SelbY!ln!l 7: 129-247

THE RELATIONSHIPS OF THE SLIPPER ORCHIDS (SUBFAMILY CYPRIPEDIOIDEAE, ORCHIDACEAE) John T. A twood, Jr. *

ABSTRACT The phylogenetic relationships of the Cypripedioideae are adumbrated by the Wagner Groundplan-divergence method. The mutual resemblance be­ tween paleotropical Paphiopedilum and neotropicalPhragmipedium is doubt­ fully superficial since both share several derived vegetative character states and similar floral bud development. For these genera a simplified classifi­ cation is proposed, and one new taxonomic combination is provided. Selenipedium grades into Cypripedium via C. irapeanum and C. caUtor­ nicum. The correlation of reduced vegetative features with increasing lati­ tude suggests that temperate climates have altered the northern taxa most. By eliminating these features from cladistic analysis, C. arietinum (for which the genus Criosanthes is recommended) is parsimoniously distinct from the 3e/enipedium-Cypripedium clade. Similar flowers of Phragmipedium schlimii and Paphiopedilum subgenus Brachypetalum relate to neoteny, since the floral parts resemble those of young flower buds of related species. Their superficial floral similarity is in­ terpreted as a convergence. Chromosome numbers are reported for five species of Phragmipedium and nine species of Paphiopedilum. Since the basal chromosome number is interpreted as 2n = 20 for Cypripedium, Criosanthes, and Phragmipedium, this is probably the basal number for the subfamily. Although centric fission accounts for the upward aneuploid series in Paphiopedilum (2n = 26-44), higher metacentric arm ratios in the aneuploids than in the non-aneuploids suggest that centric fusion has occurred as well. It is argued that both fission and fusion have karyologically repatterned Paphiopedilum section Barbata with concurrent adaptation to florest floors. Under a vicariance model, the cladograms reveal geographic patterns in Paphiopedilum, Phragmipedium, and the SeZenipedium-Cypripedium clade with the primitive taxa southernmost. To explain the seemingly incongruous patterns with a necessary Laurasian origin, it is suggested that the southern­ most populations have followed habitats similar to the ancestral ones in re­ sponse to post-Miocene cooling. This mechanism may offer a partial explana­ tion for accumulation of primitive angiosperms on the Asiatic islands which have largely arisen since the Miocene. Contrary to popular belief, some slipper orchids are very advanced, and there is little evidence that the Cypripedioideae is a relic group at an evolu­ tionary dead end. CHAPTER I: INTRODUCTION The Cypripedioideae is a subfamily of the Orchidaceae known com­ monlyas the ladyslippers (various spellings) or slipper orchids. The subfamily is easily recognized by its unique flower (Fig. 1.1) which has a synsepal, a saccate labellum, a conspicuous staminode, and two fertile stamens. The slip­ per orchids are presumed to be most closely related to the subfamily Apostasi­ oideae, but the latter occasionally has three fertile stamens and lacks synse­ pals and saccate labella.

* The Marie Selby Botanical Gardens, 811 S. Pa.hn Ave., Sarasota, FL 33577 129 130 SELBYANA [Vol 7

There are many popular and scientific accounts of slipper orchids; h w­ ever, no recent investigation has been made into relationships which m· ht elucidate evolutionary patterns within the Cypripedioideae. It is hoped t at the present work will help fill the void, and that it will provide a framew rk for understanding the subfamilial diversity. Most of the data have been 01- lected from nursery-grown plants, but a few wild populations have been ex­ amined so far as funds have permitted. Supplementary data have also been taken from literature sources, herbarium sheets, and occasionally from photo­ graphs. Because live material of Asiatic Cypripedium has been unavailable, . detailed taxonomic considerations within the plicate-leaved taxa are limited, ' but generic and infrageneric relationships are detailed among the condupli­ cate-leaved genera.

THE GENERA There are two growth patterns in the slipper orchids (Fig. 1.2 and 1.3): (1) some species have thin, plicate leaves distributed on elongate stems (Cy­ pripedium and Selenipedium), and (2) some pave thick, coriaceous leaves forming basal, distichous rosettes (Phragmipedium and Paphiopedilum). There are several floral differences among :the genera. Selenipedium and Cypripedium have persistent perianths after anthesis, while the remaining gen­ era have deciduous perianths. Sepal vernatiop is perforate in the plicate­ leaved genera, valvate in Phragmipedium, and imbricate in Paphiopedilum. Selenipedium and Phragmipedium have axile placentation, and Cypripedium and Paphiopedilum have parietal placentation. Slipper orchids are found over much of the world but are notably absent from Africa and Australia. Cypripedium is temperate with a circum boreal distribution. In the Old World it ranges from England across Siberia, to southern China and southern Europe. In the New World it occurs over most of the North American continent south to Guatemala but is absent from ex­ treme southeastern United States. The genus is perhaps best developed in China with a second center of diversity in eastern North America. Selenipedi­ um is restricted to the isthmus of Panama and northern South America. Phragmipedium ranges from southern Mexico to tropical South America, but the greatest diversity is concentrated in Andean South America from Colom­ bia to Bolivia and Peru. Paphiopedilum has a complex range in the Old World tropics. It occurs from Hongkong west to Sikkim, south to Sumatra, east to New Guinea, and Bougainville. There exists a disjunct species in the Traven­ core Hills of southern India, but the greatest development of species is found on the southeast Asian islands, especially Borneo. Due to lack of revisionary work (especially with the Asiatic populations), the actual number of slipper orchid species is difficult to estimate. Both nar­ row and broad species concepts can often be applied to single populations with seemingly equal justification. In estimating species number it is best to give both conservative and liberal estimates based on available species descrip­ tions and names provided in Index Kewensis. Cypripedium may contain only 30 or as many as 50 species. The second estimate may be somewhat inflated by available names, but the opportunities for endemism and isolation in the mountains of China provide some justifica­ tion for this unusually high estimate. Only 6 species of Selenipedium have been described, and some may eventually be relegated to subspecific status. The species are not diverse in plant habit or in floral morphology relative to the other genera. Phragmipedium contains a minimum of 10 species, although 1984] ATWOOD: CYPRIPEDIOIDEAE 131

e fig. 1.2 Fig. 1.1 ·.\..---a b

Fig. 1.3 Fig. 1.1 A generalized flower showing salient features: a. bract, b. synsepal, c. dorsal sepal, d. lateral petal with counter-clockwise spiral, e. lip or labellum, f. staminode. g. anther. Fig. 1.2 . A schematic representation of a plicate-leaved slipper orchid with spirally ar­ ranged leaves on an elongate stem: a. blade, b. sheath. Figure 1.3 A schematic representation of a conduplicate-Ieaved slipper orchid with leaves arranged in a distichous rosette. 132 SELBY ANA [Vol. 7

Garay (1979) has recognized 21. However, emphasis on singular characters (e.g., staminode shape) has undoubtedly led to taxonomic inflation. Paphio­ pedilum has a minimum of 60 species and a maximum of 80 species. Until extensive field work can be undertaken, firm bases for species recognition in Paphiopedilum will be sought in vain. There are 36 to 56 species among the plicate-leaved genera, and 70 to 101 conduplicate-Ieaved species. Altogether there exist between 106 and 157 slipper orchid species.

HABITATS Slipper orchids are found in various habitats, although often absent from areas that would appear ideal. In eastern Vermont Cypripedium pubescens is found in calcareous bogs and experiences few effects ~om drought. Although the orchids may be locally common, their habitats in eastern Vermont are not. Bordering Lake Champlain in northern New York, the same species may be found on moist hillsides underlain with limestone bedrock. In this region the species may be found also in abandoned pastures with C. arietinum and C. reginae. Cypripedium pubescens in Michigan may be found not only in these habitats but also along roadsides as "weeds" where the roadbeds are constructed from crushed limestone. Occasionally C. pubescens may be found in full flower in lawns where it has become established in the lime­ stone substrate. Cypripedium acaule prefers acidic habitats with pines, oaks, and ericaceous shrubs, especially in areas which are semishaded. In Cape Anne, Massachusetts, I have observed that a June forest fire seemed to en­ courage the growth and development of C. acaule in following years. I have also observed C. acaule growing in the cracks of a highway within 4 years after paving. Cypripedium arietinum, although usually uncommon in New England, occurs abundantly along the northern shoreline of Lake Michigan in a lakeside zone disturbed by spring ice accumulation. Cypripedium candi­ dum prefers open prairies that are periodically burned. Little is known of the preferred habitats of Selenipedium. Dunsterville and Garay (1963) note that S. steyermarkii grows among "deep grass and head-high shrubs." Dunsterville and Dunsterville (1980) note that the same species may be found growing "among rocks and fairly open scrubby trees" as well as on "densely-forested slopes." I have observed S. chica growing among palms and bambusoid grasses which it resembled. The small size of nearby trees suggests that the plants were established in second growth. The generally poor health of the few plants known from this station is probably attributed to relatively recent closing of the surrounding canopy, and I sug­ gest that S. chica is a colonizer of open habitats, not of forested environments. Phragmipedium is rather diverse in its habitat preferences. Phragmiped­ ium warscewiczianum is an epiphyte but luxuriates on dead trees in open sun and occasionally may be found on fence posts. Phragmipedium longifol­ ium occurs on disturbed embankments above the high watermark of rivers, but they seem equally well adapted for man-made roadcuts or open pastures. A small-growing ecotype of P. longifolium grows on river rocks in Panama (Atwood and Dressler, in prep.) as do P. klotschianum (Dunsterville, 1970; Gorinsky, 1972), P. pearcei, and P. caricinum (Dodson, personal communica­ tion). I have no experience with habitats ofPaphiopedilum, but many species appear to occur on cliffs and in tree crowns, although most of the mottled­ leaved species occur on forest floors (see references cited in Atwood and Williams, 1979). 1984] ATWOOD: CYPRIPEDIOIDEAE 133

From these observations and literature sources it appears that-the slip­ per orchids occur in disturbed habitats and are colonizers. The only true den­ izens of the virgin forest floor may be the mottled-leaved species of Paphio­ pedilum subgenus Barbata (Krzl.) Brieger.

BREEDING RELATIONSHIPS Breeding behavior is well-known for the conduplicate-Ieaved genera hut not for the plicate-leaved genera. According to Sanders' lists (1947), essenti­ ally any two species of Paphiopedilum may be hybridized, and many primary hybrids between very dissimilar parents have yielded progeny. Nothing is known of breeding behavior in Selenipedium. Within Cypripedium a few nat­ ural hybrids are known, mostly among the C. calceolus complex. Cypripedium pubescens hybridizes with C. candidum to form Cypripedium X andrewsii, and C. parviflorum and C. candidum hybridize to form Cypripedium X {auii­ leanum (Luer, 1975), Hybrids between more remotely related species are rare. In the Gardener's Chronicle (1892) there is a report of a hybrid between C. calceolus and C. macranthum. I have attempted to cross C. reginae with C. pubescens but the ovary aborted. Although/ intergeneric hybridization between the plicate-leaved genera is unknown, there have been several attempts at hybridizing Paphiopedilum and Phragmipedtum. Rolfe (1888) reported healthy seedlings from a cross between Phragmipedium cauda tum and Paphiopedilum barbatum, but I can­ not locate subsequent reports of flowering if such exist. Rolfe also mentioned futile attempts to obtain seed from crossing Cypripedium reginae and the Indian conduplicate-Ieaved species. Swan (1894) claimed to have recovered seedlings from crossing PaphiopedUum rothschildianum with Phragmipedium X leucorrhodum, and also Paphiopedilum parishii with Phragmipedium cari­ cinum. Plants of the latter hybrid were reported to have remained small after several years. Rolfe (1894b, 1895) mentioned hybrids between Paphiopedi­ lum lawrenceanum and Phragmipedium X sedenii but the flowering progeny were similar to the pistillate parent. He further reports that a Mr. Statter ob·· tained progeny from Phragmipedium X sedenii crossed with Paphiopedilum stonei (pistillate parent?) which resembled the latter species. Swinburne (1896) claimed he had produced plants from crossing Phragmipedium schlimii with Paphiopedilum spicerianum. He also reported a hybrid between a Ph rag­ mipedium hybrid andPaphiopedilum chamberlainianum forming fast-growing progeny resembling the pistillate parent. Hurst (1925) reported production of healthy seeds in a cross between Cypripedium calceolus and Phragmiped­ tum X sedenii, but no seedlings are known to have germinated. No reports of actual flowering of the above hybrids followed, suggesting that either the flowers were clearly like the pistillate parent (parthenogenetic origin) and the mistake never admitted, or the plants never matured to flowering. The first description of a putative intergeneric hyhrid flower was reported in 1913 (Anonymous) but no photograph was included. A hybrid between Paphio­ pedilum X harrisianum andPhragmipedium sch lim ii was apparently abnormal as the lip was smaller than that of either parent. Wilson (1961) reported the flowering of an intergeneric cross between a Phragmipedium hybrid and a Paphiopedilum hybrid, which resembled the Paphiopedilum parent but was abnormal. Wilson concluded that progeny derived from intergeneric crosses are probably derived parthenogenetically and not from fusion of parental gametes. Despite the similar appearances, sterility barriers between Phragmi~ pedium and Paphiopedilum are strong, probably reflecting geographic sepa­ ration for a very long time. 134 CSELBYANA [Vol. 7

ECONOMIC IMPORTANCE The economic importance of slipper orchids lies in their aesthetic value. Although a few accounts exist documenting medicinal usage, slipper orchids have never been widely accepted as medicinal plants and are rarely used to­ day except in preparation of folk remedies. If they are at all palatable, the slow growth rate would limit culinary usage. Although there is one account of use of capsules as a flavoring, slipper orchids have probably never been used extensively as food. Following is an account of the various reports docu­ menting the economic use of slipper orchids beginning with Gesner (1541 in Zoller, 1976). According to several herbalists, Cypripedium was not originally believed to have medicinal value. Cypripedium may have been unknown to the Greeks on whose works the herbals are based, but C. calceolus has been reported from the mountains of southwestern Greece (Boissier, 1884; Goulandris, Goulimis and Stearn, 1968). Cypripedium calc~olus was the first slipper or­ chid documented by European herbalists under the name Calceolus Mariae. Gesner (Zoller, 1976) included a very accurate illustration of C. ca lceo Ius. L'obel (1576) described C. calceolus and included an accurate if somewhat diagrammatic woodcut. Gerard (1597) indicated that he had obtained "La­ dies Shoo" from an apothecary and apparently grew it as an ornamental. He mentioned that it came from "the mountaines of Germanie, Hungarie, and Poland," apparently ignorant of its occurrence in England. Cypripedium re­ ginae was apparently known by Parkinson (1640) as he mentioned that "A sort thereof hath beene brought from the North parts of America, differing onely in being greater both in stalkes, leaves and flowers, which are not yel­ low but white, with reddish strakes through the bellies of them." As Zoller (1976) has indicated, it is surprising that Cypripedium was not mentioned in the works of Brunfels, Fuchs, Bock, and Matthioli. Use of Cypripedium in medicine by non-Asiatics was first practiced by the American Indians (Rafinesque, 1828), and all species were of presumed equal potency. Indians in New York and New England used the root as seda­ tives, nervines, and antispasmodics. Rafinesque further claimed that roots produced "beneficial effects in all nervous diseases and hysterical affections by allaying pain, quieting the nerves and promoting sleep." The use of ex­ tracts from Cypripedium by European-Americans followed, as Porcher (1863) recommended its use "as a quieting agent in place of paragoric." Johnson (1884) reports that Cypripedium roots contain "a minute quantity of vola­ tile oil, volatile acid, resin, and common vegetable principles." He indicates that the trade name registered with the United States Pharmacopoeia was Extractum cypripedii fluidum or fluid extract of Cypripedium. The use of Cypripedium as a folk medicine seems to have essentially disappeared in this century (Correll, 1950). The Asiatics apparently also used Cypripedium medicinally as Crane (1931) mentions that the Koreans made "A wonderful headache cure" from the roots of C. speciosum. . The only culinary use of slipper orchids seems to have been practiced by Panamanian Indians. Rolfe (1901) reported that the fruits of Seleniped­ ium chica in Panama were used as a flavoring called "vainilla chica" by the Spanish. The slipper orchids have been most useful to man as horticultural sub­ jects. Although Curtis (1792) indicates that Cypripedium reginae was intro­ duced to Kew in 1770, Parkinson's knowledge bf C. reginae suggests that the 1984] ATWOOD: CYPRIPEDIOIDEAE 135 species may have been cultivated in England much earlier. Cypripedium acaule was known in 1700, and is known to have been cultivated in England by 1738 (Catesby, 1771). Cypripedium pubescens was also grown in England by 1738 (Catesby, 1771), and C. macranthum was introduced to the Glasgow Botanical Gardens in 1829 (Curtis and Hooker, 1829). The first conduplicate-leaved species (Paphiopedilum uenustum) reached Europe about 1819 (Veitch, 1889), creating a sensation among horticultur­ ists. The species was closely followed by introduction of Paphiopedilum in­ signe. By mid-nineteenth century the first New World conduplicate-leaved slipper orchid was cultivated. However, it was the period from 1850 to 1900 which saw the greatest introduction of species, most described by H. G. Reichenbach filius. Fabulous prices were paid for species as Paphiopedilum spicerianum once commanded 100 guineas or about $3,000 in modern cur­ rency (Williams and Williams, 1894). Although hybridization began with Paphiopedilum X harrisianum in 1864 (Veitch, 1889), extensive hybridization developed after 1900 when species were losing favor. In the decades which followed many species be­ came lost from cultivation until their rediscovery in recent years. With the rebirth of horticultural interest in species since 1960, wild orchids have been sought worldwide. Although most slipper orchid species are now cultivated, virtually no success has been experienced with Selenipedium, and many of the Chinese Cypripedium species have been inaccessible.

A BRIEF ACCOUNT OF CLASSIFICATIONS AND NOMENCLATURES Classification systems of the Cypripedioideae have been numerous. The following section is an account of the important systems. Linnaeus (1753) recognized only one species and a few varieties which are now given specific status. The species Cypripedium calceolus was placed in Class Gynandria and Order Diandria. Olof Swartz (1800) first recognized the significance of stamen number among orchids and erected the name Diandrum for the slipper orchids. Lind­ ley later used the name Diandrae (Dressler and Dodson, 1960). Rafinesque (1836) recognized Cypripedium as a diverse group and rec­ ommended a division of the genus using the names Sacodon, Stimegas, Cor­ dula, Menophora, and Criosanthes. Unfortunately he confused the taxonomy by indicating these designations as "subgenera or rather Genera," thus the in­ tended taxonomic levels are uncertain. John Lindley (1840) recognized 22 species within the one genus Cypri­ pedium but with several subgeneric categories (see Appendix 1). In 1846 he described Uropedium based on an unusual autogamous plant from South America. Because members related to Uropedium lindenii were later described as or transferred to Phragmipedium, the priority of Uropedium became a source of confusion. In 1854 Reichenbach in recognizing the 3-chambered ovary as a char­ acter of fundamental importance described SeZenipedium, which at first in­ cluded the tropical New World plicate-leaved species and conduplicate­ leaved species. However, Blume (1858) disagreed with Reichenbach's empha­ sis on the one character as he discovered that some Cypripedium species (s. ampl.) are trilocular near the ends of the seed chamber although unilocular in the middle. In Genera Plantarum (Bentham and Hooker, 1883) the genera Cypri­ pedium and SeZenipedium (both sensu Reichenbach) are recognized. 136

Pfitzer (1886) '"'ll'·",.'".,,, with the deciduous tent perianth, He th,m cal conduplieate-le&.\10d in Cypripedium and the (1888) argued in favor species under one genus, made. By 1888 Pritzer LGe'''''''''''''''''' he divided including only the Old the multi-flowered Desbois 1984] ATWOOD: CYPRIPEDIOIDEAE 137

The classifications of Pfitzer (1903) is the most comprehensive treat­ ment available for the slipper orchids (Appendix 7). He recognized four gen­ era, consistent with Rolfe's classification (1896), but retained the -pedilum suffixes and spelled the genus Phragmopedilum according to his 1894 treat­ ment. Incorrect spellings in the following text will be corrected to reduce confusion. Pfitzer's last classification (1903) was unique in that vegetative as well as floral features were emphasized (Appendix 7). Consequently in his key to genera he separates Selenipedium and Cypripedium from Phragmipedium and Paphiopedilum on foliar vernation as well as presence of deciduous peri­ anths. Although he noted that Cypripedium irapeanum is a geographical and morphological link between Cypripedium and Selenipedium, he maintained Selenipedium as distinct from Cypripedium. Like Rolfe (1896), Pfitzer built his classification on an evolutionary framework emphasizingSelenipedium as the most primitive genus on the basis of trilocular ovaries and crustose seeds. In spite of its virtues Pfitzer's classification (1903) is taxonomically in­ flated with numerous subgeneric categories including subgenera, series, sec­ tions, and subsections. This is perhaps accounted for by his analytical ap­ proach with tendency for creating monotypic taxa. Furthermore, complex and inflexible systems were fashionable at the time. Much of the present work will examine the conceptual utility of Pfitzer's classification. By 1912 the orchid-growing public rejected the name Paphiopedilum for plants abundantly grown in conservatories. Rolfe (1912) tried to resurrect the name Cordula of Rafinesque for the tropical Asiatic species and made the appropriate combinations, but the public stubbornly retained Cypripedium. Paphiopedilum was later conserved over Cordula in 1959 (Taxon 8 (7): 242). Schlechter's system (1926) followed that of Pfitzer (1903), and the al­ tered generic spellings of Pfitzer were retained. Schlechter changed the sys­ tem very little but changed the tribal name of Cypripedilinae to Cypripedil­ oideae under subfamily Diandrae. The last and most recent classification is that by Brieger (1971). With some changes this work basically follows Pfitzer (1903). However, the spell­ ings follow the rules of nomenclature. Brieger's system, like Pfitzer's, is taxo­ nomically inflated. For instance, each of the four genera is distributed among four respective tribes and four subtribes. The generic names Cypripedium, Selenipedium, Phragmipedium, and Paphiopedilum are presently stabilized and are usually correctly used by the orchidist. Appendix 9 includes a list of taxa recognized for purposes of discussion throughout this work. The nomenclature has been updated wherever possible and includes combinations of Paphiopedilum by Stein (1892), which predate Pfitzer (1894).

THE FOCUS OF THE PRESENT WORK The purpose of this work is to expose patterns of diversity and biogeog­ raphy among the slipper orchids so that the Cypripedioideae can be better understood. Because features are obscured in herbarium specimens, primarily living or pickled specimens have been examined. A set of voucher specimens has been prepared which are listed in Appendix 10 and are deposited in the herbarium of the Marie Selby Botanical Gardens (SEL). In this work the fol­ lowing questions are posed. 138 SELBYANA [Vol. 7

How are the slipper orchids related? Morphological and anatomical fea­ tures are analyzed in chapters 2 and 3, and the data are used to construct cladograms in chapter 4. The cladograms are tested in chapters 5 and 6 for discrepancies with karyological and biogeographic evidence. Since biogeo­ graphic patterns emerge by accepting the cladograms, the inferred relation­ ships are interpreted as correct. Is the resemblance between Phragmipedium and Paphiopedilum likely to be superficial? In chapter 3 it is argued that the unique mode of labellum development together with the deciduous perianth and suite of vegetative characters suggest that it is not. Is the recognition of the four presently accepted genera justified? Al­ though Paphiopedilum and Phragmipedium are distinct, the plicate-leaved genera as now recognized are not. It will be shown that Selenipedium forms a geographical as well as morphological gradient into Cypripedium through the species C. irapeanum and C. californicum. However, C. arietinum has three primitive character states not found in other slipper orchids plus two anatomical features common to Selenipedium and should be considered as generically distinct. Are the many subgeneric taxa of Paphiopedilum and Phragmipedium as used by Pfitzer (1903) and Brieger (1971) justified? Data accumulated from morphological studies show that they are not. Several characters chosen for sectional recognition by Pfitzer and Brieger are poor taxonomic criteria since various species exhibit borderline character states. Furthermore, closely­ related species have been placed in separate sections depending on the whim of the taxonomist. Despite the striking differences in the appearance of the flowers, the similarities should be emphasized for constructing a useful classification. Is chromosome number reliable for determining affinities in the barbata group of Paphiopedilum? Although fission of metacentric chromosomes into telocentrics has undoubtedly occurred, high arm ratios among the metacen­ trics suggest that fusions have occurred as well, thus fissions are probably not simply accumulated with evolution of new species. Furthermore, it will be shown that the chromosome numbers differ greatly among seemingly closely­ allied species. Should the Cypripedioideae be considered primitive and relictual as indicated by Garay (1972) and Brieger (1971)? Although several species con­ tain character states expected of primitive orchids, the numerous specialized structures in Paphiopedilum indicate that some species are advanced and should not be considered primitive; that the presence of two fertile stamens merely indicates that the Cypripedioideae diverged very early during the evolution of orchids. Can the origin of slipper orchids be localized? In chapter 6 biogeo­ graphic patterns in Paphiopedilum, Phragmipedium and the plicate-leaved species suggest that the primitive taxa have moved southward with habitats which are most similar to the ancestral ones; that the most derived taxa oc­ cur in the northerly range limits of the three major groups. Since the northern hemisphere has undergone considerable cooling since the Miocene, it is in­ ferred that southward migrations of taxa have occurred, and that those an­ cestral species have changed the most in more northerly latitudes where they have persisted the longest. Although a northern hemisphere origin is postu­ lated, the probable widespread dispersal during Miocene time has probably blurred biogeographic patterns between the Old and New Worlds, and the origin cannot presently be further localized. 1984] ATWOOD: CYPRIPEDlOIDEAE 139

It is hoped that these questions will coax the critical reader into con­ sidering this work and to examining the evidence for alternative interpreta­ tions, for the scenarios included herein account for a number of often seem­ ingly unrelated facts.

CHAPTER II: MORPHOLOGY

In the introduction to his 1903 classification,Pfitzer did little more than describe slipper orchid morphology. The purpose of this chapter is (1) to dis­ cuss the distribution of morphological features, (2) to relate the features with known or speculated functions, and (3) to show that floral color, floral shape, and plant size are generally unimportant characters on which to base slipper orchid classification systems. Unless otherwise indicated, all observations have been made by me. Tables 2.1-2.4 summarize the distribution of morpho­ logical features among the taxa.

GROWTH HABIT All slipper orchids have a sympodial growth pattern with underground and aerial stems. The above-ground vegetative stems of the plicate-leaved species are unbranched with the exception of Selenipedium. Most slipper or­ chids are caespitose and form large clumps under favorable conditions, but a few species produce elongate rhizomes presumably for spread of individual clones. Cypripedium guttatum has a creeping rhizome and occurs in boreal forests where sexual reproduction may be relatively uncommon. Paphiopedi­ lum druryi colonizes cracks of rocks by elongated rhizomes (Mammen and Mammen, 1974). Phragmipedium pearcei forms elongate rhizomes and is well adapted for vegetative reproduction in river beds (C. H. Dodson, per­ sonal. communication).

VEGETATIVE MORPHOLOGY Roots. The roots of all mature slipper orchids grow adventitiously from the rhizome. They occasionally arise from abbreviated internodes in the condup­ licate-Ieaved species but never from inflorescences. Aside from differences in size, the roots of most slipper orchids appear externally similar. Stems. Internodes generally appear absent in distichous rosettes of the con­ duplicate-leaved genera but may occur in Paphiopedilum under subdued light. The aerial stems of North American Cypripedium do not exceed 1 meter except in C. irapeanum and C. californicum (Luer, 1975): Selenipedium usually is taller than 1 meter at maturity and may reach 5 meters. The ex­ treme height of S. chica may elevate the leaves near the stem apex to a level of sufficient light intensity for photosynthesis, and the apparently dispensable lower leaves are often decayed. Leaves. The slipper orchid leaf has a large simple blade. Only leaves of the plicate-leaved genera have sheaths surrounding the aerial stem, and the sheath margins are never fused. In Selenipedium chica the longest sheaths are found in the lowermost leaves, wheras the uppermost leaves bear little or no sheath. Apparently the longer sheaths lend support to the tall stems while maintain­ ing flexibility. Leaf abscissions are unknown in the Cypripedioideae, and the leaves simply die with senescence. + n D> ::l. ro :0. ~. D> " " "3 I "3 I I I I I I I i I I II i I I I ~ 3 V> J. <:T + + + + + + +++++++ + + + + Plants of tropical "~. regions o o .., "o Pl ant hei ght often more o .." + + + < than 1 meter 0> <.0 0> + + Vegetative leaves 12 or :;r more per sympodi um ~. ill + + + + + + + + + + + + + Leaf vernation convolute

+ + + + + + + + + + + + + + + Leaves in basal rosettes

+ + + + + + + + + + + + + Lower 1eaves with sheaths

+ + + + + + + + + + + + + + + Leaves .cbligately distichous

+ + + + + Leaves cori aceous 5: + + + + + + + + + + ...n '"0> + + + + Leaves condupl i cate ::3" + + + + + + + + + + + "re o ::!. m '< n o 0 o o o o o o o o o o o o o Leaf tips attenuate 0> + + <:T ..,o n" "- 0> n'" + + + + + + + + + + + + + + + Leaves with adaxi al keel ::!. ::r on midvein ..,'" '"n + + + + + + + + + + + + I nfl orescence multi flowered g- + J. V> ~. o 0 , o o o o .;, + + + o o o 0 0 o o o 0 Inflorescence may be lh .~ ." pani culate o o o o o o + + + + + + + o o o o o + o o o o Inflorescence successional­ ,...., flowered <: + + + + + + o Peduncle with one or more o o o o o o o o + o ","acts ~elow the inflores­ :-' cence + + + + + + Lea ves mattl ed ~. ~· ~ ~...... I3 . .., ~ "n ~. 0 " ~;·II~. .., ..,'" I I "'" '" I I I I I " n II I ~ o ,., o "~ "rI" " + ~ Pl ant of tropi ca 1 '"0. + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + regions

P1 ant hei ght often more than 1 meter

Vegetative 1 eaves 12 or more per sympodi um

Leaf vernati on convolute

+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + Leaves in basal rosettes o~ o t:I Lower leaves with sheaths

+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + Leaves ob1igate1y distichous

+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + Leaves cori aceous

+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + Leaves condup 1 i cate

o 0 0 0 0 0 0 0 0 000 000 0 0 0 0 0 0 0 0 0 0 0 000 0 0 Leaf tips attenuate

+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + Leaves wi th adaxi a 1 keel on midvein

+ + + + + + + +" + + Inflorescence multif10wered

o 0 o 0 0 o o o o 0 0 0 o 0 0 0 0 o 0 0 Tnf1 orescence may be pani cu1 ate

00+000 o o o o 0 o 0 o 0 0 0 0 o Tnf10rescence success i ana 1- flowered Peduncl e with one Qr more bracts below the i nf1 ores­ cence

+ '7' + + + + + + + + + + + + + + + ILeaves mottled 142 SELBYANA [Vol. 7

Leaf arrangement may be spiral or distichous in the plicate-leaved gen· era but is always distichous in the conduplicate-Ieaved genera. The pattern may be complex in Cypripedium as C. reginae produces distichous scales at the base of the stem which pass into spirally arranged leaves near midstem. In populations of C. reginae studied in open habitats in Michigan and New England (68 clones scored) all plants produced a counterclockwise leaf spiral from the stem base to the apex. However, in forest edges the stems may be ascending with leaves held in a nearly distichous arrangement. Cypripedium arietinum produces leaf spirals in either direction. These accounts of Cypri­ pedium do not support Pfitzer's claim (1903) that leaf arrangements of slip­ per orchids are always distichous. The leaves of Paphiopedilum and Phragmipedium are always basal, and those of Selenipedium and Cypripedium are always cauline. Although the usually paired leaves of C. acaule appear basal, internodes are conspicuous on close inspection. Leaf texture is consistently thin in the plicate-leaved genera and coria­ ceous in the conduplicate-Ieaved genera. While the coriaceous leaf is suffi­ ciently stiff and self-supporting for maintaining its shape, the thin, plicate leaf has corrugations similar to those found in metal roofing, which serve in structural support. The less massive plicate leaf probably requires less energy to produce than the thick coriacous leaf. This infers that the plicate leaf is more advantageous to plants which die down after a short growing season as in Cypripedium. The seemingly less dispensable coriaceous leaf may survive for two or more years and occurs only in the tropical genera. Although the leaves of Selenipedium chica are plicate, they are more durable and palmlike than those of Cypripedium. A year after I tagged a well-aged shoot of Seleni­ pedium chica in Panama, it still retained its upper leaves, indicating that tropical plicate leaves survive for longer periods than those of the temperate species. Leaf shapes in the plicate-leaved genera are mostly ovate or elliptic, but Selenipedium and Cypripedium irapeanum have attenuate leaf tips. The leaves of C. japonicum and closely allied species have flabellate leaves. Although the functional significance of leaf shape in Cypripedium is unknown, the at­ tenuate leaves of C. irapeanum and Selenipedium chica probably form effec­ tive drip-tips in areas of high rainfall. Leaves of the conduplicate-Ieaved gen­ era are ligular, elliptic or oblong to graminaceous. Phragmipedium has linear or even graminaceous leaves in those species inhabiting river beds. Leaves of the mottled-leaved Paphiopedilum species (especia,lly P. subgenus Barbata) are broader and oblong'to elliptic, while those of the remaining species are linear to oblong. Foliar pubescence is commonly found on both leaf surfaces of Cypri­ pedium, but aside from marginal cilia it is lacking in the conduplicate-Ieaved genera. Little is known concerning the functions of pubescence in Cypriped­ ium. The trichomes of C. reginae and C. pubescens have been known to cause dermatitis in humans (Andre, 1875; MacDougal, 1894), which suggests a pos­ sible defense against herbivory. In support of this view, Rolfe (ed.) (1894a) noted that "it has been noticed that large numbers of these plants growing in woodland pastures have been found intact, while the surrounding herbage would be closely cropped." Recently a presumed allergenic chemical has been characterized in C. calceolus (Schmalle and Hausen, 1979). The impli­ cated substance (cypripedin) is chemically related to known phytoalexins, which are defensive substances against fungal infections of various non-cypri­ pedioid orchids (Hardegger et aI., 1963; Urech et aI., 1963). 1984] ATWOOD: CYPRIPEDIOIDEAE 143

Plicate leaves lack folding around a prominent midvein with associated keel (Fig. 2.1) and must undergo a large amount of flattening on expansion. Conduplicate leaves are obligately distichous in bud and have abaxial keels on the midvein. During ontogeny each leaf half remains relatively flat and undergoes little change in shape by maturity. Occasionally the lamina is shifted asymmetrically in bud and produces two keels near the leaf margin. A more unusual situation sometimes occurs in complex hybrids, where the normally conduplicate leaf lacks abrupt folding around a major vein and de­ velops as a plicate though coriaceous leaf. It is possible that the occasional development of plicate leaves and sheaths in the conduplicate-Ieaved genera represents teratological atavisms.

MORPHOLOGY OF THE INFLORESCENCE Pfitzer (1903) correctly noted that slipper orchid inflorescences are al­ ways terminal, often solitary-flowered, and are sometimes multiflowered in racemes or panicles. They are usually pubescent, but trichomes are essentially absent in Paphiopedilum spicerianum and individual clones of Phragmipedium longifolium. Contrary to Vermeulen (1966), inflorescences are racemose though often single-flowered. The solitary flower of many Cypripedium and Paphiopedilum species is axillary and is often accompanied by an aborted terminal shoot. Occasionally a two-flowered raceme occurs in normally single­ flowered inflorescences. The Selenipedium inflorescence is a raceme, or occa­ sionally a panicle of racemes as with S. chica, but is never solitary. In Cypri­ pedium, racemes are less common than solitary flowered inflorescences, and panicles are unknown. Of the North American species, C. californicum, C. fasciculatum, and C. irapeanum normally produce racemes. Cypripedium montanum and C. reginae regularly produce two (rarely more) flowers, but other North American species rarely produce more than one. Phragmipedium inflorescences are always multiflowered and are usually racemes, but robust specimens of P. caricinum (rarely P. [ongifo[ium) produces panicles of ra­ cemes. The P. lindleyanum complex regularly produces panicles of racemes, but I have never seen panicles in the P. caudatum complex. Several species of Paphiopedilum produce multiflowered racemes but never panicles. The peduncle supporting the inflorescence may bear one or more sterile bracts. In Selenipedium and Cypripedium with cauline leaves, the leafy stem supports the inflorescence, but C. acaule, C. debile, C. elegans, C. fascicula­ tum, and C. japonicum have scapes. In the conduplicate-Ieaved genera the peduncles below the inflorescence bear none to several bracts. Phragmipedium almost always has one or more bracts below the inflorescence except in de­ pauperate specimens and the P. caudatum complex. Phragmipedium boissier­ ianum, P. longifolium, and allied species may produce up to five bracts be­ low the inflorescence. Paphiopedilum normally has scapes, but I have ob­ served bracts near the base of the peduncles in robust clones of P. concolor and P. ciliolare. Figure 2.2 illustrates the range of variation observed in slip­ per orchid inflorescences.

FLORAL MORPHOLOGY The unique floral morphology of slipper orchids provides primary evi­ dence for relationship among otherwise very differently-appearing plants. Figure 2.3 shows a floral diagram representing the floral parts of a typi~al slipper orchid including a sepal, synsepal, two lateral petals, a labellum or hp, 144 SELBYANA [Vol. 7

A B

c .,

., D

E ..

..

Fig. 2.1. Leaf vernation. A. Camera lucida drawing of a transverse section through a young shoot of Cy pripedium reginae showing convolute vernation. B. Camera lucida drawing of a transverse section through a young shoot of Phragmipedium X sedenii showing condup­ licate vernation. Alternate leaves are shaded to reduce visual confusion. C. Schematic representation of plicate leaf development. D. Schematic representation of normal con­ duplicate leaf development. E. Schematic representation of abnormal conduplicate leaf development with shifted lamina. F. Schematic representation of an otherwise condup­ licate leaf occasionally observed in complex Paphiopedilum hybrids resulting in a thick but plicate leaf. 1984] ATWOOD: CYPRIPEDIOIDEAE 145

Table 2.2. Distribution of sepal characteristics: (1) vernation, (2) presence of color patterns with lines or

spots, and (3) sepal area ratiosf The digits in parentheses are the numbers of flowers per species sampled for determi ni ng sepal area rati os. ~arentheses i ndi cate nearly borderl i ne character states. I

~ ~ ;:- s-~ ;:- s-~ -s- o", -s- o", 0 s- -0 0. 0 s- -0 0. OJ 0 Ol Ol 0 Ol +'~~ "-~ +'~~ "-~ -0 0" '" _ a. -0 '" ~ a. Ol ,., >-- Ol 0" >-- OVI 0;;;- ';;Q.lCIJ .~ r;;;cuQ} >+'+' ~" ~o >+'+' ~" :::;0 -0 "'+'~ -0 "'+'~ VlUS- ~Ol s- Vl ~us-"'''' ~Ol s- Vl ...-',..."'''' 0 ~ +' ~ Vl ...... 0 ~ ..... ~ Vl "'s-"- "'S-"- "'+' ra ra ...... · 0..0 s- "'0.0 ..... "''''OJ o.Vl ...... 0..0 s- 0.0 Ol o.Vl OlEOl WOO S-OlC OlEOl Olo. S-OlC SQecies V).,...... 0.. Vl ~ « ~- V1 .... c.. VlVl «~-

C~~ri ~edi um PaBbi 0Bedi 1 ym (cont'd) acaule P - call osum I + 4.4 (2) arie:tjnym p - ciliolare I + 2.9 (3)

cal ifornicum P - .£Q.!!.£QlQr I + 1.5 (4) candidum P - curtisii I + 2.9 (1) fasciculatum p - delenatii I - .90 (1) ~I p + exul I + 1.3 (1) itm~DYU!1 p - fS!irieanum I + 2.5 (10) la~omcum P - godefroyae I + 1.2 (1)

montanum I p - ha~na 1di anum I + 1.4 (1)

Bar~i fl or~m P - henni ~ i aDYDl I + 2.7 (3) ~ub~scens I P - hirsutissinum I - 2.0 (2) - I + 1.7 (1) reglnae u~ P insigne Se 1eni Dedi u 1awrenceanym I + 3.3 (1) I chica P - 1 inii I - 1.9 (1) Phragmioedi m J..Qlili I (- ) 1.3 (1)

boissieri num V - ma~:ters i anym I - 2.0 (1) I S;Q.(j!:;jDLlilli V - .58 (1) niveum I - 2.2 (2) ~I V - .73 (1) .Mr.i.ilii I (-) 1.2 (4) lind1evanum V - phil ippinense I + 1.1 (2)

10naifoli m V - .54 (2) I2ra~s:tan~ I + 1.0 (1)

~earcei V - purpura tum I + 2.3 (1) V - .74 (1) spi ceri anum I + 3.0 (1) .llili!!1ii i wall i si i V - .64 (1) stone; I +

warscewi zianum V - .73 (2) ~ukhaku1ii I + 1.4 (2) 'aohiooedi m tonsum I + 3.0 (2) 1.8 (3) acmodont m I + ~ I + 1.7 (3) aooleton anum I - 1.8 (2) vi ctgri a-reg; na I + 1.7 (2)

I + 2.1 (2) ~ I villosum I +

I + 3.1 (1) v;Q]a~c~DS I - 1.8 (2) ~I (1) bu11enia~um I - 2.0 I 146 SELBYANA [Vol. 7

Table 2.3. Petal Morphology. Data in pa rentheses indicates nearly bord~rl ine character states.

~ u <= s:: "0 0'<-- <:: 0 0 '_.c OJ ...-- 0:·,.... .<:: .c ..., U ..., ..., +' . +'01 Vl ';;; ~; VI ...... cc c -o~ '0 <:: OJ !.. OJ OJ, !::: V) ~s:: .,..... co...-<:::( "'.~ .~ u ~....cJ,O OJ ~ OJ'o- +,,,, !.. '" !.. '" -0 !.. E'" x E E Vls:::::rd.s::: "-E 0.'" ...oQJQ)S::: ::SVlQ)Q) <:: .---->+-> 0.>0 E 1'0.-- 0.. EE=> " ~!.."'''' !.. <:: "'<:: .--- ro \/) ...... "'~ ...,.~'" .,... ..0 -0 res S-(1)aJQ) ...,'" rd....s::: s:::-iJ x '0 'x'x ...,...,'" => ...,'" ''-'" VI,V) s- !....., !.. E -j.JVlC'C$O <'0+->.,.... So- Q) 0. 0...,.- ...0 I'd , !..w Speci es ::EroEO :EE'" '" a. => ~ (/)300.. 30."'''' 3~"'''' o..:::J.;JU')'"

C:i~ri Qedi um acaul e .21 + - (10)

ari eti num .12 californicum .38

candi dum 8.M .11 + - (10) gutta tum .45

i raQeanum M .5

montanum +

~arvi fl Drum .14 + - (TO) Qubescens 8,M .082 + - (TO)

regi nae M .38 Se1eniRedium

chi ca M .19 0 Phragmi Qedium

boissierianum .13 +

cariei num .1 + - ( 3) cauda tum .031 + - (1)

1 i ndl eyanum A (+) - (1)

lonqifolium .16 + - (20)

Rea rcei + - ( 2)

schl imii M .67 0 wa 11 is i i .028 + - (1)

warscewi ezi anum .032 + - (4) PaphioRedi 1um

acmodontum A .36 + 0 + + aRRletonianum A .35 0 +

ar~ M .25 (- ) 0 + +

ba rba tum .3 (- ) 0 +

bull eni anum A .25 (+)

callosum M,A .21 (+) +

£jl iolare M .23 (- )

concolor M .51 0 curtisii M .27 + del eniLti i M .83

exul A .25 1984] ATWOOD: CYPRIPEDIOIDEAE 147

Table 2.3. Conti nued , -"" u c c "0 04- c 0 0 ,---.c ,....- 0',.... .c .c...., '" u ...., ...., ...., ';0 s..",_ C C c ""'0> Q) Q) C VI "0- -0 C '" s.. :;:.~ '" C ·~co-<::( .- Q) '"u s-...o a OJ '" Q)'r- Q) 3~ '0. rO E''''' V) OJ s.. '"OJ 3>---=. --- -0 0. "''''''OJ '" -0 '" VlCrt:i..c s.. '" '" s.. E OJ x E E '" +..l o.E & ..oWQ.lC ::::SVlQ)Q) '" OJ .....- ___ > 0.>0 E 1'0.--- 0.. .-- en V)'r- '"E '" E " '" s.. " "'~ '" 'r- .D -0 (1j ~S '" ....,.-"''' <'O..c C+..l X -0 'x 'x ....,...., ...., .~ ~:;: ~ ....,'"s...... , s.. E +-lU'lrtSu 1'0+-=0-,- s- '" 0... •.--...0 ro OJ , s.. OJ ::;:::roEO =<=E ~g- ~ (/)300... '" OJ 3~ D..~-I-lVl Speci es '" '" '" 0. '" '"

Paphi opedil urn (Cont'd. )

fai ri eanum .26

godefro:ta~ M .76 0

hal'na 1 di anum A .19 + 0 (+) +

henni 5 i anum M,A .31 0 +

hirsutissimum A .33 +

insigne 1\ .32 0

1 aevi gatum .067 + - (1) +

1awrenceanum M .22 0 +

1 ini i A .31 + 0 (+)

lowii A .2 + (+) +

mas ters i anum M,A .34 (- ) (+)

niveum t1 .75

parishii B .1 + + (7) + +

phi 1 i Poi nense B .048 + - (5 ) +

Qraes tans .07 + + (1) +

j2ur(2uratum fl .35 + +

sQi ceri anum .38 (+)

stonei B + - (1)

sukhakul i i M .27 0 + +

tonsum A .25 0 + (+)

venus tum M,A .34 + + +

vi ctori a-reg; na B .<'B + + (10)

vi 11 osum A + 0

yi 0] ascens M .35 0 .... f-I II) ..'" ~. .... ~ I ~ ~ ~ ex> ~. iD· ~. ~. I· I ~ I: I· I i ~ I ..::l. II I I ~. 3 3 ~ 3 !§ I I i I I .. " i 5! I ~ II I i Ii f I ....'" I 3. , C" + I+ + + + + + + + + + + + Lip apex infolded <: ." .... 2:: 2:: o· Auricles present + + + + + "0 near apex .." Hollow regions ~ + + + + + + + + ...... at base ...<: + + +" + + + + + + Lateral lobes fused ~ 0 .." -;- Outpocketi ng or .... + + + +" +" + + + + ~ 2:: 2:: 2:: 2:: 2:: 2:: 2:: II horns II present II)

..C" + + + + + + + + + + Lip with broadened rim ~ <: ~ I"Il + + + Light windows [:!:J present t"' &?.... til 3 ~ .. >< <:'" ::: ,. :0. II) > ~ ...... ~. ~ i[ z "" ~. II) I~~. I~~. II) ~. <: 3. " ~ > i Iii· II) ~. '" 0> ".. ~. ... ~ 0> ~. "II) " ~ II) 0> 0> ~. I~· ~ Ii I II) '" II " '" I 3" !§ .... I ~. I r !§ "~ !§" I !§" II) I "II) ! I f i '" 3 I c:> " ~ I 0 '"~ I ".... + + + + C- Lip apex infolded "~ 0 -;- -;- -;- Auricles present ...... + + + + + + + + + + + + , II) + + + + + + + + + + near apex "II) Ho 11 ow regi ons ..... at base <:<: C" 0 Lateral lobes fused d. II) :!. + Outpocketing or + + + + + ~. "horns II present II) ....., lip wi th broadened ..g. rim ;;; < 0.... 0 + + + light windows ...II) :- present ...:] 1984] ATWOOD: CYPRIPEDIOIDEAE 149

A B c D

E F G H

J K L M

Fig. 2.2. Schematic representation of inflorescences. In some cases many more flowers would be formed per inflorescence. A. Phragmipedium longifolium. Inflorescence on the right with a lateral raceme was found in a naturally-growing specimen at El Valle, Panama. B. Phragmipedium schlimii. C. Phragmipedium warscewiczianum. D. Phragmipedium lind­ leyanum. E. Paphiopedilum philippinense. F. Paphiopedilum victoria-regina. G. Paphio­ pedilum concolor. H. Paphiopedilum ciliolare. The inflorescence on the right taken from a robust specimen is unusual in the presence of a bract near the base of the scape, as well as the two-flowered raceme. I. Cypripedium acaule. J. Cypripedium reginae. K. Cypri­ pedium californicum. L. Cypripedium irapeanum. M. Selenipedium chica. 150 SELBYANA [Vol. 7

os• os• r so '\ SD 00 00 FS, \ LP GA X X X ~lj

s~"""'''''·$'''''''/$ j A B $ os so 00

c Fig. 2.3. Floral diagrams representing various members of the Cypripedioideae. Floral parts are abbreviated as follows: S, sepal; DS, dorsal sepal; SS, synsepal; LP, lateral petal; L, labellum; FS, fertile stamen; SD, staminode; G, gynoecium. X '8 represent positions of stamens which would be found in the presumed ancestral Liliaceae. A. "Normal" flower which is widespread among the Cypripedioideae. B. Phragmipedium lindenii with a third petal in place of the labellum (labeled "L "), and a third stamen. C. Cypripedium arietinum with two separate lateral sepals in place of the synsepal. 1984] ATWOOD: CYPRIPEDIOIDEAE 151 two fertile stamens, one staminode, and a three-carpellate gynoecium. In only two known cases are there populational deviates from this floral pattern: those of Cypripedium arietinum having three distinct sepals, and Phragmi­ pedium lindenii lacking a labellum and having an extra stamen. The following sections outline the infrageneric variation encountered in the floral parts. Perianth. Pfitzer noted (1894) that perianths are deciduous in Phragmipedium and Paphiopedilum and persistent in Cypripedium and Selenipedium. How­ ever, most flowers of non-pollinated Paphiopedilum wilt before total abscis­ sion occurs, at least in cultivation. The significance of deciduous peranths is not apparent. The flowers of several Phragmipedium species often abscise when seemingly fresh. Calyx. At full anthesis the dorsal sepal of Selenipedium, Cypripedium and Phragmipedium is flat, slightly twisted, somewhat concave, or rarely convex. In Paphiopedi/um the sepal may be strongly concave to complexly folded in­ to a hood. When flattened the dorsal sepal may be oblong, orbicular, or ovate, but is rarely lanceolate as in Phragmipedium caudatum. The dorsal sepal un­ doubtedly serves to protect the sexual parts in bud. However, the large and colorful dorsal sepals of Paphiopedilum suggest roles in pollinator attraction and in preventing rain from filling the labellum. The synsepal (fused lateral sepals) is usually concave, elliptic to ovate, and least conspicuous of perianth parts. Several lines of evidence support the interpretation of the synsepal as a pair of fused lateral sepals. (1) The syn­ sepal is often apically bifurcated as in the Cypripedium calceolus complex. (2) Cultivated specimens occasionally "revert" to the aposepalous condi­ tion, and aposepaly is normal iri Cypripedium arietinum.(3) Synsepals often contain "two midribs" or keels as in Paphiopedilum praestans. (4) The syn­ sepal of Phragmipedium has nearly twice the surface area as the dorsal sepal (Table 2.2), suggesting the inheritance of a more "primitive" morphology from a putative ancestor having three distinct sepals of equal size. (5) The fu­ sion theory is consistent with the generality that monocots have three sepals. The synsepal probably functions mainly in protection in the bud stage and is especially inconspicuous in mature flowers of Paphiopedilum subgenus Bar­ bata (Fig. 2.4F). Sepal vernation is valvate in Phragmipedium, imbricate in Paphiopedi­ lum, and perforate in Selenipedium and most Cypripedium. Since Rolfe's classification of 1896 sepal vernation has been given fundamental impor­ tance in delimiting slipper orchid genera. By 1903, Pfitzer distinguished the loose, non-tangent sepals in buds of Selenipedium as "perforate" vernation. This also applies to the eastern North American Cypripedium species, since sepal margins are separated long before anthesis. The valvate sepals of Ph rag­ mipedium must develop at the same rate, otherwise bud distortion occurs, as happens in P. warscewiczianum to some extent. In Paphiopedilum imbricate vernation provides the dorsal sepal with greater protection, borne as it is within the synsepal margins, and the dorsal sepal is free to enlarge without distorting the bud. This implies that development of the presumed banner and umbrella of Paphiopedilum flowers is favored by imbricate sepal arrange­ ments. Sepal enlargement is apparently not constrained in the plicate-leaved genera, but the dorsal sepal size is rarely extreme. Table 2.2 includes the area ratios of the dorsal sepal to synsepal in the conduplicate-leaved genera. The ratios are smaller inPhragmipedium and consistently larger inPaphiopedilum. Figure 2.4 illustrates some of the variety observed in sepal areas. 152 SELBYANA [Vol. 7

Lateral Petals. Much attention has been given to petal morphology in classi­ fications by Pfitzer (1894, 1903) and Brieger (1971), particularly with refer­ ence to the mottled-leaved species. Petal morphologies range from the ordi­ nary to the most intricately specialized. Selenipedium chica has ordinary concolor (uniformly colored) petals with no unusual ornamentation. Petals of Cypripedium californicum, C. ira­ peanum, C. passerinum, and C. reginae also have similarly ordinary petals. Cypripedium acaule, C. pubescens, C. parviflorum, C. candidum, and C. arie­ tinum have lanceolate or linear petals with a green background lined or suf­ fused with purple. In all plants examined the lateral petals of C. acaule, C. candidum, C. parviflorum, and C. pubescens form unidirectionally counter­ clockwise spirals from the floral receptacle (Table 2.3). The petals are asym­ metrical in that they spiral in the same direction relative to the receptacle. No clockwise spirals are known in petals of New WorldCypripedium although they may occur in various Asiatic species. Cypripedium guttatum has the most complexly marked petals of the North American species. The petals of Phragmipedium occur in basically three forms: linear,lig­ ulate, and ovate to orbicular. Phragmipedium longifolium, P. caricinum, P. boissierianum, and P. pearcei have lanceolate petals with green background variously suffused with purple. Although no specialized color patterns occur in Phragmipedium, the ligulate petals of P. lindleyanum and P. sargentianum with partly turned apices lack the green ground color near the apex and ap­ pear similar to those of the Paphiopedilum bullenianum complex. Ovoid­ orbicular petals are unique to Phragmipedium schlimii and are white with pink suffusion. Perhaps the most unusual petals by virtue of their length are found in the Phragmipedium cauda tum complex. I have measured petals in excess of 67 cm (26 inches) in P. warscewiczianum, although claims of 3-foot petals in the P. caudatum alliance exist in commercial catalogues. The petals of P. wallisii are usually somewhat shorter. Petal spirals are always unidirec­ tionally counterclockwise as in Cypripedium. The petals of Paphiopedilum are the most diverse and complex among the slipper orchids. Petal shapes include lanceolate to linear (often variopsly spiraled as in Cypripedium and Phragmipedium), ligulate, oblanceolate, land ovoid to orbicular. Although the range of petal shapes is similar to tha~of Phragmipedium the ornamentation with trichomes, warts, and various c lor patterns is much more complex. Paphiopedilum praestans and P. phili 'Pi­ nense have Ian ceo late petals with glandularlike processes near the petal b es which are often associated with cilia. Although P. praestans and P. wilhelmin­ iae are closely related, the latter species presumably lacks glandular petal bases. The petals of P. parishii are usually lanceolate but may be broadened toward the apex. As with the Phragmipedium caudatum complex, Paphio­ pedilum sanderianum has elongate petals. Among those species with lanceo­ late to linear petals the ground color is basically green or brown variously lined or suffused with purple. The most complex petals of the lanceolate­ petaled species are those of P. parish ii, which have raised warts with long, pilose hairs. Petals of Paphiopedilum victoria-regina twist clockwise in all clones examined (Table 2.3), but counterclockwise in P. laevigatum, P. phil­ ippinense, and P. stonei. Those species with ligulate and oblanceolate petals are the most com­ plexly ornamented in Paphiopedilum. Paphiopedilum fairieanum has com­ plexly ciliate, violet-lined petals superposed on a white and green background. Among the Indian species, the petals of P. druryi and P. spicerianum have 1984] ATWOOD: CYPRIPEDIOIDEAE 153

A B

F

Fig. 2.4. Outlines of representative calyxes. The lower cucullate synsepals have been split in five species examined in order to spread them out. A. Cypripedium irapeanum. B. Se­ lenipedium chica. C. Cypripedium reginae. D. Phragmipedium longifolium. E. Paphio­ pedilum praestans. F. Paphiopedilum lawrenceanum. 154 SELBYANA [Vol. 7 yellow to green backgrounds and prominent central purple lines. The petal margins of P. spicerianum are unique in their complex crisping. The ligulate­ petaled species of P. subgenus Barbata are provided with complex patterns of warts or spots. Several species of the P. barbatum alliance only have mar­ ginal warts on the petals, while in others warts are distributed over much of the lamina. Diverse species occur with oblanceolate petals. Paphiopedilum lowii and P. haynaldianum have long, oar-shaped, oblanceolate petals spotted near the base. These species have V-shaped petal bases, in cross section, which proba­ bly serve for structural support, since a long, flat petal would be less able to hold its horizontal position. Paphiopedilum parishii also has V-shaped petal bases, but the petals are spiraled. Several species in P. subgenus Barbata (P. hookerae, P. bullenianum, P. linii, P. celebesense, P. venustum, and P. apple­ tonianum) have oblanceolate petals warted at the base. The oblanceolate petals of PaphiopedUum hirsutissimum are copiously pubescent, but those of the closely-allied P. esquirolei are presumably not. In most species with oblanceolate petals the apices are whitened and variously suffused with vio­ let, and both petals form a quarter symmetrical twist with the adaxial sur­ face upturned. The oblanceolate petals of P. tonsum and P. mastersianum are exceptional in that they do not form the upturned violet apices. The brachypetalums have ovoid to orbicular petals with obtuse apices (Pfitzer, 1903). All have ground colors of white to yellow, may be maculated with spots of various sizes and are usually abaxially pubescent. Pfitzer (1903) and Brieger (1971) gave sectional recognition to mottled­ leaved species with cilia presumably associated with warts on petal margins. However, the distinction does not deserve taxonomic consideration since the cilia are essentially equally distributed in four species examined (Fig. 2.5). The Labellum or Lip. Labellum shape is extremely variable within slipper orchid populations, and consistent differences may be observed between clones. Adams' analogy (1959) of morphological variation to variation in the human stomachs is most appropriate to the pouched labella of slipper orchids. Except for Phragmipedium lindenii the labella of all slipper orchids are saccate. The lips serve to retain and channel the insects for pollination (Dod­ son, 1966; Stoutamire, 1967; Newhouse, 1976). All slipper orchid labella studied have at least one pair of infolded lateral lobes plus a terminal lobe. Pfitzer (1903) indicated that Cypripedium lips are five-lobed, but in the spe­ cies studied, the extra lobes to which he referred appear to be outfolding of the bases of one larger pair of lateral lobes. The saccate shape of the lip is rather constant among species although the abaxial sides are occasionally prolonged into a horn as in Cypripedium arietinum and Paphiopedilum con­ color. The major differences occur in the shape and ornamentation of the orifice, features of the lip margin, and color distribution. The margins of all slipper orchid labella exhibit various degrees of infolding, lobing, outpock­ eting (as in the "horns" of Phragmipedium longifolium), and presence of auricles. Selenipedium chica has a very simple bedpan-shaped labellum (Fig. 2.6A) with two infolded lateral lobes and an infolded apical margin. Orna­ mentation is absent, but a pair of violet patches appear on the labellum mar­ gin, a feature shared by S. steyermarkii and S. aequinoctiale (Garay, 1978). As with Selenipedium, Cypripedium californicum has a simple lip with infolded margins and faint violet patches. The lip of C. irapeanum resembles that of Selenipedium chica in its simplicity and presence of violet patches. 1984] ATWOOD: CYPRIPEDIOIDEAE 155

A

o Figure 2.5. Camera lucida drawings of petal margins with cilia and warts. Although a few more trichomes tend to occur near the warts, differential cilia distibution in species with marginal warts is not worthy of sectional recognition as practiced by Pfitzer (1903) and Brieger (1971). A. Paphiopedilum barbatum, abaxial view. B. P. callosum, adaxial view. C. P. lawrenceanum, abaxial view. D. P. hennisianum, abaxial view. 156 SELBY ANA [Vol. 7

Cypripedium fasciculatum also appears to have a simply-infolded margin and a pair of small lateral lobes (Luer, 1975). Cypripedium arietinum has a simply infolded and pubescent margin with apiculate apex, but otherwise the lip appears similar to that of Selenipedium chica. The lips of Cypripedium re­ ginae, C. candidum, C. parviflorum, C. pubescens (Fig. 2.6B), and C. acaule have two infolded lateral lobes. These species exhibit various degrees of out­ pocketing, and the lip apex is minimally infolded. Unlike the other species studied, the lips of C. acuale (and perhaps the Asiatic species C. japonicum, C. formosanum, and C. cathayeanum) have compressed sides forming a fis­ sure-shaped orifice. Cypripedium guttatum is unique in that it has a broad­ ened rim at the lip orifice. Although the apex appears not to be infolded, the thickened rim may be derived from "fusion" of the infolded margins. Phragmipedium has several labellum features which are unique among the conduplicate-Ieaved slipper orchids. (1) All have infolded apical margins. (2) The basal infolded lobes are tangent and usually partially fused. (3) All have conspicuous "gibbous hollows" at the base except P. schlimii and P. besseae. (4) All have some degree of outpocketing (Fig. 2.6C, D), which is best developed in P. longifolium (Fig. 2.6C), P. bOissierianum, and P. vitta­ tum as "horns." (5) A rim 'is produced at the marginal fold which is broad and conspicuous in some species (P. cauda tum complex) and often pubescent. For their otherwise highly developed flowers, the lips of Paphiopedilum are little more complex than those of Phragmipedium. All species studied have at least one pair of lateral lobes, and many produce a secondary pair close to the apex consisting of an upturned portion of the lateral margin. The base of the lateral lobes are approximate but apparently never fused as in Phragmipedium. Unlike Phragmipedium and Cypripedium guttatum no broadened rim is formed in Paphiopedilum, and the apex is inturned only in the brachypetalum alliance (Fig. 2.6F). All the barbatas have verrucose in­ folded margins. The lip bases of Paphiopedilum haynaldianum, P. lowii, and P. parishii produce linear translucent tissue (windows) which presumably help the pollinator find the exit. Similar windows occur in the Cypripedium .. calceolus complex andPhragmipedium schlimii. The auricles of many Paphio- -­ pedilum species (Fig. 2.6G) appear as extensions of the lateral lobes. Auricles appear absent in P. philippinense, but they become evident by artificially compressing the lip apex (Fig. 2.6E). Although dorsal compression of the lips of the "earless" P. praestans fails to produce auricles, presence or absence of auricles is not fundamentally important to recognition of subgenera, as in the classification of Pfitzer (1903). Furthermore, the auricles of P. bulleni­ anum, P. tinii, and P. appletonianum are poorly developed and may represent an intermediate condition. An apicule may be found in P. insigne, P. exul, P. villosum, and P. acmodontum, and the margin of the lip is uneven in the P. bullenianum complex; otherwise the lips of Paphiopedilum have few unusual features. Table 2.4 summarizes the features of the labellum. The Androecium. With one exception the androecium of slipper orchids consists of a sterile stamen (staminode) of the outer staminal whorl and two lateral fertile stamens of the inner staminal whorl. The exceptional Phrag­ mipedium lindenii has an extra stamen in the inner whorl with elongated fila­ ment which causes contact of the anther with the stigmas and consequent autogamy. The staminodes, however, exhibit a bewildering array of forms, and many taxonomic concepts have been based on this structure (see Fig. 2.7 1984] ATWOOD: CYPRIPEDIOIDEAE 157

E

F G

Fig. 2.6. Frontal and cut profile views of labella. A. Selenipedium chica with simply­ infolded lateral lobes. B. Cypripediumpubescens with small lateral lobes, which are turned outward at the base. C. Phragmipedium longifolium with enlarged lateral lobes and out­ pocketing (horns). D. Phramipedium caricinum with large lateral lobes, and a distinct rim. E. Paphiopedilum laevigatum showing minimal development of auricles (left), which are more conspicuous when the apex is depressed (center). As with most species of Paphi­ opedilum the apex is not infolded. F. Paphiopedilum niveum with poorly developed auricles and infolded apex. G. Paphiopedilum parishii with conspicuous exerted lateral auricles. 158 SELBYANA [VoL 7

and 2.8). Since many species limits are presently being recognized on the basis of subtle staminode differences (Garay, 1979), staminode morphology will be detailed here. The presumed function of the staminode is to prevent the pollinator from escaping without collecting a pollen mass. However, Nilsson (1979) has observed that pollinators tend to fly toward the staminode, which suggests a function in modifying pollinator behavior. The basic structure of the staminode is simple. In all species examined a central thickened region connects two lateral appendages which appear to be modified anther sacs. Cypripedium arietinum staminodes resemble stamens to the extent that pollen production seems possible. The essentially orbicular staminode of this species has a conspicuous concave depression inside each theca and an apparent extension of the filament. Staminode shape ranges from orbicular to reniform, trulliform, or shield-shaped. Perhaps most staminodes are basically trilobate, but additional lobules may be found at the base or between the larger lateral lobes and apex. The basic shape is complicated by undulate margins as in the petals. The staminodes are often simply concave or convex. Selenipedium chica has a simple, essentially flat, trulliform staminode with no elaborations. It has the smallest staminodes found among the slipper orchids. Selenipedium aequinoctiale, S. palmifolium, and S. steyermarkii also have similarly small staminodes (Garay, 1978; Hoehne, 1949; Dunster­ ville and Garay, 1963). Cypripedium has relatively large staminodes compared to Selenipedium. Cypripedium arietinum has orbicular, convex staminodes, but often with ir­ regular outlines. The staminodes of C. californicum studied are orbicular, al­ though Pfitzer (1903) indicates that they are reniform. A popUlational survey might reveal much variation. Cypripedium acaule and C. irapeanum stami­ nodes are basically orbicular but with a prolonged, upturned, acute apex. Concave, trilobate, rhomboid staminodes occur in C. candidum, C. parviflo­ rum, C. pubescens and C. montanum (Luer, 1975). Although similar, C. re­ ginae has shield-shaped, slightly concave staminodes. Staminode shape in C. pubescens varies considerably (Fig. 2.7). The most unusual staminode occurs in S. guttatum which is apparently bilobed at the apex and base, arched in natural position and longitudinally grooved. The staminodes of Phragmipedium are basically trilobate with the side lobes extended laterally, although occasionally they project forward. Two basal lobules are found in all species studied except P. longifolium. Phragmi­ pedip,m caudatum, P. linden ii, P. waliisii, and P. warscewiczianum have very wide trilobate staminodes. Phragmipedium boissierianum also has wide tri­ lobate staminodes, but its shape is apparent only when flattened. Phragmi­ pedium longifolium, however, is more complex as it has staminodes grading

Fig. 2.7. Slipper orchid staminodia. Several examples showing variation (F, H, P) do not support use of staminodes as fundamentally important structures to species recognition. A. Selenipedium chica (X 4.2). B. Cypripedium irapeanum (X 1.5). C. C. reginae (X 1.7). D. C. californicum (X 3.3). E. C. arietinum (X3.3) showing dorsal (left) and ventral (right) surfaces. A filament apex appears between the anther sacs, which is also seen in ex­ amples V-X. F. C.pubescens (X 1.7). G. C.parviflorum (X 1.7). H. C. candidum (X 1.5). I. C. guttatum (X 3.3), showing longitudinal lamellae. J. C. acaule (X 1.7). K. Phragmi­ pedium schlimii (X 1.7). L. P. warscewiczianum (X 1.7 ). M. P. cauda tum (X 1.7). N. P. wallisii (X 1.7). 0.. P. caricinum (X 1.7). P. P. longifolium (X 1.7). The specimen on the left is entity gracile, the others are entity longifolium. Q. P. boissierianum (X 1.7). R. Ventral view of P. laevigatum (X 1.7) showing inpocketed regions filled externally with trichomes. S. P. philippinense (X 1.7), which lacks the inpocketed regions on the ventral surface (left). T. P. praestans (X 1.4) showing development of lateral pUbescence. Note 1984] ATWOOD: CYPRIPEDIOIDEAE 159

inpocketed regions on the ventral side (right). U. P, victoria-regina subsp. primulinum (left) and subsp. chamberlainianum (right) (X 1.6), V, W. and X. P. haynaldianum, P. /owii, and P. parishii (all X 1.5), Note the filament apex on the ventral surfaces. The acute umbo is barely distinguishable in frontal view but is evident in the slightly tilted ventral view of P. parishii. 160 SELBYANA [Vol. 7 in shape between trilobate and reniform, varying considerably in pubescence. The small-growing plants of the Caribbean slope growing on river rocks ex­ hibit trilobate staminodes often lacking pubescence. The resemblance of these staminodes to those of immature buds of the larger-growing embank­ ment plants of the Pacific slope suggests neoteny. Phragmipedium lindley­ anum has more elongate, trilobate, pubescent staminodes, while P. schlimii has simply trilobate staminodes with a pair of basal lobules. As indicated by Pfitzer (1903) Paphiopedilum exhibits the most diverse and complex staminodes. Those of most species appear basically trilobate, but unlike Phragmipedium the lateral lobes are directed forward near the apex. Additional lobules occur at the base or between the lateral and apical lobes. The simplest staminodes occur in P. subgenus Brachypetalum. The apex occasionally exhibits three acute lobes, but this feature is not consistent. The staminodes of Paphiopedilum delenatii are among the largest in the genus and are orbicular-convex and shallowly trilobate at the apex. Paphiopedilum concolor may form a long attenuate apex, but this feature is not consistent. Paphiopedilum victoria-regina (cochlopetalum group) also has simple three to five lobate, convex staminodes. Which pair of lateral lobes corre­ sponds to the single pair observed in most other species is not clear, if such homology exists. Within the coryopedilum group (P. subgenus AnotopedUum Pfitz.), P. laevigatum and P. praestans have convex, trilobate staminodes with lateral concave chambers filled with trichomes. Curiously, the chambers are lacking in the single clone of P. philippinense studied. Whether this difference is con­ sistent in natural populations is not known. The staminodes of these species bear a resemblance to insects. Within the pardalopetalum group (P. section Pardalopetalum Pfitz.), staminodes are trilobate and also bear a remarkable resemblance to insects. These species have an acute umbo dorsally and an apex on the filament. The lateral lobes are closed in P. haynaldianum but spreading in P. lowii and P. parishii. The Himalayan non-mottled-Ieaved species have staminodes similar to those of the pardalopetalum group, but the umbo is blunt and located cen­ trally or near the apex. The lateral lobes are often indistinct, and there is no filament apex. All staminodes are trilobate but often have extra basal lobules. Staminodes of Paphiopedilum spicerianum are unique in their deep violet color, smaller size and twisted margin, but an inconspicuous umbo is present near the apex. Paphiopedilum druryi (not seen) and P. exul also have a blunt umbo, but the latter species lacks the acute apices on the lateral lobes. This difference may not be consistent in natural populations. Paphiopedilum charlesworthii (as observed from display material of orchid exhibits) has a similar staminode with rounded umbo. The lunate staminode of P. fairieanum appears very similar to those of P. subgenus Barbata, but it has a distinct umbo, consistent with the other non-mottled-Ieaved species. The staminodes of P. hirsutissimum are unique in their truncate apices and rounded processes near the base of each anther sac in addition to the less conspicuous umbo. The biological significance of the umbo is unknown. The last group (Paphiopedilum subgenus Barbata) has basically trilobate lunate (moon-shaped) staminodes, with occasional development of lobules at the base or between the lateral lobes and apex. All bear some resemblance to insects, and none has a conspicuous umbo. The differences in staminode mor­ phology include presence or absence of extra lobules, the posture of the lat- 1984] ATWOOD: CYPRIPEDIOIDEAE 161

p

Fig. 2.8. Staminodia of Paphiopedilum. Intraspecific variation is shown in I, L, 0, Q, and W. All examples are shown about 1.7 times natural size. A. P. concolor. B. P. niveum. C. P. delenatii. Note very large size and orbicular shape. D. P. niveum. Compare with B. E. P. godefroyae. Note similarity to D. F. P. hirsutissimum. The umbo is only slightly developed in the mid region of the staminode. G. P. exul showing a well-developed umbo. Paphiopedilum villosum and P. insigne (not shown) are similar, but the examined material has trilobate apices. Paphiopedilum druryi also has a similar shape and a conspicuous um­ bo. H. P. spicerianum. The contorted margin obscures the shape, and a central somewhat elongated but inconspicuous umbo is present. 1. P. fairieanum. The umbo is visible in the sinus between the basal lobes (at the top). J. P. hennisianum. N. P. curtisii. 0. P. cilio­ lare. P. P. tonsum. Q. P. acmodontum. R. P. sukhakulii. S. P. purpuratum. T. P. viola­ scens. U. P. mastersianum. V. P. venustum, W. P. appletonianum. X. P, bullenianum. Y. P. linii. 162 SELBYANA [Vol. 7

erallobes, and pubescence. The lateral lobes of staminodes in P. !inii, P. bul­ ienianum, and P. appletonianum are somewhat laterally compressed with lobules only at the base. Paphiopedilum venustum, P. sukhakulii, and P. pur­ puratum exhibit similar staminodes, but the side lobes are extended. Paphio­ pedilum curtisii and P. ciliolare have broad, pubescent staminodes. Paphio­ pedilum barbatum, P. hennisianum, P. mastersianum, P. argus, P. lawrencea­ num, P. violascens, P. tonsum, P. ciliolare, and P. curtisii all produce one or two pairs of lobules between the lateral and apical lobes. The Gynoecium. The gynoecium of slipper orchids consists of three fused carpels. Three variously compressed stigma lobes are present, the dorsal of which is usually larger. The degree of confluence exhibited by the three stig­ ma lobes may be intraspecific ally variable. In Paphiopedilum acmodontum I have observed clones with orbicular stigmas, whereas in others the lobes may be distinct. Curiously, the lateral stigma lobes curl upward touching the sta­ mens and probably cause self-pollination. The ovaries are inferior and develop into capsules following pollination, but these may be fleshy and possibly baccate in Selenipedium (Dressler, 1981). The capsules of Selenipedium and Phragmipedium are oblong to linear while those of Cypripedium and Paphiopedilum are broader. Capsule maturation time is variable in the slipper orchids. The northern Cypripedium species require at most three months, while Paphiopedilum spe­ cies may require more than nine. Seed Morphology. Although seed morphology has been recently studied (Barthlott, 1976; Arditti et al., 1979; Ziegler, 1981), a thorough study has never been accomplished, probably because the seeds are not available. Most slipper orchids, like other orchids, have seeds with thin testas and minute embryos lacking endosperm. Selenipedium has been known to have crustose seeds (Pfitzer, 1888; Dressler, personal communication). Arditti et al. (1979) have recently researched seed characters in Cypri­ pedium and Paphiopedilum for their taxonomic utility. They found that C. acaule has the largest seeds with the largest embryos while seeds of Paphio­ pedilum are the smallest. Their data do not show variability of seed charac­ ters, and the number of plants per species examined was apparently less than eight. Their data show that among the Cypripedium species examined C. cali­ tornicum may have the longest or shortest seeds (excluding C. acaule) sug­ gesting that intraspecific variation is high. Cypripedium candidum has mark­ edly smaller seeds and embryos than the closely related C. parviflorum and C. pubescens, which implies that populational studies might reveal specific differences. They indicate correctly that relatively longer testas should cause greater seed buoyancy. Indeed, C. candidum with smaller testas occurs in prairie environments where the relatively abundant winds would be expected to minimize the requirement for larger testas. The data of Arditti et al. sug­ gest that the relationships of seed characteristics to environment are worthy of examination. Ziegler (1981) has shown that seed morphology is consistent at the ge­ neric level. Testas of Cypripedium seeds are inflated, and the conduplicate­ leaved slipper orchids have smaller and dark brown seeds with shorter testa cells. Phragmipedium (two species sampled) is unique in that the testa has an exterior brim. The testas of P. caudatum but not of P. longifolium are thick­ ened as in Selenipedium, and the infrageneric differences bring into question the presumed fundamental distinction between Selenipedium and Cypriped­ ium on the basis of seed structure. In this regard the seeds of C. irapeanum 1984] ATWOOD: CYPRIPEDIOIDEAE 163

and C. californicum should be investigated as both exhibit several character states otherwise unique to Selenipedium. Floral Color Patterns. Color patterns of slipper orchids appear ordinary to bizarre, drab to ostentatious. Color ranges include shades of violet to lilac, white, yellow, green, and nearly black, but rarely red and never blue. The simplest color pattern is found in Selenipedium which is rather drab and often has a pair of dark patches on each side of the labellum orifice. Cypripedium irapeanum and C. californicum also have a pair of dark patches on their labellum margins. Cypripedium color patterns are usually little more complex. Cypripedi­ um reginae has a white perianth with purple labellum. Cypripedium pubes­ cens and C. parviflorum have madder-purple perianths with yellow labella often spotted purple dorsally. Flowers of C. candidum and C. montanum differ from the latter two species mostly in the white lip, which provides a convenient species marker for taxonomists. Cypripedium acaule usually has a madder-purple perianth with concolor-purple labellum, but the latter is oc­ casionally dotted with white, may be white veined with purple, or pure white. A more complex pattern occurs in C. guttatum which is white with purple spots. In Phragmipedium the greens predominate with suffusion of purple or red. Phragmipedium boissierianum is nearly pure green, but P. longifolium and P. lindleyanum have some suffusion of purple. The latter species also has purple-lined petal apices. Phragmipedium wallisii has a white lip with faint lavender spots and a yellow rim. Phragmipedium warscewiczianum is basi­ cally yellow-green, but the petals are dark, and the lip is complexly veined in red. Ph1YllJmipedium schlimii is unusual in its white sepals and petals and pink labellum, and P. besseae is coral red. Ptlphiopedilum has the most complex color patterns of the slipper or­ chids. Ptlphiopedilum philippinense, P. laevigatum, and P. praestans have se­ pals lined with .purple to bronze and may have darker glandular processes near the base. The labella are colored various shades of brown to green. Paphio­ pedilum Iowii andP. haynaldianum have dorsal sepals lined or spotted with purple; the petals have large spots near the base, and the apices are lavender. Paphiopedilum parishii lacks purple coloring in the sepals, but the petals are extremely dark and have the spots expressed as black, pubescent warts. In section Cochlop~talum, all subspecies have very different color patterns in the sepals; the petals may have either vertical or horizontal blotches, and the lip may be lavender, white or greenish spotted with lavender. The color pat­ terns of the Himalayan species are very diverse and complex, especially those of the dorsal sepals. Paphiopedilum fairieanum has white, violet-lined dorsal sepals, lateral petals lined and reticulated with green and purple, and brown­ reticulated labella. Paphiopedilum insigne has variously colored sepals spotted with purple or brown. Paphiopedilum villosum and P. boxallii differ only in the very different color patterns, and both may be conspecific. Paphiopedi­ lum spicerianum and P. druryi both have central purple lines on the dorsal sepals and lateral petals. Paphiopedilum charlesworthii has a pink dorsal se­ pal with darker venation. Paphiopedilum subgenus Barbata exhibits basically green and white flowers lined with purple, and the petals are black-warted. Paphiopedilum mastersianum is notably different, however, in the predomi­ nation of red in the lip and lateral petals. Paphiopedilum subgenus Brachy­ petalum has the least complex color patterns in Paphiopedilum and has pink, 164 SELBYANA [Vol. 7

yellow, or white flowers dotted with magenta spots of various sizes which are apparently distributed randomly or in lines. Color patterns are extremely labile in some populations of Cypripedium acaule, which may have pink or magenta labella reticulated with darker ma­ genta, or occasionally spotted with white. In northern New England, entire populations occur with white lab ella., Elsewhere albino forms are rare. I am not aware of the existence of albino flowers among C. parviflorum or C. pu­ bescens, although the closely related C. kentuckiense, C. montanum, and C. candidum have white labella. Given the extreme intraspecific color variation as observed in Paphio­ pedilum insigne, P. villosum and Cypripedium acaule, very different color patterns between closely allied taxa are expected, and close affinities may be easily overlooked. Floral Fragrance. Odor production in the slipper orchids is difficult to assess because (1) sufficient vocabulary describing fragrance components is lacking, and (2) human noses are not equally sensitive. Furthermore, odors may be minimally produced in depauperate plants. Selenipedium palmifolium has been reported as having a strong, sweet fragrance (Crueger in Darwin, 1885). The flowers of most North American Cypripedium species are pleasantly scented, although the fragrance is often subtle. Cypripedium irapeanum is re­ ported by Stoutamire (1967, as observed by Dickinson) to "have a strong, sweet odor 'too sweet to be used in Chanel No.5.' " Among the western North American taxa, Cypripedium californicum is known to be "slightly but noticeably fragrant" (Anonymous, 1944), and Hooker (1891, 1893) noted fragrance in C. montanum and C. fasciculatum. In Michigan the fra­ grances of C. parviflorum and C, pubescens are markedly different, and I find this character to be reliably species-specific, although Stoutamire noted no clear difference in odor quality between these segregates of C. calceolus. Differences in odor perception or low odor production in inclement weather may account for these different views. Presumably most of the Eurasian spe­ cies are also fragrant. Nilsson (1979) reports fragrance in C. calceolus and Stoutamire noted a sweet fragrance in C. macranthum, but a funguslike odor in C. debile. Odor production, as perceived by the human nose, is less noticeable in Phragmipedium. Vogel (1962) indicates that the petals of some Phragmipedi­ um species have the odor of urine, and I have noted a faint urinous aroma to petal apices of P. iongifolium and P. caricinum. Phragmipedium schlimii has a distinctly spicy fragrance. Paphiopedilum has relatively few species with noticeable odors. Paphio­ pedilum subgenus Brachypetalum includes four species which often produce an odor similar to butyric acid. However, P. delenatii is exceptional in having a pungent and pleasant odor and appears to be the strongest scented species in Paphiopedilum. I have noted slight but pleasant odors in P. spicerianum and p, fairieanum, but otherwise odor production in Paphiopedilum is not common. Only Cypripedium calceolus has been analyzed by gas chromatography for odor composition (Nilsson, 1979). Besides ordinary fragrance attract­ ants, C. calceolus produces compounds corresponding with substances in Andrena bee pheromones. The site of odor production in Cypripedium arietinum, C. pubescens, C. parviflorum, and C. candidum is mostly in the sepals and lateral petals with 1984] ATWOOD: CYPRIPEDIOIDEAE 165 little or no production in the labellum and column (Stoutamire, 1967). How­ ever, Swanson et al. (1980) believe that the glandular trichomes of the ovary may contribute to odor production.

VARIATION IN PLANT SIZE One of the most striking intraspecific or intracomplex differences is size. In northern Michigan some Cypripedium pubescens clones found in bogs and woodlands remain reproductively immature before attaining a height of at least three decimeters, whereas others growing in open roadsides constructed on crushed limestone may flower when less than one decimeter tall. It is not known if the observed size variation is controlled by genetic or by environmental factors, but C. planipetalum (C. calceolus var. planipeta­ lum) rarely exceeds 20 cm high (Luer, 1975). Phragmipedium longifolium exhibits a similar extreme in size (Atwood and Dressler, in prep.). Some plants of embankments or landslides produce leaves up to 90 cm long and 5 cm wide, but the smaller plants of river beds produce graminaceous leaves up to 50 cm long and 1.2 cm wide. Both ex­ tremes have correspondingly large or small flowers, and intermediate popula­ tions have been found. The variation is genetic, since the size characteristics do not change significantly in conservatory-grown specimens. The larger plants occur on disturbed embankments but never are subjected to flash­ flooding, whereas the dwarf populations always occur within the high water zone and are subjected to periodic flash-flooding. The narrower, more wiry leaves with more closely spaced vascular bundles are less breakable and seem better adapted to flash-flooding than the larger less durable leaves of plants found on the embankments. When hybrid populations between the extremes are found within the high water zone, all plants are small, although the floral morphologies may vary between the extremes. At least two striking examples of genetically controlled size variation are known in Paphiopedilum. Paphiopedilum victoria-regina subsp. victoria­ regina may possess leaves up to 40 cm long and 6 cm wide, while subsp.primu­ linum rarely reaches 33 cm long and 4 cm wide (Wood, 1976). The dwarf vegetative features are apparently consistent in subsp. primulinum, and the smaller floral size accompanies the smaller leaf size. The Paphiopedilum phi­ lippinense complex has similar size variation. Paphiopedilum philippinense is typically a large plant with leaf length often exceeding 5 dm, while leaves of P. laevigatum rarely exceed 25 cm at maturity. Since leaf width increases correspondingly with length, area variation is more striking than the linear variation but is more difficult to assess. Similar size variation seems to dis­ tinguish members of the P. glanduliferum complex, which may be best han­ dled as a single species with several subspecies. Striking differences in plant size and in shape and color patterns of flowers, which are evident even among closely related taxa, are taxonomically confusing. The most useful features on which to base a classification of the Cypripedioideae are those which are unique. Those which I would emphasize include inflorescence features, sepal vernation, perianth articulations, light windows, tangent lateral lobes in labella, and umbos of staminodes. In un­ usual cases floral color patterns may be given some weight when correlated with other features. In this regard the presence of darkened patches on la­ bella of Cypripedium irapeanum and C. californicum support several other morphological features as evidence for relationship with Selenipedium. 166 SELBY ANA [Vol. 7

FLORAL MORPHOLOGY AND POLLINATION BIOLOGY Slipper orchids have classic deception flowers since they off r no food reward for their pollinators. The variability of floral morpholog' s may be related to the deception flower syndrome. The requirements for a successful pollination system have bee reviewed most recently by Nilsson (1979). The necessary events include: (1) attraction and subsequent approach of the pollinator, (2) alighting behavi r, (3) en­ trance into the labellum, and (4) captivity and escape. Two visits e required by the same visitor in order to pollinate one flower successfully. Little is known of the significance of visual cues to pollinator attraction and alighting behavior. The maximum distance at which a pollina r may be attracted is undoubtedly affected by flower size, inflorescence siz ,inflores­ cence number, distance at which odors may be perceived, and open ess of the habitat. Movement of flowers in a breeze may increase probability f pollina­ tor attraction, especially if supported on long peduncles. In Paph 'opedilum the complex floral patterns undoubtedly serve some function in pollinator attraction. Furthermore, the similarity of staminodes to insects s ggests the possibility of pseudocopulation or a pseudo-prey relationship. Nil son noted that bees are attracted to the staminode before falling into the I bellum of Cypripedium calceolus. Odor production undoubtedly serves in directing pollinator behavior. Stoutamire has suggested that odors orient an approaching vis ito , but the presence of pheromone compounds in addition to ordinary fragran e attract­ ants in Cypripedium calceolus (Nilsson, 1979) suggests that odo s may be effective at greater distances. The pollinator must be enticed to enter. In Cypripedium the ollinators usually slip into the labellum. In Phragmipedium longifolium var. hartwegii a pair of glandular dots on the infolded lateral lobes attract pollina ors which lose their footing and slip into the lip (Dodson, 1966). Although t e dots are absent from Paphiopedilum, the rounded umbo of staminodes as 0 served in P. uillosum and P. insigne may serve a similar function. Once the pollinator has entered the lip, it must be retaine a d allowed to exit in such a way that a pollen mass is collected. The in 01 ed lateral lobes of the labellum appear to restrict the pollinator's choice of ex ts. Escape must be made via the stigma and pollen mass at one of the t 0 b sal exits. Nilsson indicates that the pollinator wedges itself between the s igzha and lip and that the tight passageway is important to pollen depositio . In Phragmi­ pedium the basal hollows seem well adapted for pressing the poll'nator against the stigma. The light windows observed in the Cypripedium calceolus com­ plex have been presumed to help illuminate the exit for the pollinator; how­ ever, Daumann (1968) has indicated that successful exits are made even by covering the light windows. Despite this evidence, light windows as observed in Cypripedium, Phragmipedium schlimii, Paphiopedilum lowii, P. haynaldia­ num, and P. parishii may increase the probability of successful pollination. The pollinator is apparently not offered food reward by slipper orchids. The idea that trichomes in the labellum supply food has been dispelled by Nilsson for Cypripedium calceolus. When contacting the oily hairs, the polli­ nator grooms itself, as though trying to eliminate the oil. Nilsson believes that oil serves to cover the sensory organs of the pollinator, thus dulling its senses and somehow causing behavior change. Despite the lack of food re­ ward the first successful pollination by a vector can occur only with a second visit. 1984] ATWOOD: CYPRIPEDIOIDEAE 167

The pollination system of Cypripedium pubescens is hampered by pres­ ence of spiders (Newhouse, 1976; Stoutamire, 1967), which prey upon the pollinators. Furthermore, snails and other non-pollinators may damage the flowers when trapped. No accounts of pollination in Selenipedium are known. Pollination may be rare among some populations as all mature plants of one population of S. chica lacked seed capsules when I visited it in 1977 or 1978. Most observations of pollination have been made on Cypripedium (see Stoutamire, 1967; Newhouse, 1976; Nilsson, 1979; and van der Pijl and Dod­ son, 1966 for reviews). In Cypripedium the labellum may trap any small ar­ thropod, but only solitary bees (families Andrenidae, Halictidae, and Mega­ chilidae) and bumble bees (Apidae) are known to be effective pollinators. Only the halictid bee Ceratina acantha has been observed emerging from C. californicum with pollen smears, although other visitors are known (Kipping, 1971). In the Cypripedium calceolus complex, pollen smears have been found on Andrena, Agapostemon, Ceratina, Lasioglossum, and Osmia. In C. candi­ dum, a pollen smear was observed on Andrena by Stoutamire (1967), and Catling and Knerer (1980) found pollen smears carried by females of the genera Augochlorelia, Halie tus, and Dialictus. Although Raffill (1913) re­ ported bumble bee pollination in Cypripedium reginae based on an observa­ tion of a cultivated specimen at Kew, the exit of the orchid is too small for the presumed pollinator. More likely, C. reginae is pollinated by smaller bees, and Guignard (1886) has reported Megachile carrying pollen smears of this orchid. Newhouse reported the visitation of C. reginae by the syrphid fly Sericomyia, but no pollen smear was found. Cypripedium acaule is pollinated by females of at least one Bombus species, and C. arietinum is pollinated by Lasioglossum (Stoutamire, 1967). Apis mellifera has been observed with pol­ len smears in C. pubescens, but I have observed dead specimens in the label­ lumas has Ames (1932). Stoutamire (1967) speculates that C. debile is pol­ linated by fungus-seeking insects, since the flowers have a fungal odor. Within Phragmipedium, pollination has been observed only in P. longi­ folium (Dodson, 1966). Dodson placed plugs at the exits and found many in­ sect species had been caught in the labella, but most were too small for effec­ tive pollination. He noted that the halictid bee Chlerogelia and syrphid flies were found with pollen smears. The resemblance of odors of P. longifolium to urine may not be superficial, as sweat bees (also halictids) are attracted to human perspiration, which is similar to urine. Despite the intriguing possi­ bilities, there are no validated reports of pollination of the P. cauda tum com­ plex. Snail pollination has been implicated by Knuth (1909), but this is highly unlikely. Although van der Pijl and Dodson (1966) have suggested that Paphio­ pedilum is pollinated by flies as evidenced by the presence of warts and spot­ ting patterns, no accounts of pollination in natural populations are known. I have observed flies trapped in the lips of cultivated P. concolor and P. cilio­ lare, but the insects seemed to make no attempt at escaping by the basal exits and were hopelessly trapped. Given the bizarre forms observed in Paphioped­ ilum, pollination studies can only lead to some surprises. However, pollina­ tion may be rare as suggested by the great longevity of conservatory-grown flowers, and field study will likely tax the patience of its observers. Aside from the few accounts of pollination success in Phragmipedium longifolium by syrphid flies, there is little evidence for myophily in the Cy­ pripedioideae. The reports of pollination suggest that Cypripedium is polli- 168 SELBYANA [Vol. 7 nated by related groups of bees restricted in morphology and behavior pat­ terns, but this generality cannot be extended to the conduplicate-Ieaved genera. Pollination success is undoubtedly low when all visitations are noted, as indicated by Nilsson. However, it must be sufficiently high to maintain wild populations. Perhaps the rarity of Cypripedium calceolus in England is in part accounted for by low pollination success. Newhouse believes that in­ clement weather was responsible for lack of visitation during her study in 1975, and the prevalence of capsules which I have observed in more climati­ cally favorable years supports this view. At one C. acaule station in Michigan I found that the pollen masses had been disturbed in several examined flow­ ers, and pollen smears were observed on the stigmas. Observations of C. pu­ bescens growing abundantly on roadsides and residential lawns in northern Michigan also attest to success of the pollination system. Veitch and Sons (1889) believed that pollination success was low in Paphiopedilum since very few wild-collected plants entered their nursery with seed capsules. This was clearly a poor measure of success, as capsules are usually removed from plants intended for cultivation. Extreme inbreeding has been noted for various slipper orchids, and in these plants floral appearance and odor production would be of little impor­ tance. Veitch and Sons (1889) report self-pollination in Phragmipedium schlimii, which was a minor economic nuisance to their firm since the flowers failed to mature properly. Phragmipedium lindenii is an autogamous segre­ gate of the P. caudatum complex. Day (1972) reported self-pollination in Paphiopedilum lowii and P. mastersianum, but pollination could have been accomplished by an unkn3wn vector. I have noticed capsule formation in P. argus infested with mealy bugs. The most reliable account of autogamy has been made by Catling (1980) on Cypripedium passerinum, in which the anthers come into contact with the stigma with age. However, as Catling noted, allogamy is not excluded in younger flowers. Although outcrossing is encouraged by the entrance-exit system, multi­ flowered inflorescences and large plants would be expected to increase in­ breeding, since more flowers would be available for pollination from the same clone. Small plants with single flowers would be expected to be outcrossing. In Selenipedium, Cypripedium irapeanum, several Phragmipedium species, and Paphiopedilum victoria-regina, all plants flower successionallr with usu­ ally one flower open at once, a feature which encourages allogamy!. The precise size and shape relationships between pollinator ~d flower together with observed variability in natural populations sug~est that a switch to a different sized pollinator could change the floral morPhology of a population as Nilsson believes occurs in the Cypripedium calc~olus com­ plex. This may explain much of the intraspecific size variatio observed among local populations. Alternatively, Heinrich (1975) has sug ested that differences in floral appearance may delay the pollinator from earning to avoid deception flowers. Extensive field studies will be required t test both hypotheses. 1984] ATWOOD: CYPRIPEDIOIDEAE 169

CHAPTER III: COMPARATIVE ANATOMY

Works on slipper orchid anatomy include those ofPfitzer (1903), Swamy (1948), Davis (1966), Rosso (1966), Abe (1972), Rao (1974), Atwood and Williams (1978, 1979), and Williams (1979). Many other references are cited in these works.

VEGETATIVE ANATOMY Rosso (1966) studied the vegetative anatomy of 22 species of slipper orchids including 2 Selenipedium, 11 Cypripedium, 4 Phragmipedium, and 5 Paphiopedilum. Table 3.1 summarizes his findings. Anatomical features are conservatively similar at the generic level except in Cypripedium. Cypripedium irapeanum has eight anatomical features found in Selenipedium, and C. cali­ fornicum has two. Although Rosso's sample size in the conduplicate-Ieaved genera is small, the diversity of species selected suggests that a larger sample size would likely have been redundant. Except where otherwise indicated the following account is based on Rosso. The roots of Selenipedium, Phragmipedium, and Paphiopedilum have a number of features not shared by Cypripedium. Except in Cypripedium, a "well-defined, multiseriate layer of highly-thickened sclerenchyma cells im­ mediately adjacent to the endodermis" is found. Selenipedium, Phragmiped­ ium, and Paphiopedilum roots have piths which are absent in Cypripedium (except for small piths in C. irapeanum and C. californicum). All genera ex­ cept Cypripedium have an exodermis, multiseriate pericycle, and velamen, (however, an exodermis and velamen are also present in C. irapeanum). The velamen of Selenipedium and Cypripedium irapeanum is uniseriate, whereas it is multiseriate in the conduplicate-Ieaved genera. The number of alternating xylem-phloem groups is highest in Selenipedium palmifolium, somewhat lower in Phragmipedium and least in Paphiopedilum and Cypripedium (not shown in Table 3.1). The number of xylem-phloem groups within Cypriped­ ium appears to be highest in C. irapeanum and C. californicum, somewhat less in the remaining species, and C. guttatum has the least. Tracheary ele­ ments comprise vessels and tracheids in Selenipedium, Phragmipedium and Cypripedium, but Paphiopedilum apparently has only tracheids. The rhizome is difficult to distinguish from the aerial stem except in Cypripedium, but Rosso treats them separately. The endodermis of the rhi­ zome appears to be simple in the plicate-leaved genera but multiple in the conduplicate-Ieaved genera. The vascular bundles are collateral in Seleniped­ ium and fused in Cypripedium (except C. irapeanum and C. californicum where they are amphivasal, a feature also shared by Phragmipedium and Paphiopedilum). The cortex is well developed in Cypripedium, including C. irapeanum and C. californicum but is less well developed in the remaining genera. In the aerial stem of Phragmipedium and Paphiopedilum, Rosso found a distinct endodermis, but endodermal cells are scattered and not suberized in the plicate-leaved genera. Stomates were found in aerial stems in the plicate­ leaved genera but not in the conduplicate-Ieaved genera. The vascular bun­ dles are collateral in Selenipedium and Cypripedium but amphivasal in the 170 SELBYANA [Vol. 7

Table 3.1. Distribution of anatomical features according to Rosso, 1966 (the di stri bution of vascu1 ar bund1 es in the shoot has been rei nvesti gated). Parentheses indicate nearly border1 i ne character states.

Root Aerial Stem Leaf .... ." '" +> N c '" s.. '" .'">: ...c ~ , s.. "'0 Cl) ... Cl) Cl) .~Vl ... :J: Cl) u c .c~ ".;::; c Cl) .0 :;!.~'" Cl) Cl) s.. .,,+'~-- ."C Cl) :J: :;:;'" '"s...... Qi :; +> "- ~ c~~ co -" Cl) "- ou 5 c ::I ::I s.. ~ .o~~ .o~ "- ~ Cl)." Cl) "- :J: '" '" ~Cl) Cl) "~ s..s..","'''' s..s..'" c c "E'" uc 'OCl) s.. "E "'Cl)'" "'Cl) L s.. "'Cl) ,., ... "- Cl) Cl) ~ ... > ~.c .,c 5 5 Cl) u-" U'" '" ." ." :::s(O.,... ::1"- ."., '" '" .., "~ u 0S:"S: .c 0 0 U~.<: u"~ 0'" 0 s.."~ ... .., .., "'~ "- ",s.. ."., Qj"' Qi'" Cl).c .,., ~., c "'05 cs.. Species :> :> .;:: a. ... a.'" :> a: ~ w :>u'" :>"'., "- w "-

C~Qri Qedi ym

~ + 0 + arietinum (-) + +

~a1ifornicum + + 0 + +

candidum + + fasci culatum + + gutta tum + +

i raQeanum + + + + + + + +

!lJacranthym - + + rna rgari taceum + +

Qubescens + 0 + reginae + +

Se1eniQediu!!l

~ + + + + + + + + + + l2almifoliym + + + + + + + + +

Phragmi Qedi Y!!I

~ + + + + + + + + + + lir.!l!wii + + + + + + + + + + ]ongifolium + + + + + + + + + +

sch1imii + + + + + + + + + PaQhioQedilym

bellatu1um + + + + + + + + + insigne + + + + + + + + +

ni veum + + + + + + + + +

[Qthschil di anym + + + + + + + + venustum + + + + + + + + + 1984] ATWOOD: CYPRIPEDIOIDEAE 171

remaining genera. The aerial stem of Selenipedium chica has vascular bundles only in the periphery, and the center is filled with a white pith similar to that of Sambucus. In all northeastern Cypripedium species which I have ob­ served, the vascular bundles are scattered throughout the stem. However, I have noted that C. californicum has peripheral bundles as does Selenipedium, but the center of the stem is hollow. Cypripedium irapeanum is anatomically similar to the northeastern cypripediums in this regard. Rosso noted several differences in leaf anatomy between the condupli­ cate-Ieaved and plicate-leaved genera. The epidermal cells are smallest in Selenipedium and largest in Paphiopedilum. In the plicate-leaved genera,' epidermal cells over the vascular bundles are often much elongated. Stomates are present on the abaxial leaf surface of all species, but on the adaxial surface they were found only in Selenipedium, Cypripedium ira­ peanum, C. californicum, and C. reginae. Williams (1979) found subsidiary cells inPhragmipedium longifolium, P. lindenii, P. sargentianum, Selenipedium chica, and S. palmifolium but not in Cypripedium acaule nor in a Paphioped­ ilum hybrid. Palisade mesophyll appears to be lacking from Cypripedium, and presence of palisades in Phragmipedium and Paphiopedilum is variable, even within the same leaf (Rosso, 1966). Rosso also noted the similarity in sclerenchyma abundance between Cypripedium irapeanum and Selenipedium, a feature not shared by other Cypripedium species. In Cypripedium there is no endodermis associated with the vascular bundles except in C. irapeanum. The endodermal cells are U-shaped in Selenipedium, O-shaped in Phragmi­ pedium, and U-shaped in Paphiopedilum. The endodermal cells are suberized only in the conduplicate-Ieaved genera. Atwood & Williams (1978, 1979) studied the distribution of sculpturing patterns of epidermal cells in the conduplicate-Ieaved genera. We found that Paphiopedilum subgenus Barbata (excluding P. fairieanum) and Phragmi­ pedium schlimii have micropapillose or rugose epidermal cells. The remaining species sampled lack sculpturing. Additional Anatomical Evidence. Several anatomical features seemed worthy of further investigation, so the following features have been studied: (1) the distribution of sinuous anticlinal walls of epidermal cells, (2) epidermal cell volume, (3) distribution of palisade mesophyll, (4) spacing of vascular bun­ dles in leaves, and (5) stomatal development. The methods employed for observation are outlined as follows. (1) The adaxial epidermal cells wer~ prepared in the plicate-leaved genera by scraping away the subepidermal tissues with a razor blade. A wet mount of the epi­ dermis was made for observation under a Wild M20 binocular microscope. In the conduplicate-leaved genera the epidermis was simply removed with a razor blade, mounted and observed. Camera lucida drawings were made from several preparations. (2) Cell volume was measured by multiplying length, width, and height of epidermal cells. Length and width measurements were averaged over a minimum of ten cells. Height was relatively uniform in ob­ served transverse sections. In the plicate-leaved genera height was measured by focusing the objJctive on the top and bottom of epidermal cells and re­ cording the differenbe. In some cases cell volume was determined from data provided by Rosso. (3) Distribution of palisades in the conduplicate-Ieaved genera was determi~ed from thin hand sections stained lightly in toluidine. Among several mottled-leaved species of Paphiopedilum, hand sections of light and dark patches were prepared separately. (4) To measure vascular bundle spacing, leaves were scraped until the vascular bundles were clearly in 172 SELBY ANA [Vol. 7 view. Visibility of vascular bundles was enhanced by eliminating air spaces in a desiccator and subsequently by normalizing air pressure with the weighted leaf submerged in water. The spacing of vascular bundles was measured by placing a millimeter ruler across the leaf, counting a minimum of five bun­ dles, and expressing bundle spacing in bundles per millimeter. (5) Stomatal development was determined in young shoots of Phragmipedium X sedenii prepared according to Williams (1979). The data for epidermal cells are recorded in Table 3.2. The anticlinal walls are sinuous in the plicate-leaved genera and straight or simply curved in the conduplicate-leaved genera, a feature also noted by Rosso (see Figure 3.1). The anticlinal walls of Cypripedium acaule are shallowly sinuous despite Solereder et al. (1930) and references cited therein. Epidermal cell volume is smallest in Selenipedium chica but larger in S. palmifolium. The cells of Cy­ pripedium are generally larger than those of Selenipedium, except in C. arie­ tinum. Epidermal cells are largest in Paphiopedilum, particularly among P. subgenus Brachypetalum, P. section Pardalopetalum, and the "earless" multi­ flowered species. The epidermal cells of Phragmipedium are nearly as small as those of the plicate-leaved genera, except in the Phragmipedium cauda tum complex, where they are comparable in size to those of Paphiopedilum. Tables 3.3 and 3.4 summarize palisade distribution among Paphiopedi­ tum. Palisades are unknown in the plicate-leaved genera (Rosso), and within Phragmipedium I noticed a very slight difference in cell organization in meso­ phyll of the P. cauda tum complex. Palisade mesophyll is absent or poorly differentiated in Paphiopedilum lowii and P. insigne. Presence of palisades does not follow any taxonomic pattern in Paphiopedilum, and variation is perhaps attributed to conservatory conditions. However, distribution of pali­ sades may offer at least a partial explanation for mottled leaf patterns. The dark patches of at least the strongly tesselated species have markedly greater development of palisades (Fig. 3.2) than do the light patches. This is most striking in P. concolor but not in the weakly tesselated P. victoria-regina subsp. liemiana. Pfitzer (1903) may have known this as he indicated that the light patches contain mesophyll with more intercellular air spaces, but he did not verbalize the difference in palisade distribution. Rosso also noted palisade differences in singular leaves of Paphiopedilum bellatulum, which probably correspond to the mottled patterns. The data for bundle spacing are more fragmentary in the plicate-leaved genera and are given separately from those of the conduplicate-leaved genera (Fig. 3.3,3.4). Since spacing is affected by shrinkage in herbarium specimens, Fig. 3.3 shows bundle spacing as determined from live, pickled, and herbarium material. Among the plicate-leaved species, Seienipedium chica has the closest vascular bundles, followed by Cypripedium irapeanum. Most Cypripedium species have less than 6 bundles per millimeter. Within the C. calceolus com­ plex, C. candidum has the closest spacing of vascular bundles, and this spe­ cies may be readily distinguished on this feature. Among the conduplicate­ leaved genera Phragmipedium has the closest vascular bundles, Paphiopedilum the most distant, and there is little overlap. The riverine Phragmipedium spe­ cies have the closest vascular bundles, and the P. cauda tum complex exhibits relatively broad spacing. Within Paphiopedilum spacing is closest in the non­ tesselated species, wider inP. section Cochlopetaium, and widest in subgenera Brachypetalum and Barbata. Stomates have similar developmental patterns (Fig. 3.5) in Phragmiped­ ium X sedenii as in other orchids (Williams, 1979). Rows of abaxial proto- 1984] ATWOOD: CYPRIPEDIOIDEAE 173

o E

Fig. 3.1. Camera lucida drawings of adaxial epidermal cells. The scale equals 0.2 mm ex­ cept in the larger magnification of A, which is 0.05 mm (0.2 mm in the insert). A. Seleni­ pedium chica. B. Cypripedium californicum. C. Cypripedium guttatum. D. Cypripedium acaule. E. Phragmipedium longifolium with small epidermal cells relative to those of other conduplicate-Ieaved species. F. Phragmipedium warscewiczianum. G. Paphiopedilum par­ ishii. H. Paphiopedilum victoria-regina subsp. liemianum. I. Paphiopedilum concolor. 174 SELBYANA [Vol. 7

Table 3.2. Characteristics of leaf epidenna1 cells. Data marked with an asterisk (*) are based on Rosso (1966) .

v> v> .... ';;; ~ :;t QJ~ ...Um Um :s:'" ~::; .s::'" ';;; ~::; .s::"'" ';;; "'~ " "'~ t:", ...,,,"..., .!= til E", ...,,,"..., .~ (I) 0. ... ~uo " "'E 0.'" ~"uo ~§ ~V> ~V> .~ " -0" .~ " .~~ ...,,, .~~ ...,,, 0.0 us.. ".~ 0.0 us.. UJ> "'" UJ> "'" ".~ S ecies '" 0. S ecies '" 0.

~~g[igediYm Paghiogedilum (Cont' d.)

acau1e .000,476 (+) de1anatii

arietinym .000,088 + exu1 .014,1

ca 1 i forni cum .000,292 + fai rieanum .022,7

candidum (immature) + godefro~ae

guttatum .000,129 + ha~na 1 di anum .009,83

iraQeanum 000,135 (+) hennisianum .002,26 +

~a[~if1orum .000,14 + hirsutissimum .003,63

Qybescens .000,284 + insigne .002,69

reginae .000,322 + 1 aWrenceanum .003,50 + Phragmi Qedi um 1 inii .002,57 + boissierianum .000,395 10wii .005,67

~ari cinum .000,227 (-) mastersi anum .006,39 + cauda tum .005,65 niveum .033 10ngifo1 ium garishii .054,1 ssp. gracile .000,127 ghil i gginense .007,29

ssp. 10ngif • .000,098 Qraestans .021,6

Qearcei .000,229 (-) QurQuratum .001 ,81 +

sch1 imii .000,392 + randsi i .014

wallisii .009,48 rothschil di anym .004,04

wars~ewi~zianym .012,7 sgi&eri anum .003,16 ei!ghiQQ~gi]ym sukhaku1 i i .006,33 +

acmodontum .003,19 + tonsum .002,69 +

aQQ1etonianum .002,04 + venus tum ',003,18 + argus .006,92 + vi ctori a-regi na .002,71

~ + .002,17 bellatu1um .030 .002,36

by]l eni anum .001 .79 + violascens .002,72 +

ci1 io1are .003,99 + ~~leniQediym

conco10r .010,6 chica .000,004,560 (+)

curtisii .001,4 + galmifoliym .000,098,260