cheiranthus cheiri) Petri, 1869 (Cheiranthus cheiri) Duchartre, 1870-1871 (Cheiranthus cheiri) Baillon, 1872 (Cheiranthus cheiri) Eichler, 1872 (Cheiranthus cheiri) Marchand, 1896 (Cheiranthus cheiri) Gerber, 1904a, 1905a, 1905b (Cheiranthus

cheiri) Chittenden, 1914 (Cheiranthus cheiri) Worsdell, 1916 (Cheiranthus cheiri) Gudrin, 1924 (Cheiranthus cheir0 Sirks, 1924 (Cheiranthus cheiri) Saunders, 1928a (Cheiranthus cheiri) Nelson, 1929 (Cheiranthus cheiri) Gudd~s, 1964 (Cheiranthus cheiri) RSbbelen, 1965 (Arabidopsis thaliana) Polowick & Sawhney, 1987 (Brassica napus) Pruitt et al., 1987 (Arabidopsis thaliana) Komaki et al., 1988 (Arabidopsis thaliana) Bowman et al., 1989 (Arabidopsis thaliana) Hill & Lord, 1989 (Arabidopsis thaliana) Okada et al., 1989 (Arabidopsis thaliana) Meyerowitz et al., 1989 (Arabidopsis thaliana) Irish & Sussex, 1990 (Arabidopsis thaliana) Bowman et al., 1991 (Arabidopsis thaliana) Coen & Meyerowitz, 1991 (Arabidopsisthaliana) Jack et al., 1992, 1994 (Arabidopsis thaliana) Bowman, 1994 (Arabidopsis thaliana) Clark & Meyerowitz, 1994 (Arabidopsisthaliana) Goto & Meyerowitz, 1994 (Arabidopsisthaliana) Krizek & Meyerowitz, 1996 (Arabidopsis thali-

ana) Berberidaceae Marchand, 1863-1864 (Epimedium grandiflorum) Halsted, 1894 (Podophyllum peltatum) Stevens, 1894 (Podophyllumpeltatum) Sawyer, 1926 (Podophyllumpeltatum) Saunders, 1928b (Mahonia aquifoHum) Baron in Guides, 1972 (Mahonia sp.)

304

THE BOTANICAL REVIEW

XIV. Appendix 4: Chronological List of Literature and Unpublished Observations Concerning Additional Encaptic Carpel Whorls Papaverageae Schimper, 1829 (Papaver somniferum) ROper in De Candolle, 1833 (Papaver somni-

ferum) Wiegmann, 1836 (Papaver somniferum) Lankester, 1848, 1849 (Papaver somniferum) Clos, 1862 (Papaver somniferum) Koch, 1869 (Papaver spp.) Scheffer, 1869 (Papaver somniferum) Braun in Magnus, 1876 (Papaver somniferum) Wartenberg, 1886 (Papaver somniferum) Wittmack, 1887 (Papaver somniferum) Fanta, 1894 (Papaver somniferum) Gonnermann, 1979, 1980 (Dicranostigma erec-

turn) Bokenfohr & Cass, 1988 (Dicranostigma erec-

turn) Teppner, unpubl. (Dicranostigma erectum)

Fumariaceae BriJckner, unpubl. ( Corydalis turtschaninovii)

Brassicaceae Fermond, 1864 (Brassica napus) Dickie, 1867 (Cheiranthus cheiri)

Engler, 1872 (Barbarea vulgaris) Duthie, 1882 (Brassica sp.) Watson, 1882 (Tropidocarpum sp.) Cuboni, 1889 (Diplotaxis erucoides) Robinson, 1896 (Tropidocarpum capparideum) Gerber, 1904b, 1904c, 1904d, 1905d (Lepidium

villarsii) Vandendries, 1910 (Cardaminepratensis) Baccarini, 1918 (Aethionema saxatile) Saunders, 1928b (Capsella viguieri--form of

C. bursa-pastoris) Arber, 193lb (Capsella viguieri--form of C.

bursa-pastoris) Gazet du Chatelier, 1946 (Diplotaxis tenuifolia) Yen, 1957 (Brassica oleracea var. botrytis) Fox Maule, 1970 (Sinapis alba) Clark & Meyerowitz, 1994 (Arabidopsis thali-

Grid) Capparaceae Payer, 1857 (Cleome spinosa) Eichler, 1865 (Cleome spinosa) Morini, 1891 (Capparis spinosa) Narayana, 1965 ( Cadaba indica)

CARPEL NUMBER

305

XV. Appendix 5: Chronological List of Literature and Unpublished Observations Dealing with Increased Carpel Numbers in the Gynoecial Whorl Papaveraceae Jepson, 1894 (Eschscholzia crocea) Gilnther, 1975 (Dicranostigma erectum ) Beck, 1977 ( Chelidonium majus) Gonnermann, 1979, 1980 (Dicranostigma erec-

turn) Bokenfohr & Cass, 1988 (Dicranostigma erectum) Brilekner, unpubl. (Macleaya microcarpa)

Fumariaeeae Brilckner, unpubl. ( Corydalis gotlandica, C.

lineariloba, C. flexuosa) Capparaceae Eiehler, 1865 ( Cleome spinosa), 1872 ( Cleome

gynandra) Raghavan, 1939 ( Cleome gynandra) Raghavan & Venkatasubban, 1941a (Crateva

religiosa) Purl, 1951, 1952 (Crateva adansonii--"C, re-

ligiosa") Murty, 1953 ( Cleome gynandra) Narayana, 1965 ( Cadaba indica) B~ekner, unpubl. (Polanisia dodecandra)

Berberidaeeae Trimble, 1882 (Podophyllum peltatum) Citerne, 1892 (Nandina domestica, Podophyllure spp.) Saunders, 1928b (Nandina domestica) Chapman, 1936 (Epimedium sp.) Eckardt, 1937 (Nandina domestica) Leinfellner, 1956 (Berberis vulgaris) Kaute, 1963 (several taxa)

Brassicaceae Schlotterbeck, 1755 ( Cheiranthus cheiri) Schimper, 1829 (several taxa) Bernhardi, 1838a (several taxa) Braun, 1841 (Lepidium sativum ) De Candolle & De Candolle, 1841 (Lepidium

_sativum) Moquin-Tandon, 1842 (Iberis spp.) Schnizlein, 1843-1870 (Raphanus sp., Diplotaxis sp.) Wesmael, 1861 (Erophila verna) Godron, 1864, 1874 (several taxa)

Eichler, 1865 (Draba kusnetsovii), 1872 (Bras-

sica napus) Crepin, 1866 (Lunaria annua) Barnsby, 1868 (Raphanus caudatus ) Buchenau, 1871 (Brassica sp.) Peyritseh, 1872 (Arabis alpina) Moore, 1875 (Megacarpaea sp.) Borb~s, 1879 (Rorippa sp.) Duthie, 1882 (Brassica sp.) Watson, 1882 (Tropidocarpum sp.) Kronfeld, 1886 (Lunaria annua) Wille, 1886 ( Capsella bursa-pastoris) Wittmack, 1886 (Brassica spp.) Camus, 1888 ( Capsella bursa-pastoris) Potoni~, 1892 (Diplotaxis tenuifolia) Robinson, 1896 ( Tropidocarpum capparideum ) Chodat & Lendner, 1897 (Cheiranthus cheiri) Gerber, 1899c, 1899d (Rorippa barbareifolia), 1904b, 1904c, 1904d, 1905d (Lepidium villar-

sii) Krause, 1900 ( Cheiranthus cheiri) Solms-Laubach, 1900 (Rorippa barbareifolia,

Draba kusnetsovii) Blaringhem & Viguier, 1910 (Capsella viguieri--form of C. bursa-pastoris) Blaringhem, 1910 (Capsella viguieri--form of

C. bursa-pastoris) Monnet, 1913 (Brassica oleracea ) Saunders, 1923 (Matthiola incana), 1928b ( Capsella viguieri--form of C. bursa-

pastoris) Shull, 1929 ( Capsella bursa-pastoris) Arber, 193 lb (Capsella viguieri--form of C.

bursa-pastoris) Yen, 1957 ($rassiea oleracea vat. botrytis) Rathore & Singh, 1968 (Brassica campestris) Fox Maule, 1970 (Sinapis alba) Podkolzina, 1974 (species of Brassica) Kadkol et al., 1986 (Brassica campestris) Pruitt et al., 1987 (Arabidopsis thaliana) Komaki et al., 1988 (Arabidopsis thaliana) Okada et al., 1989 (Arabidopsis thaliana) Eyde, 1990 (Rorippa barbareifolia) Gladis & Hammer, 1992 (Brassica rapa) Bowman, 1994 (Arabidopsis thaliana) Br~lckner, unpubl. ( Cheiranthus cheiri) Teppner, unpubl. ( Capsella bursa-pastoris)

THE BOTANICAL REVIEW

306

Pe3~oMe

];once 170 HCT npo)~on}KacTc~ KHCKyCCHH0 CTpOCHHHcTpyqKa. ~TOT n n o ~ THnHqCH RH~ CCMCI~CTBa KpCCTOI~BCTHbIX ( B r a s s i c a c e a e ) ~OpMe) ~

, a TaKAKr (B MO~M~HI~HpOBaHHOI~I

HCKOTOpbIX IIpe~CTaBHTCaCH KaHcpCOBblX (Capparaceae)

(Papaveraceae) (Berberidaceae)

H ~IMffHKOBBIX ( F u m a r i a c e a e ) . BCTpCqar

, MaKOBbIX

IIocKOnbKF y 6ap6apHcoBbIX

TH~ FJIo~a, KOTOpbIII NOXO)K Ha "no~ycTpyHOK",

I~IHeI~CH 9TOIX) CCMCI~ICTBaTaK~Kr o 6 c y ~ a ~ T C ~ B CBff3H CO CTpyKTypOI~ITHIIHtlHOI~) cTpy~Ka. CyIIXCCrByIOTpa3HHxIH~r TCOpHH 0 HHCJICIIJIO~OJIHCTHKOBB cTpyHKr ZOTOpbI~I B THIIHtIHOMCYI~IaC C0CTOHTH3 ~BFX K/IaIXCHT, o6pa3y~n~x paMKy H CB~I3aHI~IX ncpcr0po~KOR, a TaX~r H3 ~ByX CTCpHJI~H~IX CTBOpOK,O T ~ I K I I X C ~ I npH pacKpbIBaHHH 0T paMKH, HCCy]~CII CCMCHa. (B TCplKIIHax 0~HOI~ TcopHH M 0 ~ O 06~ICHHT~ CTpOCHIIC M0~H~bHII~pOBaHH~IX cTpyzIKOBKaK 6C3 ncpcropoKK~, c OKHOfi~ H ~4HOr~MH CTB0pKa_t4H~ TaK H HCBCKpbIBaIO~CSI IIJIO~OB.) I~pHpo~a H 9BOJIIOIXH~IIJIO~OJIHCTHKay IIOKpbITOCCMCWR%IXr

HC~OCTaTOHItOH3y~ICHbI,H I~09TOMyylvI03pHTC3IbHbI.HCCMOTp~I

Ha 9TO, pa3~IH~bIC TC0pHH l~pc~iio3iaraiOT, HTO ~ByXCTBopHaTbII~CTpyHoK COCTOHTH3 ~ByX, ~ICTblpeXHJIH I~CCTH IIJIO~OJIHCTHKOB.KpOMC T01X), cylI~CCTByCTTOHKa 3petIH~, HTO H~0)XO~HCrHKH H0~H0CT~IO pcKy~HpoBaH~ H I~E0~0~ I~p0K30me~ H3 KBCTO~O}Ka. B 3aBHCHMOCTH OT CIIOC06a O~s

HOJI0}KCHH~I,~)opl~I, CTepI~bH0CTH HJIH

~CpTH~IbHOCTH FHHOTCTHHCCKHXIIJIO~OJIHCTHKOB,TCOpHH ItMCIOTpa3JIHHHI~ICBapHaHT~I. B pa60TC IIpHBOKHTC~0630p cyII~CCTByIOIKHXTC0pHH H HX KpHTHHCCKHI~I&HaJIH3. ~JL~ apITMCHTaILHH IIOCTOs

HCIIOJI]~SyK)TCYIC~yIOIKHCIIpH3HaKH ITIHCI~C2["C1~OpMyp~I3IbI~a~

xapaKTep paCKpblBaHHYIcTpyHKa, CTpOCHHC IIYIaKCHTapHbIX06YlaCTCI~I,aHaTOMHIO HpOBO)L~IKCI~ICHCTCMbI, 3a~I0}KCHHC I~IHCIXC~IH TcpaTo~IOI~IqCCICHCH3MCHCHH~I. BCC ~TH

CARPEL NUMBER HpH3HaKH cpaBHHBaIOTC~ y Hpr162

307

Ka~J~OFO COMeI:iCTBa, a apryMOHTa~Hff

Ka~K~O~[ H3 TOOpHH B3BCIIIHBaCTC~H HpoBcp~ICTC~I. B pc3yHbTaTC HCCJ]C~0BaHH~/ BblSICHCHO, tlTO 60HOe COBpCM{~HHbIr TeOpHH HC IIpHBO~EI" BCCOMbIX apryMCHTOB IIpOTHB IoIaccHHcCKOI~TcopPIH Po6cpTa ]~poyHa ( B r o w n ,

1817),cor~acHo

KOTOpOI~

cTpyqOK o6pfl3OBaH ]~ByM~ ]laTepZJIbHO pacIIOHO)KCHHBIMHIUIO]~OJIHCTHKaMH,paMKa ~IBJIHCTC~ IIpo]/~YKTOMCJIH~IHH~[HX KpaCB, ncpcropo]~Ka ~BngCTC~ nOYKHO~H 06pa30BaHa B pc3yJIBTaTr cpaCTaHH~ ]~ByXBBIpOCTOBIIJIaI~CHT, a CTBOpKH - ~TO JIHUlb KpoIoIKHr o6pa3OBaHH~. CpaBHHTe~BHbIH aHaJIH3 npH3HaKOB BbI~BHJI, ~ITOJIOIIaCTH pblJlbL~a HC BO BCeX cJIyHa~x COOTBr

KOHHHKaMILrIO~OJIHCTHKOB,HX HHCJIO HC Bccr]~a COBna~acl

c HHCJIOM II/IO~OHHCTHKOB.IHBLI, IIO KOTOpbIM BCKpbIBaK)TC~cTpyHKH, ~[B~I~IIOTC~[Hr

rpaHn~aM~ nnOnOa~CTHKOB, a BTOpH~IHMMHOT~eJIHTr

TKaIL~MH BHyTpH

nnaCTHHK~ n~OnO~mCTHKOB.MaCCHBH~IeRna~eHTapH~IC paI:IOH~I CO CYIOHC~I~IM H~L~IKOBaHHCM He 06~[3aTeJ~bn0 ~ 0 n ~ H ~ 6blTb II~O~OJIHCTHKaMH3aKpbITOr0 THHa. ~IHcJIO H XO~ IIpOBOKKIKHXIIFqKOB60JIr162H0~IHHO HHTCpIIpCTHpOBaTbB TCpMHHaX 3aYlo~KCHI~ H qbyHKI~H opraHOB, Te0pH~I 06 9BO~IIOKHOHHOMKOHCepBaTH3MCIIpOBO~qIKeI~I CHCT~MbI HCO6OCHOBaHa. 30H~I p0cTa Ha KOJIbIXeBOM3a~aTKr rHHeKC~ HC BCeI'~a M0~KHO C~IHTaTbIIpHMop~H.~H IIYIOKOJIHCTHKOB.TCpaT~ H~OKOB (HaIIpHMCp, ~eT~peXCTBOp~aT~e crpy~m) He HB~D/IOTC~IaTaBHZMOM. TaKHM 06pa3oM, ~IHCYIO /IJIO~OJIMG~IKOB H~BO3MOHQ/0~OCTOB~pHOoIIp~fl~eJIHTb"OHO 6o~t~mc, ~CM ~HC~O IL~a~CHT B I'HHCI~Cr I'HHeKC~ 6ap6apHc0B~IX ~r

H~ea~ma~

CTpyKTypa rHHCI~C~y n p c ~ c T a ~ e ~ i

O~HaKOBOrO O 6 ~ c H e m ~

~

MOHOMCpHbII~I.

o6cyx~aeM~iX CeMCHCTBTpe5yCT

p a ~ a x O~HO~ ~ TO~ ~ e TeOp~m. Heo6xo~m~a

IIepepaSoTxa MIIorHx y~e6HIlKOB, qb~IOpHCTII~IeCKHXH ~[~pyrHx 063OpHI,IX CBO~OKB npcAJIO)KCHHOM aCHeKTC.