P1. Syst. Evol. 175" 161-223 (1991) --Plant Systematics and Evolution © Springer-Verlag 1991 Printed in Austria The phylogeny and evolution of the Pterygodium-Corycium complex ( Coryciinae, Orchidaceae) H. KURZWEIL, H. P. LINDER, and P. CHESSELET Received August 13, 1990; in revised version October 18, 1990 Key words: Angiosperms, Orchidaceae, Coryciinae, Ceratandra, Evota, Evotella, Ommatodium, Pterygodium, Anochilus, Corycium.- Morphology, phylogeny, cladistics, phytogeography, speciation, classification.- Flora of southern Africa. Abstract: The leaf-anatomy, palynology, seed-morphology, vegetative morphology and especially the highly complicated floral morphology of the Coryciinae s. str. (Diseae: Orchidoideae: Orchidaceae) are described and illustrated in detail. On the basis of these characters the presumed phylogeny, based on a rigorous cladistic analysis, is presented. The cladistic biogeographical analysis of the Coryciinae s. str. shows that it is a member of the Afrotemperate Track, with a pattern of vicariance events typical of the members of this track. An analysis of the patterns of speciation shows that allopatric speciation appears to be rare, and that parapatric speciation across edaphic boundaries may be the most important factor. Proceeding from the information presented, a new classification of the group is proposed in which we recognize the four genera Ceratandra, Evotella, Pterygodium and Corycium. The new monotypic genus Evotella comprises a species originally described as Pterygodium rubiginosum. The three species of the genus Anochilus are transferred to Corycium and Pterygodium. P. magnum, which was originally described as Pterygodium but was transferred to Corycium lately, is placed in a monotypic section of Pterygodium. The Coryciinae have amongst the most bizarre floral structures k n o w n in the orchids. A l t h o u g h the sepals and petals are relatively simple, the lip is highly specialized and bears a complex appendage. While lip calli are fairly c o m m o n in the orchids in general, the complex lip appendage of the Coryciinae, mostly larger than the lip blade itself, is unique in the family. The gynostemium is twisted and turned out of recognition, so that the structures will provide even the m o s t eminent orchidologists with headaches (DRESSLER 1981 : 197). There are a few detailed descriptions of these complex structures available (in particular that of VO~EL 1959), and none that are comparative for the whole group. BoLus (1888, 1 8 9 3 - 1896, 1911, 1913, 1918) illustrated a range of the gynostemia, but his interpretations were often not accurate. The complex floral structures are related to the pollination by oil-collecting bees (STEINER 1989, and unpubl.). The origins of these structures are not known. VOCEL (1959) suggested that the ap- 162 H. KURZWEIL 8,~ al.: ..,.a ¢-q o~ 0..l, .< [... ct~ © e,.) C, © Z: r..9 ° ,..., <~o z ,...a ,..2 © Z < Phylogeny of the Coryciinae 163 .~ ~ ~ ~ . ~.~.~ N ~ 164 H. KURZWEIL• al.: pendage may be derived from the abaxial stamens. The alternative is that they are new structures. The generic classification of the Coryciinae has generally been unstable. While only a few monographic treatments of the entire subtribe Coryciinae have been published (KRXNZLIN 1897, SCHLECHTER 1898), several revisions of the southern African species are available (BoLus 1888, 1893--1896, 1911, 1913, 1918; ROLFE 1913; LEWIS 1950; STEWART & al. 1982). The different classifications proposed so far differ mainly in the recognition of Corycium, which was either maintained as a separate genus or included as a section in Pterygodium. Also the exact placement of several "problem species" caused difficulties in previous classifications. These were then either shifted around between the major genera or split off as separate genera (see Table 1). ROLFE (1913) separated Ceratandragrandiflora and C. globosa into Ceratandropsis, but his main argument, the presence of resupinate flowers, is based on an incorrect observation. Ceratandra bicolor and C. harveyana differ from other Ceratandra species by having resupinate flowers and a prominent lip appendage and were separated by some authors (ROLFE 1913, LEWIS 1950, SCHELPE 1966, BOND & GOLDBLATT 1984, STEWART& al. 1982) into Evota. As shown below, the two species are linked with the remainder of Ceratandra by a number of characters. The apparently isolated species Ceratandra venosa was included in Ceratandra (SCHLECHTER 1898), Evota (ANTHONY & SCHELPE 1982, STEWART & al. 1982), Pterygodium (LINDLEY1833, BOLUS 1893--1896, KRS.NZLIN 1897) and Corycium (ROLFE 1913) at different stages. The species originally described as Pterygodium inversum and P. flanaganii were separated into Anochilus by ROLFE (1913) on account of having non-resupinate flowers. A third species was later added (Anochilus hallii; Pterygodium hallii in the present paper). The generic concept of Anochilus, however, appears doubtful, since the species are quite different in their flower structure (see also VOGEL 1959). On the basis of having flowers with erect anthers, LINDLEY (1833) separated Pterygodium volucris into Ommatodium. The genus is not upheld here since several advanced characters are shared with other Pterygodium species. The generic delimitation of Disperis has never been problematic, and the genus is regarded by all authorities as being isolated within the Coryciinae (KRANZLIN 1897, SCHLECHTER 1898, ROLFE 1913, VOGEL 1959, LINDER 1986). The genus is here considered as being the sister group of the Coryciinae s. str., and will be dealt with in a separate paper (MANNING& LINDER 1991). The subtribal classification has been more or less stable, with the Coryciinae being recognized as a distinct group since BENTHAM& HOOKER (1883). In spite of being extraordinarily distorted, the architecture of the gynostemium (on which much of the orchid taxonomy relies) clearly indicates the systematic position of the Coryciinae in the tribe Diseae within subfam. Orchidoideae (DRESSLER1981). Most earlier workers also included the Corycium-Disperis-alliance in the OrchisHabenaria-Disa-group, although the names for both these taxa varied (Corycieae: Ophrydeae, BENTHAM & HOOKER 1883; Corycieae: Ophrydinae, PFITZER 1889; Disperideae." Ophrydoideae, SCHLECHTER 1926; Disperidinae: Orchidoideae, SENGHAS 1973-- 1974). Here we investigate only the Coryciinae s. str., that is, excluding Disperis. The group includes 39 species, distributed in southern Africa with one outlier species (Pterygodium ukingense) recorded from Tanzania. All the species inhabit fynbos Phylogeny of the Coryciinae 165 and marshes in the Cape Province of South Africa, and thrive in m o n t a n e grassland further north (South Africa's summer-rainfall area, and in Tanzania). The Coryciinae are part of the remarkably rich southern African flora (GOLDBLATT 1978; GIBBS-RUSSELL 1985). GOLDBLATT (1978) summarized the probable speciation mechanisms that m a y have led to this rich flora, and suggests that "Geographical speciation is undoubtedly of primary importance in the southern African flora ...". In contrast, LINDER (1985) suggested that allopatry m a y be incidental in speciation in the Cape flora, and that the steep ecological gradients m a y be more important. Possibly due to the "morphological-geographical m e t h o d " (DAVIS& HEYWOOD 1963) and the expectation of allopatric speciation, botanists generally emphasize allopatry in their analysis of speciation (i.e. GOLDBLATT 1982, BREMER 1976) while not listing cases of parapatry or sympatry. In addition, sistergroups have rarely been established using rigorous cladistic methodology (but see BREMER 1976, GOLDBLATT 1987, LINDER & VLOK 1991). In the present paper we attempt to resolve the taxonomic problems of the Coryciinae by a cladistic analysis of all available morphological, anatomical and palynological data. On the basis of the resulting cladogram, we propose a new generic classification for the subtribe. Material and methods The gross morphology of vegetative structures was generally examined on herbarium material held in the Bolus Herbarium, University of Cape Town (BOL). Material for the detailed morphological and anatomical investigations was fixed and stored in Kew Cocktail (H20 :alcohol:formaldehyde: glycerine = 15.3 : 16:1 : 1) or FAA (70% ETOH:40% formalin : absolute glacial acetic acid = 85 : 10 : 5). Both newly collected material and the existing holdings of the Bolus Herbarium were used. Pollen and seed samples were generally taken from herbarium sheets from BOL. The voucher specimens are listed in the appendix. Macrophotographs of living flowers were taken with a WILD photomicroscope. Floral morphology was studied from liquid preserved material by either drawing with a WILD dissecting microscope, or observing and photographing in a CAMBRIDGE STEREOSCAN S 200 scanning electron microscope after being critical-point-dried according to a standard preparation technique (GERSTBER~ER& LEINS 1978) and sputter-coated with Au/ Pd. Dry seeds were directly mounted on aluminium stubs, and without further preparation coated with Au/Pd in a BALZERS sputter coater. Samples were viewed and photographed in a CAMBRIDGE STEREOSCAN S 200 scanning electron microscope at 10kV. Characters studied Morphology and anatomy Underground organs. All the species of the Coryciinae s. str. are terrestrial and inhabit areas with a seasonally dry climate. They are deciduous and survive the dry season as dormant underground storage organs. Two distinct groups of such organs can be distinguished (see also Fig. 1): (1) In Pterygodiurn and Corycium the underground organs are of the tuberoid type ("root-stem tuberoid" sensu DRESSLER 1981). Such root-stem tuberoids are well known from European genera of the closely related tribe Orchideae where their anatomy and seasonal growth cycle has also been studied in detail (e.g., SHARMAN 1939, where also the older literature on this topic is reviewed): they are mostly made up of root-tissue, but the apical portion (termed "neck") has a sheath of root tissue around a core of stem tissue 166 H. KURZWEIL c~; al.: ' 0-'9- ~ N "~ / / ¢ e/\ ,. // o O~ D o © o / I\ f o ,~\ I / / d C~ c c~ % :: ,,, 7 ° } O Q ~~' / 3 a Fig. 1. Different plant habits in the Coryciinae s. str., and their possible evolution in the group. The habits e - g are found in Ceratandra, c in Pterygodium and b in Corycium. a and d are hypothetical forms Phylogeny of the Coryciinae 167 with an apical bud. In the growing season the bud grows into a shoot, and one of the axially buds forms a new tuberoid, which then will form the next year's shoot. Superficially similar underground organs occur also in the tribe Diurideae (DRESSLER 1981). On the basis of this structure the Diurideae are placed in subfam. Orchidoideae together with the tribes Orchideae and Diseae. However, it has not yet been demonstrated that the underground organs of the Diurideae are homologous to those of the Orchideae and Diseae. In Pterygodium and Coryciurn the tuberoids are always undivided like in all other Diseae, while palmate tuberoids occur in a number of Orchideae (for example in Gymnadenia, Dactylorhiza and Brachycorythis). The shape of the tubers of Pterygodium and Corycium ranges from spheroid to elliptic, although the elongate oblong tuberoids which are found in some Disperis spp. do not occur here. The size of the tuberoids varies extensively (possibly with the nutritional state of the plant and the season). In most species the tubers are sessile on the stem base just below the soil surface. Occasionally they occur at a greater depth and are separated from the above-ground shoot by a short wiry underground stem. On the outside the tubers are mostly slightly pubescent or woolly. A plant has usually only one pair of tubers (a new one and an old one). (2) In Ceratandra the underground organs usually are referred to as a cluster of "thickened roots" (SCHLECHTER 1898, LEWIS 1950, SEN~HAS 1973--1974). These "roots" are usually branched and have a woolly integument. However, it is interesting to note that occasionally (but very rarely) root-stem tuberoids occur as well. In a specimen of Ceratandra atrata (BEAN 757), an elongate tuberoid arises from one of these "thickened roots". In the specimen of Ceratandra grandiflora illustrated by BoLus (1913), both a cluster of"thickened roots" and a stalked spheroid tuberoid are shown. These cases suggest that the "thickened roots" - or some of t h e m - are partly made up of stem tissue, and that the underground organs of Ceratandra are not a completely new development but rather a modification of the tuberoid type. In Ceratandra atrata a bunch of up to five stems can arise from a single cluster of "thickened roots". Stem. The stems of the Coryciinae s. str. are usually solitary (except Ceratandra atrata, see above), erect and unbranched. While a stem length of 1 5 - 30 cm is the most common situation, "dwarfs" of 5 - 10 cm (PterygodiumplatypetaIum, poor specimens of other species) as well as "giants" of 6 0 - 90 cm (Pterygodium magnum, P. hallii) occur occasionally. The stems may be robust (up to 15 mm diameter in P. hallii) or very thin (1 mm diameter in Pterygodium platypetalum), but flexuose wiry stems like in Schizodium (Disinae) (LINDER 1981 b) do not occur. A significant swelling on the stem base is found only in Ceratandra venosa. In many Disperis spp. a similar swelling on the stem base occurs. It remains to be shown whether these swellings are homologous in both taxa. At present, nothing is known of the anatomy and function of the swelling of the stem base in both. The stem is glabrous in all species, although densely hairy stems occur in several Disperis species. [,eaves. All species have entire sheathed leaves spirally inserted on the stem. These are generally convolute in their vernation, conduplicate when fully expanded and not jointed at the base as is the case in most epiphytic orchids. The foliage leaves are of rather thin texture as in most other terrestrial orchids, but are fleshy and subterete in Ceratandra bicolor, C. harveyana and EvoteIla rubiginosa. Short scale-like leaves below the foliage leaves are present in all species. Towards the apex of the stem, the foliage leaves grade into the bracts via several intermediate leaves. The anatomy of the leaves is rather uniform (CHESSELET 1989: 67). The flat and wide leaves of Pterygodium and Corycium are virtually indistinguishable in terms of leaf anatomy. The epidermis is uniseriate, glabrous, and sometimes with epidermal papillae. The anticlinal walls of the epidermis vary from straight to undulating, with the abaxial epidermis often with more undulation than the adaxial surface, Disperis, Ceratandra, Evotella rubiginosa, and Corycium dracornontanum have undulating walls, while the rest of Corycium species have straight walls. Pterygodium is variable for wall undulation. The cuticles are either 168 H. KURZWEIL& al.: striate or smooth, and the mesophyll cells often have well developed arm cells. Raphide idioblasts occur throughout the study group. In most of the Coryciinae the vascular bundles have parenchymatous bundle sheaths, but Ceratandra atrata, C. grandiflora, and C. globosa can be characterized by the possession of fibre caps at the poles of the primary and secondary vascular bundles. The arrangement of the foliage leaves is characteristic of the genera: Cera tandra is characterized by having equally long narrow leaves scattered over the stems. While these are narrowly lanceolate in shape and dense in C. venosa, both C. bicolor and C. harveyana have a foliage made up of few, lax and linear leaves. The remaining Ceratandra spp. have numerous linear and frequently marginally ciliate or coarsely denticulate leaves. An interesting feature of most specimens of Ceratandra (except C. venosa) is the occurrence of a cluster of linear radical leaves. These are frequently dry at flowering time (SCHLECHTER1898). A similar feature is found in some Disa species, notably in D. patens and D. filicornis. In Ceratandra this cluster of radical leaves is the only visible part which is developed in non-flowering plants. The degree of development of this cluster is variable. Usually the radical leaves are very short and inconspicuous, but in some specimens they reach up to the inflorescence (e.g., Ceratandra atrata, BEAN 757). Specimens without radical leaves also occur. Such a variation may be found even within populations (e.g., in C. atrata, LINDER 7476). The radical leaves often arise from a furry and fibrous dark sheath. The function and ecological significance of the radical leaves and the basal sheath has not yet been studied. Basal clusters of radical leaves in Ceratandra and in some Disinae are obviously parallel developments due to convergent evolution. Ev o t ella rubiginosa, like Ceratandra harveyana and C. bicolor, has few lax, linear and semi-terete leaves. Most P t e r y g o d i u m spp. have a lax foliage made up of few lanceolate and cauline leaves: up to five occur in robust specimens of P. leucanthum (e.g., LINDER 954), but the majority of species has only one to three leaves. An unusual foliage occurs in P. alatum with many short but largely basally clustered leaves. The foliage of P. magnum is similarly unusual as the species has lanceolate imbricate leaves equally scattered over the stem. The leaves of P. hallii and P. inversum are narrowly lanceolate, imbricate, and slightly clustered in the lower stem portion. Corycium is characterized by having few to many imbricate and lanceolate foliage leaves. C. flanaganii has equally large broadly lanceolate and imbricate leaves inserted up the entire length of the stem. In contrast, all the other species have narrowly lanceolate leaves in the lower stem portion. A successive reduction can be observed in the genus. The greatest reduction of leaves is found in C. deflexum and C. excisum. It is interesting that these two species inhabit comparatively arid areas (dry karoid vegetation and sandy flats, respectively), while species with a more extensive foliage often occupy mountain marshes and moderately wet grassland (e.g., C. carnosum, C. nigrescens, C. dracomontanum). Another feature of Corycium species occupying dry areas is the withering of the foliage leaves before flowering. This can be seen in sun-exposed specimens of several species, and C. microglossum only flowers with its leaves already withered. However, typical hysteranthous species in which the leaves completely dry off several months before flowering (like in some Disinae) do not occur in the Coryciinae. The reduction of leaf size and number as well as the increasing withering of the leaves at the time of flowering are obviously correlated with an ecological shift from mesophytic growing conditions to dry shrub or sandveld habitats. The primitive habit of the Coryciinae s. str. appears to be a slender plant with a lax foliage made up of few cauline leaves (Fig. 1 a), since this is the general condition in the outgroup Disperis. A lax foliage with few cauline leaves occurs also in the isolated genus Brownleea, which is usually included in the Disinae but which may be rnore ciosely related to the Coryciinae (LINDER & KURZWEIL,unpubl.). This suggests that the most likely trend Phylogeny of the Coryciinae 169 in the vegetative architecture leads from a few- and lax-leaved habit to a dense- and manyleaved habit, and is demonstrated in Fig. 1 by solid arrows. When this character is optimized on the cladogram, the first change is from the gracile habit of the outgroup to a robust habit, followed by independent reversals back to the gracile habit. Intuitively, the robust habit is "primitive". Changing the outgroup coding for this character does not affect the cladogram topology, but shortens the tree by one step. This polarization is shown by the broken arrows in Fig. 1. Inflorescence. The inflorescences of the Coryciinae s. str. are generally simple, unbranched racemes and are borne in a terminal position on the stem. In most species they are either lax, few- and large-flowered (most species of Plerygodium, Ceratandra bicolor, C. harveyana), or dense, many- and small-flowered (Corycium). Pterygodium hallii, P. inversum, P. magnum, Ceratandra atrata, C. grandiJlora, and C. globosa are somewhat intermediate with fairly dense but medium- to large-flowered inflorescences. The flowers are subtended by conspicuous bracts which, although considerably smaller, are usually similar in shape to the foliage leaves. In most species they are as long as the ovary or slightly longer. Exceptions are Pterygodium magnum and Coryciumflanaganii, in which the bracts are much longer than the flowers. The bracts are usually lanceolate, but rarely wide and almost ovate (Ceratandra bicolor, C. harveyana). They are always erect and adpressed to the ovary except in Pterygodium volucris, P. schelpei, P. cooperi, P. ukingense, and P. magnum, where they are strongly deflexed. This bract orientation is unique in the Coryciinae and also does not occur in the Disinae, but is very common in the Satyriinae. The bracts are always persistent and fresh at the time of anthesis and thus never scale-like and dry like in some Disa species. However, in Corycium flanaganii and C. microglossum they frequently have dry apices. Polarising the evolution of this character has exactly the same problems as the vegetative habit, and has similar effects on the cladogram (see also Fig. 1). Flowers. A b b r e v i a t i o n s and t e r m i n o l o g y . The flowers of the Coryciinae are by far the most complicated in orchids. In order to make their architecture comprehensible the following terms and abbreviations are used: AN C CO GYN L LA LC P1-2 RO 81 82-3 STG theca belonging to the fertile anther (belonging to the stamen in position A1) callus on the lip blade of some species (previously termed "tubercle" in BOLUS 1893- 1896, 1913; ROLFE 1913; LEWIS 1950; "laminaler Anhang" in VOOEL 1959; "Schwiele" in SEN~HAS 1973-- 1974) connective of the anther gynostemium (entity composed of the fertile stamen and the epi-gynoeceal organs of the gynostemium; usually in orchids also incorporating the staminodes al and a2) blade of the lip (the median inner petal) basal appendage of the lip which is found in most Coryciinae s. str.; usually very large (previously termed "appendage" or "appendix" by BOLUS1893 -- 1896, 1911, 1918; ROLFE 1913; LEWIS 1950; DRESSLER1981; OLIVER 1986; LINDER 1986, 1988 a; "Anhang" or "basaler Anhang" by VOGEL 1959, SENGHAS 1973-- 1974) lip claw petals rostellum (the modified non-receptive portion of the median carpel apex; in the Coryciinae s. str. the structure bearing the stigmas which are here also derived from the median carpel apex; not clearly three-lobed and consequently not differentiated into lateral and central portions like in related subtribes; the expressions "terminal portion" for the outermost part, and "inner portion" for the part between them are used to circumscribe the condition) median sepal lateral sepals stigmas (here derived from the median carpel; KURZWEIL 1991) 170 H. KURZWEIL ~¢ al.: Phylogeny of the Coryciinae 171 Fig. 3. Flower and gynostemium in Evotella rubiginosa (KuRzWEIL s.n.), b: SEM micrograph. Bars: a: 5mm, b." 0.5mm. For abbreviations see p. 169 G e n e r a l shape. Ceratandra and Pterygodium have comparatively open flowers. These vary from almost flat (Ceratandra grandiflora, Fig. 2 a - c; C. bicolor, Fig. 2 d) to shallowly cup-like (C. harveyana Fig. 2 e - f ; Evotella rubiginosa, Fig. 3 a; all Pterygodium species, here shown in Pterygodium caffrum, Fig. 4 a; P. acutifolium, Fig. 4 b; P. hallii, Fig. 4 c; P. inversum, Fig. 4 d). P. platypetalum with its deeply galeate flowers is an exception in the genus. The flowers do not open at all in the cleistogamous Pterygodium newdigatae vat. cleistogamum. The perianth lobes are also almost closed in P. connivens. In contrast, almost all species of Corycium have subglobose or globose flowers ( C. orobanchoides, Fig. 5 a; C. crispum, Fig. 5 b; C. carnosum, Fig. 5 c; C. nigrescens, Fig. 5 d). These range from horizontally bell-shaped with a wide flower entrance (C. crispum; C. ingeanum; C. deflexum; C. microglossum), to globular with its opening considerably smaller than the maximum flower diameter (C. orobanchoides; C. excisum). An exception is C. flanaganii with spreading perianth lobes. The shape of the perianth is traditionally a critical character in the systematics of the Coryciinae s. str. In particular, it is considered an imporant character to differentiate the genera Pterygodium and Corycium, and was used by SCI-ILECI-ITER(1898), ROI,FE (1913), SEQUOIAS(1973-- 1974), STEWART& al. (1982). Considerable variation occurs in the colour of the perianth, but in most species it is dominated by yellow or green (or rarely brownish or whitish) colours. Exceptions are EvoteIla rubiginosa with red petals and a striking white lip, Corycium carnosum with pink flowers and C. tricuspidatum with a purple perianth. Dry perianth lobes usually turn dark brown or black (C. microglossum, C. nigrescens, C. dracomontanum). R e s u p i n a t i o n . The flowers are resupinate in most Coryciinae s. str. (=the lip faces down). Resupination is like in most other terrestrial orchids the result of torsion of the ovary through 180°. There are only a few species with non-resupinate flowers in three apparently unrelated groups. Fig. 2. Flowers in Ceratandra. a - c C. grandiJlora (S3:~NER s.n.), a Inflorescence, b flower, c gynostemium and lip in side view. d C. bicolor, flower in side view (S~rE~NERs.n.), e - - f C. harveyana, flower in ventral and lateral view (KuRzWHL 1020). Bars: 5ram. For abbreviations see p. 169 172 H. KURZWEIL 8¢ al.: Fig. 4. Flowers in Pterygodium. a P. caffrum (KuRzWEIL 955). b P. acutifolium (KuRzWEIL 1019). c P. hallii (BOLl, 1817). d P. inversum (BOLp 1830). Bars: 5ram. For abbreviations see p. 169 Resupination is very common in orchids in general and seems to be the basic condition in the family. The reverse torsion found in Ceratandra atrata (see below) is apparently a case of "attempted" resupination and supports the view that a general tendency towards the formation of resupinate flower seems to occur in orchids. The lack of resupination in some orchid groups is probably due to secondary loss (see also DRESSLEP, 1981). The flowers are non-resupinate in Ceratandra atrata, C. grandiflora and C. globosa, which is superficially seen to be due to a lack of torsion. On closer examination it becomes evident that the ovaries are in fact slightly twisted in the lower ovary part and twisted back in the apical ovary part (this is most obvious in C. atrata). This feature is referred to as "reverse torsion" and was described by VOGEL (1959) and VAN DER PIJL & DODSON (1966). The flowers are resupinate in the other species of Ceratandra which were previously separated as genus Evota (STEWART & al. 1982). In the past, there was considerable confusion about the flower orientation of Ceratandra atrata, C. grandiflora, and C. globosa. C. globosa was referred to by several authors as having resupinate flowers (BoLus 1888, 1918; SCHLECHTER1898; ROLFE 1913; LEWIS 1950; Phylogeny of the Coryciinae 173 Fig. 5. Flowers in Corycium. a C. orobanchoides, inflorescence (KuRZWEILs.n.), b C. crispum, flower in lateral view (JACKSON s.n,), c C. carnosum, ventral view (KuRZWHL 903). d C. nigrescens, ventral view (KuRzwEIL 1308). Bars: a, c - d: 5 ram, b: 2 mm. For abbreviations see p. 169 SCHELPE 1966). SCHLECHTER,ROLFE and SCHELPE stated the same for C. grandiflora, although BoLus (1888: 193) already described the flowers of the species as non-resupinate. All of the above-mentioned authors, however, agreed in the non-resupinate orientation of the flowers of C. atrata. LINDLEY(1833) and KR~NZLIN(1897), on the other hand, made no mention of the flower orientation of Ceratandra species. Based on his incorrect observation, ROLFE (1913) then separated Ceratandra globosa and C. grandiflora into a new genus Ceratandropsls as opposed to Ceratandra atrata with non-resupinate flowers. He also considered the "resupinate" flowers of the two species to be associated with the total lack of a lip appendage, while this is present in the form of a minute structure in C. atrala. Non-resupinate flowers occur also in Pterygodium hallii and P. inversum, which is due to hyperresupination (= torsion through 360 °) in both species. This torsion occurs in the uppermost, sharply constricted ovary portion. The flowers of Coryciumflanaganii are also 174 H. KURZWEIL& al.: non-resupinate: this is, however, due to a lack of torsion. BOLUS'S (1893-- 1896) drawing shows a plant of this species with hyperresupinate flowers. The drawing is based on a specimen collected by FLANAGAN (No. 1639), in which the ovaries are untwisted. The mistake is therefore due to an incorrect observation made by BoLus. Previously, on account of their non-resupinate flowers Corycium flanaganii and Pterygodium inversum were considered to be closely related, and were placed into Pterygodium sect. Anochilus (ScHLECHTER 1898). This section was raised to generic rank by ROLVE (1913). A third species, Anoehilus hallii, was later added (ANTHONY & SCIJEI.PE 1982). VOCEL (1959) already pointed out that the species are apparently not closely related. Furthermore, the lack of resupination is an inadequate character to delimit the genus Anochilus, as it is not homologous in the three species. Ovary. The ovary is elongate in all species of the group, although its thickness at the beginning of anthesis varies. Whereas Ceratandra spp. have very slender ovaries, these are somewhat stout in some species of Pterygodium. A significant pedicel is very weakly or not at all developed. Usually the six ribs of the ovary are very prominent, and are generally almost equally wide at the ovary base. However, they are equally wide over the whole length of the ovary only in few species (Ceratandra atrata, C. globosa, C. bicolor, C. harveyana, Corycium carnosum). In most species the placental ribs in the central portion of the ovary are three to ten times as wide as the sterile ribs. In some species the reduction of the sterile ribs is even more pronounced and in Corycium flanaganii the lateral sterile ribs are not developed at all. In Corycium carnosum, the placental ribs terminate in obscure bulges at the ovary apex, which appear to be vestiges of a "calyculus"as known from Polystachya, BulbophylIum, Epistephium, and Lecanorchis (VERMEVI~BN 1966, KURZWEIL 1987a). This is missing in most other Orchidoideae, where it is only found in some species of the genus Disperis. Sepals. The sepals of the Coryciinae s. str. are comparatively simple structures compared to other Diseae.'The sepals are always entire and usually flat or shallowly galeate (except in Pterygodium :inversum and P. hallii where they are deeply naviculate). The median sepal is, however, frequently longitudinally strongly curved. The sepals usually have subacute apices, but mucros like in the Disinae (= "apiculi") do not occur here. In most species all three sepals are more or less lanceolate. A tendency towards the formation of very narrow median sepals is found in Corycium. This is developed to its extreme in C. excisum and C. ingeanum where the median sepal is confined to a narrow strap connecting the petals, which is also found in many tropical Disperis species. Strongly oblique lateral sepals like in the genus Habenaria (Orchideae) are generally not found in the Coryciinae. The median sepal exhibits a minute claw in Ceratandra venosa. However, elongate claws like in Disa sect. Micranthae do not occur in the Coryciinae. The sepals are fresh in the majority of species, although they are sometimes very thin in texture. They frequently have dry apices, and in C. nigrescens and C. dracomontanum all three sepals are dry at the time of anthesis. In C. microglossum, the three sepals as well as the petals are dry. In Pterygodium catholicum, P. acutifoIium, P. newdigatae, P. platypetalum, and to a lesser extent also in P. hastatum, the median sepal has a significant saccate spur. This sac is unique in the study group and'probably indicates a close relationship between the five species. A saccate or spurred median sepal is the basic condition in the Disinae, and the lack of this structure in some species is interpreted as secondary loss (LINDER 1981 a--f). Saccate or spurred median sepals are also known in some Disperis spp. (Vocal 1959). These cases suggest that such median sepal spurs evolved independently at least three times in the tribe Diseae. The median sepal is usually firmly adnate to the petals throughout anthesis. This postgenital fusion is very weak in Ceratandra gIobosa and Corycium flanaganii (BoLvs 1893- 1896, 1913; VO6EL 1959). In both species the median sepal seems first to be firmly Phylogeny of the Coryciinae 175 i i e f g ® h Fig. 6. Petal shapes in the Coryciinae s. str. a Ceratandra grandiJlora (KuRzWEIL 1226). b C. harveyana (KuRzWEIL 1020). C C. venosa (MosTERT s.n.), d Evotella rubiginosa (BOL e 85). e Pterygodium magnum (ScIqELPE7196).fP. volucris (BOLe 947). g P. hastatum (BATTEN 1). h Corycium microglossum (EsTERIqUIZEN 35545). i C. orobanchoides (KuRzWEIL 957). Bar: 5 mm adnate to the petals and often becomes free in later stages of anthesis. The postgenital fusion median sepal/petals occurs also in Disperis, in Brownleea (Disinae) and to a lesser extent in some Orchideae (particularly in Cynorkis). The lateral sepals are generally spreading and free, but descending to pendent and congenitally fused in most Corycium species. In C. bicolorum, C. crispum, C. deflexum, C. excisum, C. microglossum, C. bifidum, C. ingeanum, and C. orobanchoides the lateral sepals are fused for more than two thirds of their length. In a group of obviously closely related species (C. tricuspidatum, C. nigrescens, C. dracomontanum, and C. alticola) the lateral sepals are (possibly secondarily) weakly or not at all fused: in C. nigrescens they are congenitally fused in their lower portion, and cohering or postgenitally adnate in the upper half; in C. alticola and C. dracomontanum they are also basally fused but free above; in C. tricuspidatum the lateral sepals are completely free. Free lateral sepals occur also in C. carnosum. Basally fused lateral sepals are found in C. flanaganii, and this is similar to what is found in many Disperis species. The fusion of the lateral sepals is traditionally an important character to delimit Corycium (ScHLECHTER 1898; ROLFE 1913; SEN~I~AS 1973--1974). However, in the past the taxonomic significance was somewhat overemphasized. SCHLEC~ITER(1898), who treated Corycium as a section of Pterygodium, separated all Corycium spp. with free sepals into Pterygodium sect. Eleutero-Corycium. Petals. The petals of the Coryciinae s. str. are rather complicated structures, and are almost always attached to the median sepal to form a hood. Their shape varies considerably 176 H. KURZWEIL& al.: (Fig. 6). In most species the petals are broadly lanceolate or 4- ovate. From the position of the midnerve it is evident that the petals are extremely asymmetrical with the largest portion derived from the anterior petal half while the posterior half is developed as a narrow strap only. Similarly asymmetrical petals also occur in the genus Disperis where, however, symmetrical petals also occur. Generally, the petals of the Coryciinae s. str. are undivided although they are slightly lobed with an obscure basal anterior lobe in many species. Comparatively significant anterior petal lobes occur very rarely and are found only in Ceratandra grandiflora and Evotella rubiginosa. Prominent anterior lobes at the petal base are also found in some tropical Disperis species. Among other Diseae, significant anterior petal lobes are very common in the Disinae where they were used in the classification of the subtribe (LINDER 1981 a--f). They are not found in the Satyriinae, where the petals are very simple structures. The petals are deeply concave to semi-globular in most Coryeium spp., where they have an anterior sac and sometimes also a posterior sac ("pouch") enclosing the processes of the lip appendage. This is developed to its highest degree in Corycium orobanchoides and C. ingeanum, where the dorsal processes of the lip appendage and consequently also the petal pouches are extraordinarily strongly developed. The petals of the Coryciinae s. str. are entire in most species, but deeply serrate in Ceratandra venosa and Pterygodium magnum and undulate in Ceratandra globosa, C. harveyana, C. bicolor, Pterygodium alatum, P. caffrum, and P. hastatum. The petals are unstalked in most Coryciinae s. str. Short petal stalks occur only in Ceratandra. In C. bicolor, C. harveyana, and C. venosa these claws may be comparatively long being up to one third of the petals in length. In the other Diseae stalked petals are only known in Disperis where they are minute. In the Orchideae, petal stalks are found in the form of minute structures in Brachycorythis, but are prominent and very elongate in Huttonaea. Significant extraordinarily thickened petal nerves - narrow median portions of the petals which are visible on the closed bud between the sepals- occur in many Coryciinae s. str. and are also found in the genus Disperis (although frequently confined to the lower petal portion). Similar petal nerves among other Orchidoideae are also found in Brownleea and Brachycorythis. Extraordinarly strongly developed petal nerves occur in other orchid groups, e.g., in Apostasia and Vanilla. However, too few data on their distribution in the Orchidoideae are available to evaluate their taxonomic significance. Lip. The lip of the Coryciinae s. str. is a highly complex organ compared to the rest of the Orchidoideae. This is largely due to (1) the occurrence of an often enormous lip appendage and to (2) the basal fusion of the lip to the front of the gynostemium. Both this features are frequently used as diagnostic characters of the group (RoLvE 1913, SCHLECI-ITER 1926, SENGHAS 1973-- 1974, DRESSLEe, 1981). The question of the homology of the lip appendage was raised by VO~EL (1959: 147). He interpreted the lip appendage in the group as being a newly evolved and complex structure, which therefore cannot be homologized with any other organ. VOGEL pointed out that the lip appendage is almost always initiated in the form of two distinct structures, which is reminiscent of the expected position of the lateral stamens of the outer staminal whorl. This might suggest the homology of the lip appendage with staminodes A2 and A 3. The suspected homology of the laminal lip callus of some Coryciinae s. str. with the median inner staminode (a3) is even more hypothetical. In view of the general absence of vestiges of the adaxial stamens in other Orehidoideae, an occurrence of such structures in the Coryciinae s. str. is highly unlikely. VOGEL'S(1959) statement of the two-partite primordium of the lip appendage corresponds well with similar observations made in Corycium excisum and C. nigrescens by one of us (KuRzWEIL 1991). The occurrence of deeply two-lobed lip appendages in adult flowers of many Disperis species as well as the two-lobed primordial stage of the lip appendage in D. lindleyana (KuRzwHL, unpubl.) suggest that a two-lobed lip appendage is the basic condition in the Coryciinae s.1. 177 Phylogeny of the Coryciinae a e g ¢ @ Fig. 7. Lip shapes in the Coryciinae s. str. a Ceratandra atrata (JACKSONs.n.) b C. grandiflora (KuRzWEIL 1226). C C. harveyana(KuRzwEIL 1020). d C. venosa (MoSTERT s.n.), e Evotella rubiginosa (BOLp 85). f Pterygodium magnum (ScnELPE 7196). g P. cooperi (LINDER 1050). h P. inversum (BOLp 1830). i P. caffrum (KURZWEIL 1217). j P. cruciferum (BOLp 1002). k P. catholicum (BOLp 1008). l P. leucanthum (LINDER 954). m Corycium microglossum (ESTERHUIZEN35545). n C. bicolorum (DucKITT s.n.), o C. nigrescens (HALL 962). Bar: a - m, o." 5 ram, n." 3 mm 178 H. KURZWEIL8¢ al.: / i f IN g ,< \ / G b a Fig. 8. Hypothetical evolution of the lip in the Coryciinae s. str. All gynostemla in side view (diagrammatic). The lip b is found in Ceratandra, d and f in Corycium and g - i in Pterygodium, a, c, and e are hypothetical forms. Black = lip appendage, dotted = lip blade, hatched = lip claw (in d median keel on the lip appendage which the lip claw of Corycium is probably derived from) Phylogeny of the Coryciinae 179 Fig. 9. Lip shapes in Ceratan&a. a - b C. a t r a t a (JACKSONs.n.), c C. harveyana (KuRZWEIL 1020). Bars: 3mm. For abbreviations see p. 169 The fusion of the lip to the gynostemium does not vary among the study species. Since a spur on the lip is missing in the Coryciinae s. str., this fusion does not impair the functioning of the flower. Such a fusion is generally quoted as an important morphological character of the genus Disperis. However, in a number of Disperis spp. the lip is almost or completely free. The lip in the genus Brownleea is also fused to the gynostemium, and this was quoted by SCnLECn~rER (1902), VOGEL (1959) and SEN~HAS (1973- 1974) as a character "linking" the Coryciinae and Brownleea. The lip is also basally fused to the gynostemium in many Orchideae. The kind of fusion, however, is quite different: the base of the lip is only marginally fused to the lateral parts of the gynostemium and leaves an entrance to the lip spur. There is evidence that the tissue involved in this fusion is of staminodial origin (KuRZWEn~ & WEBER, unpubl.). The lip blade is clawed in several species. A flat and wide claw which is structurally different from the lip blade occurs in Ceratandra (Figs. 7 a - d , 9 a, c), Evotella rubiginosa (Fig. 7 e), Pterygodium inversum (Figs. 7 h, 10 f), P. hallii, and Corycium flanaganii, and is probably due to convergent evolution. The lip blade in Corycium spp. has a narrow pendent stalk (here referred to as "upper claw"; see below). The lip blade generally has entire margins except in Pterygodium hastatum where these are crenulate, in P. newdigatae where the apex of the lip is lacerate and in P. magnum where the entire lip margin is fimbriate. Ceratandra spp. have a more or less undulate or dentate lip margin. The diversity of lip shapes in the Coryciinae s. str. is extensive (Fig. 7 a - o). It appears that a simple lip with a well developed lip appendage is the primitive state in the study group, and that a two- or tree-lobed lip blade as well as an insignificant lip appendage are derived. This is suggested by the simple lip blade being the more general condition in the entire tribe Diseae. A prominent lip appendage is apparently the basic condition of the Coryciinae s. str., and its insignificant appearance in Ceratandra sect. Ceratandra is probably due to secondary loss. The bilobed nature of the lip appendage of the Coryciinae s. str. early in the ontogeny suggests that the two-lobed lip appendage as found in Ceratandra bicolor, C. harveyana, and C. venosa is primitive to the unlobed appendage in Pterygodium and Corycium. The hypothesis is also supported by the occurrence of a basically two-lobed lip appendage in the outgroup Disperis, although the distribution of this character state has not yet been assessed. A two-lobed lip appendage is found in the Ceratandra-Evotella 180 H. KURZWEIL& al.: clade: A prominent and deeply two-lobed lip appendage occurs in Ceratandra harveyana, C. bicolor, and C. venosa, while the lip appendage of C. atrata is very insignificant (Fig. 8 b). The lip appendage of Evotella rubiginosa is entire, but is apically deeply bilobulate. In the Pterygodium-Corycium clade an entire lip appendage (Fig. 8 c) is found; this may be interpreted as being fused from two separate lobes. Various reflexed (some species of Pterygodium sect. Pterygodium; Fig. 8 g, i) and funnel-shaped (associated with an apparently secondarily three-lobed lip blade; Pterygodium sect. Ommatodium; Fig. 8 h) appendages are probably derived from a hypothetical ancestral undifferentiated but erect and solid lip appendage (Fig. 8 c). Also the erect, solid and unlobed lip appendage with a ventral median keel as seen in Corycium sect. Mieroglossum (Fig. 8 d) is here interpreted as being derived from the hypothetical ancestral Pterygodium-Corycium lip appendage. From the lip appendage as found in Corycium sect. Microglossum the shield-like lip appendage of Corycium sect. Corycium may have evolved, and the median keel on the lip appendage of a Corycium microglossum-like ancestor apparently has evolved into the "upper claw" found in Corycium sect. Corycium. Ceratandra spp. (Figs. 7 a - d , 9 a - c ) have a descending anchor-shaped to hemispherical lip blade. The lip blade has a prominent laminal callus in all species except in C. globosa and some specimens of C. grandiflora. A prominent sessile two-lobed lip appendage is present in C. venosa, C. bicolor (in both species in the form of two elongate horns) and in C. harveyana (in the form of two wall-like lobes with anterior processes). The occurrence of a prominent lip appendage in the three species is also correlated with the resupinate flower orientation, while the flowers are non-resupinate in the other species. A very small bilobed or three-lobed appendage is also present in C. atrata (Fig. 9 a - b ) . In C. globosa and C. grandiflora such an appendage is minute or not developed at all. Previously, the structure of the lip appendage was used as a taxonomic character. C. bicolor and C. harveyana with their prominent lip appendages were separated into Ceratandra sect. Evota (LINDLEY 1833, SCHLECHTER1898). This section was raised to generic rank by ROLFE (1913). The third species with a prominent lip appendage, C. venosa, was at times placed in Pterygodium, Corycium, Ceratandra, and Evota (Table 1). However, the structure of the lip appendage did not play a role in the confusion concerning its systematic position. C. grandiflora and C. globosa were separated in a separate genus, Ceratandropsis (ROLFE 1913, LEW~S 1950, C. globosa only), on account of lacking a lip appendage as opposed to Ceratandra (only C. atrata) with an obscure appendage. Evotella rubiginosa has a broadly triangular lip blade with a flat and wide claw and a sessile, sub-erect and elongate appendage. The incised apex of this lip appendage has a pendent process. This is usually very short, but in one specimen (OLIVER s.n., BOLp 98) it is as long as the lip appendage itself and furthermore bilobed. In Pterygodium the lip shape varies from unlobed to bilobed and three-lobed (Fig. 7 f-1). The enormous lip appendage is generally elongate, sessile and undivided (though occasionally with lateral processes). Its stoutness varies extensively between the species. The lip appendage is mostly erect or suberect, but reflexed (= related to the main part of the ovary) in some Pterygodium spp., and this is developed strongly in P. platypetalum (Fig. 10 b). In all species the lip appendage is elaborated further bearing apical callosities and cavities (apparently with elaiophores; STEINER,pers. comm.). The very apex is frequently inflexed (Fig. 10d). Unlobed lips occur in a number of Pterygodium species. P. catholicum (Fig. 7 k), P. acutifolium (Fig. 10 d), and P. platypetalum (Fig. 10 b) have lanceolate lip blades associated with a reflexed solid lip appendage. The lip of P. newdigatae is ovate-oblong and apically fimbriate, and has an appendage very similar to that of P. acutifolium. In P. hastatum the lip is ovate-crenulate. Its apically somewhat trifid appendage has a big cavity in the front side. P. eruciferum (Fig. 7j) and P. connivens have linear lip blades and a cross-shaped appendage. In P. hallii and P. inversum (Figs. 7 h, 10 e) the lip is semi-ovate. The lip blade Phylogeny of the Coryciinae t 81 Fig. 10. Lip shapes in Pterygodium. a P. alatum, ventral view (KuRzWEiL 949). b P. platypetalum, lateral view (KuRzwHL 872). C P. caffrum, ventral view (KuRzwEIL 955). d P. acutifolium, lateral view (KuazwEm 1019). e - f P . inversum, ventral view and section (BOLe 1830). g P. hallii, dorsal view (BOLe 1817). Bars: 2 ram. For abbreviations see p. 169 182 H. KURZWEIL& al.: Fig. 11. Lip shapes in Corycium. a C. excisum lateral view (KuRzWEIL 1026). b C. bicolorum, lateral view (DucKrra- s.n.), c C. crispum, lip appendage in dorsal view (JACKSON s.n.), d C. nigrescens, lip appendage in dorsal view (KuRzWEIL 1308). e C. flanaganii (KuRzWEIL 1318). SEM micrographs. Bars: 1 mm. For abbreviations see p. 169 may be slightly emarginate in P. hallii and some specimens of P. inversum. In both species the lip blade has a flat, wide and hairy claw (Fig. 10 f). The lip appendage is elongate and solid and is apically bifid in P. inversum. It has two minute lateral teeth at its base (arrows in Fig. 10 e). P. magnum (Fig. 7 f) with its sessile, ovate and fimbriate lip blade and entire, unstalked and deeply bifid lip appendage is very unusual in the genus. Phylogeny of the Coryciinae 183 In Pterygodium cooperi, P. ukingense, P. schelpei, and P. voIucris the lip is three-lobed. In P. cooperi (Fig. 7 g) and P. ukingense the side lobes are fleshy throughout. In P. volucris and P. schelpei the lateral lobes of the three-lobed lip have tooth-like calli projecting to the sides, which appear as additional lip lobes. In P. volucris they are longer than the obscure side lobes, while in P. schelpei they are minute. The presence of a three-lobed lip in these species is associated with the occurrence of a cup-like lip appendage. A three-lobed lip is extremely common in the Orchidaceae in general, and is obviously the basic condition in the family. However, there is evidence that the three-lobed lips of the four Pterygodium spp. are a secondary condition. Large two-lobed lips spanning the width of the flower are found in Pterygodium datum (Fig. 10 a), P. caffrum (Figs. 7 i, 10 c), and P. pentherianum, A rudimentary lip midlobe is found in P. alatum while this is missing in the other species. The lip of P. leucanthum (Fig. 71) is also two-lobed but is narrow and very short. The lip appendage is short and stout in P. caffrum and P. pentherianum, but is fiddle-shaped in P. alatum. In P. leucanthum it is elongate and apically strongly inflexed. The majority of Corycium spp. have lips longitudinally divided into four parts: (1) The lowermost part of the lip is a short claw. This is generally very insignificant and may also be considered as being already a part of the lip appendage. (2) It is followed by the appendage which is an ascending or horizontal shield with lateral processes, the shapes of which are characteristic of the species. The simplest form is found in C. excisum (Fig. 11 a) where the lateral processes are not clearly developed. However, this may well be due to secondary reduction, since the lateral processes can be seen clearly in the early ontogeny (KuRzwEIL 1991). The lateral processes are horn-like and projecting to the sides in C. bicolorum(Figs. 7 n, 11 b), decurved and wide in C. crispum (Fig. 11 c) and decurved plus reflexed in C. deflexum. In C. orobanchoides and C. ingeanum the lateral processes are very elongate horns projecting backwards. The lip appendage of C. ingeanum has an additional small tongue-like appendage arising between and just above the stigmas (OLIVER 1986). The highest degree of complexity is found in C. tricuspidatum, C. nigrescens (Figs. 7 o, 11 d), C. alticola, and C. dracomontanum, where the decurved and rolled up processes of the lip appendage are fused to the anther and the rostellum. (3) The lip appendage is separated from the flat blade by a narrow pendent stalk ("upper claw"). (4) The lip blades of C. excisum, C, bicolorum, C. orobanchoides, C. deflexum, C. crisum, and C. ingeanum are bilobed and apparently derived from large lip side lobes, while the midlobe is lost. Unlobed lips occur in C. dracomontanum, C. nigrescens, and C. alticola. The square lip of the latter species is notched at the apex, suggesting that the lip is probably derived from lip side lobes in this group of obviously closely related species. An apically trifid lip occurs in C. tricuspidatum only. Lips of a few Corycium spp. are very different from the above. In C. microglossum (Fig. 7 m) and C. bifidum the lip blade is sessile and linear-acute and has an unstalked, elongate and apically dilated appendage with a fleshy median keel on the front face. A similar unstalked and elongate lip appendage with a fleshy keel on the front face is also found in C. flanaganii (Fig. 11 e), where the lip blade, however, is oblong. In this species the blade has a significant flat, wide, and papillose claw. The lip of C. carnosum with its sessile and hood-like lip appendage with an anterior process is unique in Corycium. Gynostemium. The undifferentiated portion of the gynostemium ( = column-part) typical of the Epidendroideae and Vandoideae is present in the Coryciinae s. str. only in Ceratandra sect. Ceratandra where, however, it is very short. A column-part occurs in various unrelated groups in the subfamily Orchidoideae: (1) short in Ceratandra, (2) well developed in Satyr&m, (3) minute in some species of Habenaria (SUMMERHAYES1968). The scattered distribution suggests that this character evolved independently in these genera. The orientation of the anther is traditionally an important feature in the classification of the Diseae, separating the Disinae and Coryciinae with horizontally reflexed anthers 184 H. KURZWEIL& al.: from the Satyriinae with pendent anthers. However, its usage as a strict key character may lead to difficulties as several exceptions occur. In the Disinae with basically horizontally reflexed anthers, both secondarily erect (Disa sect. Micranthae, D. neglecta) and semipendent anthers (D. rosea, D. draconis) occur (LINDER & KURZWEIL 1990). In the Satyriinae with usually pendent anthers, the anther is only slightly reflexed in Satyrium bicallosum. In the Coryciinae s. str. the anther is usually horizontally reflexed or semi-pendent, but pendent and erect anthers also occur (see below). In the genus Disperis the anther is usually horizontal but occasionally ascending. Reflexed anthers occur occasionally also in the Orchideae (Neobolusia virginea, Habenaria sect. Kryptostoma), although an erect anther is usually quoted as the main diagnostic character of the group. The thecae are generally separated by an enlarged connective. A connective separating the two thecae is also found in several other Orchidoideae. This is most prominent in Habenaria (e.g., sectt. Kryptostoma and Multipartitae) and Roeperocharis, but occurs to a lesser degree also in many other genera. A slightly enlarged connective occurs occasionally in Satyrium which led GARAY (1960) to interpret the thecae as two separate anthers. In the Coryciinae s. str. the separation of the thecae acts as a device to position the viscidia next to the lip appendage (the thecae would otherwise be obscured by the latter). Auricles, the lateral gynostemium appendages of the Orchidoideaeinterpreted as filament appendages (KuRzWEIL 1987b), were not observed in the Coryciinae s. str. They may, however, be incorporated in the median callus on the connective of Ceratandra (Vo~EL 1959, KURZWEIL 1991), and this is probably also the case in the other genera. Significant lateral appendages are found in many Disperis spp., and their sculptured appearance suggests an involvement of auricles (KuRZWEIL, unpubl.). Obscure staminodes ("basal bulges", KURZWEIL 1987 b) were observed in some Coryciinae s. str. in the early flower ontogeny and are most prominent in Ceratandra atrata (KuRzWEIL 1991). In the later ontogeny they lose their prominent appearance and become incorporated into the gynostemium, and are no longer distinct. This is perhaps the general condition in the subtribe. The rostellum forms the viscidia in its terminal portions. It is either strap-shaped (Corycium, Pterygodium), or flat and covers the thecae separately (Ceratandra). A flat rostellum covering both thecae is only found in Disperis. Unlike the situation in most other Diseae and in the Orchideae, the rostellum is not deeply three-lobed in the Coryciinae. During the early ontogeny, the median carpel apex is either very shallowly three-lobed, bilobed or completely unlobed (KuRzwEIL 1991). A two-lobed median carpel apex is also found in the early ontogeny of Brownleeaparviflora (KuRzWEIL 1990) and Disperisfanniniae (KuRzwHL 1991). The lack of a deeply three-lobed median carpel apex is probably a derived feature in the Diseae, although it is curious that the lobing is lost right back to the early ontogeny. The receptive area is born on two separate stigma cushions which are situated on the rostellum. In contrast to most other orchids, in the Coryciinae s.1. they are entirely derived from the median carpel apex (KuRzwEIL 1991). This kind of stigma is also found in Huttonaea (KuRzWEIL 1989) and in some species of Brownleea (LIND~ & KURZWEIL, unpubl.). In the majority of species the stigmas are situated on the inner portion of the rostellum, although they are borne occasionally on the terminal portions. The scattered occurrence of the latter feature (Evotella rubiginosa, Pterygodium alatum, P. magnum, Corycium carnosum) suggests that it evolved independently four times in the group. The lateral carpel apices which form the stigmas in other orchids can be detected in the early ontogeny of some species (KuRzWEIL 1991). In the mature flowers they are either completely incorporated into the gynostemium and then not visible as distinct structures, or present in the form of vestigial bulges or teeth. The detailed structure of the gynostemium is best explained individually for the respective genera. Phylogeny of the Coryciinae 185 Fig. 12. Gynostemium in Ceratandra. a - b C. venosa, in ventral and dorsal view (MOSTERT s.n.). Left lobe of the lip appendage removed, c C. grandiflora, lower portion ofgynostemium in dorsal view (KuRZWE!L1226). d C. bicolor, in dorsal view (STEINER s.n.), e C. harveyana, lower portion of gynostemium in dorsal view (KuRzWEIL 1020). SEM micrographs. Bars: 1 ram. For abbreviations see p. 169 The thecae are pendent in Ceratandra (Figs. 9 a - c , 12a-e). In Ceratandra sectt. Ceratandra and Evota they are clearly differentiated into narrow apical anther canals and basal thecae sacs. The anther canals are free from one another and project upwards, while the anther sacs are positioned closely to each other and are connected by a very narrow connective, looking like a minute bulge (Fig. 12 c - e ) . C. venosa is somewhat different, as 186 H. KURZWEIL& al.: Fig. 13. Gynostemium in Pterygodium sect. Pterygodium. a P. acutifolium (BOLp 980). b P. catholicum, central portion (BOLp 1008). c P. leucanthum, right half of gynostemium (LINDER 954). d P. eruciferum (DucKITT s.n.), e P. alatum (KuRzWE~L 949). f P. caffrum, right half of gynostemium (KuRzWE~L 1217). SEM micrographs. Bars: 0.5 ram. All gynostemia are shown in top view. For abbreviations see p. 169 Phylogeny of the Coryci#~ae 187 Fig. 14. Gynostemium in Pterygodium sectt. Pterygodium and Ommatodium. a - b P. pentherianum, top view and right theca (LxNDER4198). C P. volucris, longitudinal section (BOLp 1843). d - e P. cooperi, front view and left theca (L~NDER 956). SEM micrographs. Bars: 0.5ram. For abbreviations see p. 169 188 H. KURZWEIL& al.: Fig. 15. Gynostemium in Pterygodium sectt. Anochilus and Magnum. a - b P. hallii (BOLp 1817). c - d P. magnum, gynostemium and close-up of stigmas (SCHELPE 7196). SEM micrographs. Bars: 0.5ram. All gynostemia are shown in top view. For abbreviations see p. 169 elongate and free anther canals are not present in this species (Fig. 12 a, b). Furthermore, the thecae are separated by a comparatively wide and strap-shaped connective in the species. Whereas the thecae are overall fusiform in Ceratandra atrata, C. grandiflora, and C. globosa, they are flattened and rather wide in C. bicolor (Fig. 12d) and C. harveyana. A distinct feature of all Ceratandra spp. are the flat lateral rostellum arms which cover the dorsal side of the thecae (Fig. 12 c). The cushion-like stigmas are situated in the inner portion of the rostellum in Ceratandra sectt. Ceratandra and Evota (Fig. 12 c - e ) , while in C. venosa they are situated on the lateral parts (Fig. 12b). Significant lateral carpel apices are not visible. The structure of the gynostemium of Evotella rubiginosa (Fig. 3b) is remarkably different, and is very similar to that of some Pterygodium species. This is also correlated with the occurrence of a similar (sessile and elongate) lip appendage, which necessitates this gynostemium structure. The connective is very wide and neither curved nor twisted, and the parallel thecae are thus far apart. The flat stigmas are borne in the terminal rostellum portion. Vestiges of the lateral carpel apices are clearly developed. In the majority of P t e r y g o d i u m spp. (Figs. 13-15) the anthers are horizontally reflexed to semi-pendent. The erect anthers of Pterygodium cooperi, P. ukingense, P. volucris, Phylogeny of the Coryciinae 189 and P. schelpei are probably derived from horizontal anthers. The thecae are generally situated at the far ends of a long strap-shaped and slightly curved connective (this is rather short in P. volucris and P. schelpei). The central portion of this connective is frequently thickened and bears calli (Figs. 13 a - c , 14 a - b ) . In many species the median thickened portion of the connective is furthermore incised (Figs. 13 a, d, f, 14 d). In P. pentherianum the whole of the connective is covered with papillae (Fig. 14 a, b). The thecae are not clearly differentiated into anther sacs and anther canals. Their anterior (originally basal) portion has a wide lobe derived from the outer anther wall, and this is variously twisted in front of the viscidia. This lobe is probably of importance in the pollination of the species or alternatively may act as a protection. The thecae are parallel in most species. The rostellum of Pterygodium is basically a strap of tissue between the lip appendage and the connective. In contrast to Ceratandra is does not cover the thecae on their dorsal sides. It is rather variable in width but generally serves as a carrier of stigmas and viscidia. The terminal portion is a fleshy wide lobe, and is bent backwards and inwards in most species. This lobe is remarkably large in Pterygodium catholicum, P. acutifolium (Fig. 13 a), P. platypetalum, and P. hallii (Fig. 15 a, b). In P. leucanthum, this terminal portion is very narrow (Fig. 13 c). In most Pterygodium spp. the stigmas face upwards and are borne on the inner portion of the rostellum, which also exhibits a non-receptive portion. The stigmas are slightly convex cushions, and are orbicular to lanceolate in shape. Several modifications can be recognized in the genus. In P. alatum (Fig. 13 e) the stigmas are situated on the terminal lobes of the rostellum and face inwards. Another unique stigmatic condition is found in P. volucris (Fig. 14c), P. cooperi (Fig. 1 4 d - e ) , P. scheIpei, and P. ukingense. In these species the comparatively small stigmas are situated on bulges on the rostellum and face forwards. They are obscured by the terminal rostellum portion which is sharply bent back. In P. magnum the stigmas are situated in the terminal rostellum portions, and are strongly convex and almost club-shaped (Fig. 15 c, d). Plerygodium newdigatae var. cleistogamum generally has an abnormal gynostemium (this was described in detail by BoLus 1893 - 1896). Both specimens of P. connivens studied here were found to have abnormal gynostemia with an enormous structure obscuring the rostellum. Vestiges of lateral carpel apices are occasionally visible in Pterygodium. These may take the form of teeth (P. catholicum; arrows in Fig. 13 b), minute round bulges (P. platypetalum, P. alatum), or wide ridge-like structures (P. hallii; arrow in Fig. 15 b). The gynostemium of Corycium spp. is similar to that of Pterygodium, except that it is considerably smaller and more compact, as its parts are very close to each other. The structure of the gynostemium is remarkably uniform in the genus. C. carnosum, C.flanaganii, and C. microglossum exhibit exceptions, and these will also be described below. The gynostemium of C. bifidum was not examined, as material for study could not be obtained (the species was not collected in the last few decades). Adequate information on this species is also not available in the literature. In typical Corycium species, the thecae are rather short. The rostellum is a narrow strap fused to the connective. The stigmas are comparatively large and occupy most of the rostellum except the terminal portions. On account of the twisted nature of the rostellum, the stigmas are twisted in some species (especially in C. deflexum and C. nigrescens). A non-receptive portion on the central rostellum lobe is developed in some species (C. bicolorum, Fig. 16 a - b; C. crispum, Fig. 16 d; C. deflexum, Fig. 16 f). The terminal portions of the rostellum are angled up and backward '(here shown in Fig. 16 b, d - e). The viscidia are exposed on their dorsal side, and the pollinia are attached to them on the front side. Vestiges of the lateral carpel apices occur in none of the species. A tendency towards successive (1) curving of the connective around the "lower claw" of the lip appendage, and (2) twisting of connective and rostellum can be recognized in 190 H. KURZWEIL& al.: Fig. 16. Gynostemium in Corycium sect. Corycium. a - b C. bicolorum, in top and lateral view (DtJcKITT s.n.), c-- e C. crispum (JACKSONs.n.), c gynostemium and lip appendage in ventral view. d - e gynostemium in top and lateral view. f C. deflexum (Sc~ELPE 8088). SEM micrographs. Bars: 0.5 mm. For abbreviations see p. 169 Corycium. This is also correlated with the anther becoming strongly semi-pendent. In C. excisum and C. bicolorum (Fig. 16 a, b) the connective is straight. The rostellum and connective retain their "normal" position relative to each other: the anther and the connective are the outermost organs of the gynostemium, while the rostellum is situated between them Phylogeny of the Coryciinae 191 Fig. 17. Gynostemium in Corycium sect. Microglossum. a C. microglossum (ESTERHUIZEN 35545). b C.flanaganii (KURZWEm 1318). SEM micrographs. Bars: 0.5 ram: Both gynostemia are shown in top view. For abbreviations see p. 169 and the lip. In other species the connective is strongly curved around the lower claw of the lip appendage, and the two thecae are therefore born in front of the gynostemium under the shield of the lip appendage (here shown in C. crispum, Fig. 16 c). Both rostellum and connective are twisted, resulting in an inverted position of the organs: the thecae are situated in the central part of the gynostemium, while the terminal rostellum lobes are positioned on the outer margin of the gynostemium. A further consequence of this inversion is that the enlarged originally outer anther walls now face inward. In C. tricuspidatum, C. nigrescens, C. dracomontanum, and C. alticola the rostellum and the connective are not only variously twisted, but also fused to the lip appendage. The gynostemium of C. flanaganii and C. microglossum is atypical of the genus, and reveals several affinities to Pterygodium (Fig. 17 a, b; enlarged and untwisted connective, incised central connective portion, horizontal thecae). The terminal rostellum portions are not angled back here like in Pterygodium. The gynostemium of C. carnosum is peculiar on account of the strongly cohering thecae and its wide terminal rostellum portions bearing the stigmas and the viscidia. Seed ultrastrueture The sculpturing of the seed coat is usually regarded as being systematically and phylogenetically significant (CLwvORD & SMITH 1969, BARTHLOTT 1976, BARTHEOTT& ZIEGLER 1981, DRESSLER 1981), particularly at the subtribal, tribal and subfamily levels (BARTHLOTT 1976, BARTHLOTT& ZIEGLER 1981). In the present study, 87 SEM micrographs of 20 species of the Coryciinae s. str. were taken and analyzed with regard to their taxonomic implications. The specimens studied, as well as some important features of their seed structure are listed in Table 2. The seeds are fusiform in most species, and are 2.5 - 4 times longer than wide. Exceptions are Ceratandra globosa and Pterygodium alarum with comparatively short seeds (ratio seed length/thickness < 2.5), and Pterygodium catholicum (McLouOHIN 29) and P. newdigatae with very narrow seeds (ratio > 4). The actual seed length ranges from 285 gm (Ceratandra globosa) to 679 gm (Pterygodium catholicum, McLou~HIN 29). The main part of the seed coat is made up of 7 - 13 large elongate cells covering the embryo, which is situated in the central portion of the seed. These "central cells" are longitudinally arranged and occupy approximately 5 0 - 70% of the total seed length. The 192 6> H. KURZWEIL & al.: I s~'u.mozio!ql Ieu!Ia!JM j o "oN ~" ~ " ,, ca I I I "7. I ~ , I m, 7 ' 17'77 t.= ~u.umdln3s 's[Ie~ [eu.qa!aOd ~5 ssou~D!q:~ 'SlIe~ IeU!iO!:mV II I I I I I I I Slp3A~ IeU!pt.mV qlp!a~ ~od SHOOjo "oM ql~ua[ ~od Slpa j o "oN I I I I1'11 II II ez0 ql~uo I ilaa/ql~uai peas o!1~ H r~ ~d (IlaO I~aluao) ql~ua I [IO~ ssam,D!ql/ql~u~[ paos o!leH ..= "m ssau>D!ql P ~ S m < ql~uaI p~as o m '.o Z b~ .g7 oo t~ 8~ ~s ¢/3 Ca b ~eq Phylogeny of the Coryciinae 193 Fig. 18. Ultrastructure of the seed surface in Ceratandra. a C. grandiflora (FouRCADE 975). b C. globosa (EsTEe,HUIZEN 7299). C C. venosa (BOL 5281). SEM micrographs. Bars: 50 lam remaining space is filled up with short but equally wide cells on each end. Usually one row of such "terminal cells" is found on each end except in Ceratandra venosa, Pterygodium cooperi, P. alatum, and Corycium dracomontanum with two to three rows. The terminal parts of the seed are narrower. These terminal portions are beak-like in Pterygodium catholicum and P. newdigatae, while Ceratandra sect. Ceratandra and Corycium carnosum are characterised by having seeds with only one such beak (on the distal side). In Pterygodium catholicum (both specimens studied here) and P. newdigatae this beak is extraordinarily narrow and furthermore strongly twisted. In ColTcium microglossum the entire seed coat is spirally twisted. The point of fusion of the anticlinal walls is significantly deepened, and thus clearly visible in all species. This can particularly clearly be seen in the form of a triangular cavity at the point of fusion of three cells. The anticlinal walls are rather uniform in thickness and measure approximately 3 - 6 gin. These anticlinal walls are straight in the majority of species, but strongly undulate in Ceratandra sect. Ceratandra (Fig. 18 a, b) and in Corycium carnosum. In Pterygodium alarum, Corycium nigrescens (SCHELPE6324) and C. nigrescens (LINDER 4665) some of the anticlinal walls are irregularly slightly thickened ("beaded" according to CLIFFORD& SMITH 1969). The periclinal walls exhibit a fine sculpturing in most species. Various different kinds of such fine sculpture can be found, but this does not reveal a clear distribution pattern. A comparatively coarse sculpturing occurs in Ceratandra, Pterygodium caffrum, and Corycium carnosum. "Periclinal thickenings" are found in almost all species, and vary greatly in abundance, distance to each other, angle to the anticlinals etc. (Figs. 18 a - c , 19 a - h ) . However, it is difficult to quantify such features. The periclinal walls of Pterygodium magnum (Fig. 19 d) have marginal holes in a row. 194 H. KURZWEIL& al.: Fig. 19. Ultrastructure of the seed surface in Pterygodium and Corycium. a Pterygodium catholicum (McLouGHIN 29). b P. leucanthum (BOARDMAN245). c P. caffrum (GIFFEN109). d P. magnum (ScHELPB 6336). e P. inversum (LEWYNS 3272). f Corycium orobanchoides (CASSADY21). g C. carnosum (LINDER 1689). h C. flanaganii (LINDER4852). SEM micrographs. Bars: a - c , f - g : 50gm; d - e : 10~tm Phylogeny of the Coryciinae 195 Seed surface morphology is not very informative of the phylogeny of the Coryciinae s. str., although some peculiarities may be used as autapomorphies for certain species (e.g., twisting of the seed in Corycium microglossum, marginal holes on the periclinal walls of Pterygodium magnum). Only the strongly undulate anticlinal walls found in Ceratandra sect. Ceratandra as well as in Corycium carnosum are phylogenetically interesting, and this could be interpreted as suggestive of a close relationship. This is also favoured by the coarsely wavy periclinal walls, and by the occurrence of a single beak-like terminal portion of the seed in both taxa. Pollen ultrastructure The palynology of the Diseae has been studied in detail by CHESSELET(1989) and CHESSELET & LINDER (unpubl). As is typical for the Orchidoideae, the pollen-grains are in tetrads which are aggregated into massulae. These massulae are fused into sectile pollinia. The pollinia extend to the viscidium which develops in the rostellum. The exine lacks a foot layer, being constructed of bacculae forming a tectum, although some atectate forms are found in the tribe. The massulae are generally elongate and fasciculate, the only exception being Pterygodium hallii, which has fan-shaped massulae. The tetrads are linear in the whole group. This is a synapomorphy for the group, as the tetrads in the Disinae and in Disperis are isodiametrical, rarely with linear tetrads (Disperis circumflexa). These linear tetrads may be responsible for the elongate appearance of the massulae. The structure of the exine in the Co~2vciinaes. str. is constant. The walls are tectate with rather elongate columellae. CnESSELET (1989: 77) interprets the rectum of the Coryciinae s. str. as being secondary. The columellae taper towards the base, and are generally massive, thus leaving but little space between them. The ornamentation of the tectum varies from almost smooth to rugulate (Ceratandra bicolor) or verrucose (Corycium dracomontanum and C. nigrescens). Most commonly the tectum is smooth, occasionally very finely striate with scattered punctae, or with foveolate perforations (Pterygodium acutifolium). In contrast, the sister-group, Disperis, can be atectate or have a reticulate tectum. Cladistic analysis On the basis of the detailed morphological investigations presented above, 56 characters which are potentially phylogenetically informative were selected (see Tables 3 and 4). These occur in more than one species, but not in the whole group. Determining homologies for these characters is in some cases simple, but in others very difficult. There are two major problem areas. The first is in determining whether apparently similar structures are due to convergence or to c o m m o n ancestry, the second is in the delimitation of character states, that is, the subdivision of the variation range into characters. Fortunately, due to the complex nature of the orchid flowers, there are ample highly unique characters which can be used. The polarisation of the characters is always a difficult problem. The two generally accepted criteria are outgroup comparison and ontogeny (MADDISON• al. 1984, LINDER 1988 b). The arguments for Disperis being the sister group of the Coryciinae s. str. are presented elsewhere (LINDER1986). Disperis was therefore used as the outgroup and m a n y of the characters could be polarised using this hypothesis. Some characters, such as the shape of the lip appendage, were polarised from ontogenetic data. The data were then analysed using the software H E N N I G 86 (FARRIS1988), implemented on a Bondwell PC. The data set was too large for the Table 3. Character list for Coryciinae s. str. cladistic analysis. The first state is 0 and the scored 1. Where there are three states, the first state is 0, the second 1 and the third 2 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. Underground organs: thickened roots/undivided tuberoids Cauline leaves: less than six (usually 2-3)/numerous Cauline leaves: all the same size/grading from large at the base to small at the apex Clusters of basal linear leaves: present/absent Fibre caps at the poles of the vascular bundles of the leaves: present/absent Floral bracts: strongly deflexed/erect and adpressed to the ovary Inflorescence: lax and few-flowered/dense and many-flowered Inflorescence: capitate/racemose Flowers: deeply globose/relatively open Flowers: not resupinated (ovary 4- untwisted)/resupinated Flowers: hyperresupinated/resupinated Sepals at anthesis: dried/fresh Median sepal: saccate/flat to concave Lateral sepals: free/somewhat fused/fused for more than two thirds of their lengths Petals: apically more or less bilobed/apically unlobed Petals: with anterior lobes/without such lobes Petals: with prominent pouches for the appendage horns/not so Lip blade: with a wide, flat claw/without a distinct claw Lip: three-lobed/not three-lobed Lip: longitudinally divided/not so Lip: two-lobed sessile blade/not so Lip: two-lobed with central vestigial lobe lost/not so Lip: three-lobed with fleshy side lobes/not so Lip: three-lobed with calli on the side lobes/not so Lip: lateral lobes much enlarged/not so Lip: ± anchor-shaped/not so Lip: deltate/not so Lip blade: with a prominent callus/not so Lip appendage: very insignificant or lost/prominent Lip appendage: two-lobed/not two-lobed Lobes of the lip appendage ifit is two-lobed: elongate horns/concave wall-like structures Lip appendage: entire or shallowly bifid/deeply bifid Lip appendage: elongate and undivided/not so Lip appendage: square/not so Lip appendage: apically inflexed/not so Lip appendage: suberect and elongate and with a fleshy keel/not so Lip appendage: solid/not so Lip appendage: solid with two lateral teeth at the base/not so Lip appendage: cup- or funnel-shaped/not so Lip appendage: cross-shaped/not so Lip appendage: an ascending or horizontal shield with lateral processes/not so Lip appendage: arched over the gynostemium/not so Lip appendage: with an upper claw/not so Lip appendage: with lateral processes/not so Lip appendage: lateral processes very elongate horns/not so Lip appendage: lateral processes strongly deflexed/not so Lip appendage: lateral processes expanded apically/not so Lip appendage: lateral processes fused with the anther and rostellum/not so Undifferentiated column-part: present/absent Anthers: ascending to erect/reflexed Anther: pendent/semi-pendent, horizontal or sometimes erect Rostellum: strap-shaped/flat and covering the thecae Stigmas: borne on the terminal portions of the rostellum/on the inner portion Stigmas: seated on bulges, small, and facing forwards/not so Seed walls: undulate/straight Pollen wall ornamentation: verrucose bacculate/finely striate H. KURZWEIL& al.: Phylogeny of the Coryciinae 197 Table 4. Data table for the Coryciinae s. str. cladistic analysis. The characters are according to Table 3 with the unknown states and inapplicable characters labelled as 9 Disperis C. atrata C. grandiflora C. globosa C. bicolor C. harveyana C. venosa E. rubiginosa P. magnum P. inversum P. hallii P. volucris P. schelpei P. cooperi P. ukingense P. caffrum P. alatum P. pentherianurn P. acutifolium P. catholicum P. newdigatae P. platypetalurn P. cruciferum P. connivens P. hastatum P. leucanthum C. microglossum C. bifidum C. flanaganii C. carnosum C. bicolorum C. excisum C. orobanchoides C. ingeanum C. dejTexum C. crispum C. tricuspidatum C. nigrescens C. dracomontanum C. aIticola 19111 01000 01000 01000 00001 00001 01011 10011 11111 11111 11111 101ll 10111 10111 10111 10111 19111 10111 10111 10111 10111 10111 10111 10111 10111 10111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 19111 11110 11010 11010 10111 10111 10111 10111 01111 11111 11111 01111 01111 00111 00111 10111 10111 10111 10111 10111 10111 10111 10111 10111 10111 10111 11101 11101 11110 11101 11101 11101 11101 11101 11101 11101 11101 11101 11101 11101 11001 11101 11101 11101 11101 11101 11101 11101 11101 01101 01101 11100 11100 11101 11101 11101 11101 11101 11001 11001 11001 11001 11101 11101 11001 11101 10121 10121 10111 11101 11121 11121 11121 11121 11121 11121 11101 10111 10111 11111 11111 11011 01011 11011 11011 11011 11011 01011 11111 11011 11011 11101 11101 11101 11101 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11011 11111 11110 11110 10110 10110 11110 11110 11110 11110 11110 11110 11111 11111 11111 11111 11111 11111 11111 91111 91111 11111 11111 11101 11101 11011 11011 00110 01110 00110 11111 11111 11111 11111 11111 11111 11111 01111 lllll 11111 11111 11111 11111 11111 11111 11111 lllll 11111 11111 11111 11111 11111 11110 01009 01909 01109 01010 01010 01010 09111 19111 11111 11111 11111 11111 11111 11111 11111 11111 11111 10lll 10111 10111 10lll 11111 11111 I1111 11111 11111 11111 11111 t1111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 91111 99111 99111 99111 01111 111ll 01111 91011 91011 90011 90011 90011 90011 90011 90011 90001 90011 90001 90010 90010 90010 90011 90011 90011 90011 90010 90111 90111 90111 90111 90111 90111 90111 90111 90111 90111 90111 90111 90111 90111 11111 11111 11111 11111 11111 11111 11111 11111 11111 10011 10011 11101 11101 11101 11101 10111 10111 10111 10111 10111 10111 10111 10110 10110 10101 10111 01111 01111 01111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 11111 10111 10111 11111 11111 01001 01001 01000 01000 01001 01001 01001 01001 01001 01001 19111 19101 19101 19101 19111 19111 19111 19111 19111 19111 19111 19110 19110 19110 19110 19111 19111 19111 19111 19111 19111 19111 19111 19111 19111 19111 19111 19111 19111 19111 t1111 11111 11111 01111 01111 01111 00011 00011 00011 00011 11111 01110 01110 01110 01111 01111 00lll 10011 10011 10111 10111 10101 10101 10101 10101 10111 10011 i0111 10111 10111 10111 10111 10111 10111 10111 10111 10111 10111 10111 10010 10111 10111 10111 10111 10Ill 10111 10111 10111 10111 10111 implementation of an exact algorithm, like "'ie*". Consequently, approximation approaches were used: the routines "mhennig*" and "bb*". Although these may not find all the shortest trees, they are likely to find most of them (PLATNICK 1989). The analysis located 56 equal length trees, with a consistency index of 67, and a length of 85 steps for 56 characters. There are two suggested ways of locating a single tree from multiple trees. CARPENTER (1988) suggests a successive weighting 9 1 1 9 1 9 1 9 1 9 1 9 9 9 9 9 9 1 1 9 9 1 9 9 9 9 9 1 9 9 9 1 1 9 1 9 9 0 0 9 198 H. KURZWEIL& al.: E ~ ~ ~ x . . . . ~ , ~?° ,.-, C~ 0 0 : ~ ~ E. 0 ~', o 0 ...... 0 0 0 0 ~ a H , 0 0 0 0 )c ~.6 • 31 l= 5 E ~ x ~ ~ E ~ ~ , 0 : 16:2 E 0 0 0 o 0 (J 0 0 l: .. ~ . . . . o. o. o. , : : . . o. o. cL o. . ® > = _~ ~ 3 .~ 0 : n o. cL o. ~. o. o. o. n cL o. ~ :: |~ 6=:18 20 D r .42 ~, "-3 _'z4 ~ . . y t= 0 ~ : :3 m27 ~1 ~ • m13 -~ Fig. 20. Cladogram for the species of the Coryciinae s. str. A solid line indicates a unique synapomorphy, a double line a parallelism and a cross a reversal. The characters are numbered according to Table 3 approach, in which those characters that support the chosen tree receive a higher weighting. The alternative approach is to calculate a strict consensus tree. This is a more conservative approach, as it is based on the information shared by all the competing trees. This latter option was followed in the present study. The strict consensus tree was calculated using the routine "nelsen" in H E N N I G 86, and is presented in Fig. 20. The length of the consensus tree is 87 steps. Although this tree was not systematically tested for robustness, various analyses were performed reversing the polarity on some of the less clearcut characters. These resulted in only marginal changes in the terminal portions of the cladogram. The basal nodes appeared to be stable. Analyses using the successive weighting approach by im- Phylogeny of the Coryciinae 199 plementing the "xsteps w" routine in H E N N I G 86 did not result in much more resolved trees. This may be because the lack of resolution is largely due to absence of data, rather than conflict in the data set. Phylogeny and evolution Relationships of taxa* The basal dichotomy in the cladogram indicates the existence of two clades which corresponds to the classification of SCI~LEC~TER (1898) who recognized two groups (the Evota-Ceratandra group, and the Pterygodium-Corycium group). Ceratandra. Ceratandra forms a distinct and obviously monophyletic genus, characterised by cylindrical thickened roots (Char. 1) and pendent thecae (Char. 5!). Evotella rubiginosa which is here considered as being the sister group of Ceratandra (see cladogram), shares the cauline leaves being the same size up the length of the stem (Char. 3) and the spathulate and + anchor-shaped lip (Char. 26) with Ceratandra. Ceratandra sect. Ceratandra is defined by lack of resupination (Char. 10), the possession of an undifferentiated column-part (Char. 49), the insignificant or lost lip appendage (Char. 29), the undulate seed coat walls (Char. 55) and the fibre caps in the leaf anatomy (Char. 5). Within the group, C. grandiflora and C. globosa are related by the capitate inflorescences (Char. 8) and the lost lip appendage. In C. atrata the lip appendage is minute. The degree of median sepal/petal fusion is very weak in C. globosa. This does not necessarily have to be a primitive condition (as such interpreted by VO~EL 1959), since petals adnate to the median sepal are probably primitive in the Coryciinae. The petal claws are unique in the Coryciinae s. str., and may be another synapomorphy. The occurrence of prominent anterior lobes (Char. 16) which are rare in the other Coryciinae s. str. represents an interesting similarity to EvotelIa rubiginosa. The two species of Ceratandra sect. Evota (C. harveyana and C. bicolor) are linked by the reduced number of leaves (Char. 2) and the few-flowered inflorescences, and the two species share the cluster of radical, linear leaves (Char. 4) with Ceratandra sect. Ceratandra. In addition, the gynostemium is rather uniform in Ceratandra sectt. Ceratandra and Evota, having prominent anther canals distinctly projecting above the lip insertion and a narrow sculptured connective between the two anther sacs. The main difference is the form of the anther canals as these are round in sect. Ceratandra and laterally flattened in sect. Evota. The latter is probably the derived condition, since round anther canals are the more general state in the Orchidoideae. The gynostemium of Ceratandra venosais quite different from the above (thecae not clearly projecting beyond the point of lip insertion, wide strap-like connective), suggesting that the species is rather isolated. C. venosa, which was originally described as Pterygodium venosum and was recently transferred to Evota, is here considered as being a sister group to the Ceratandra sect. Ceratandra/sect. Evota clade. Evotella rubiginosa. The morphological peculiarity of the species is reflected in its complicated taxonomic history. BoLus (1893-1896) and SCHI~ECI~TER(1898) * The numbers in brackets, (Char . . . . ), refer to the numbers of the character in the cladogram in Fig. 20. 200 H. KURZWEIL~:; al.: Table 5. Differentiating characters of Ceratandra sect. Evota, C. sect. Venosa and Evotella rubiginosa Distribution Habitat Underground organ Leaves Inflorescence Flower size Petal stalk Pet. ant. lobe Lip blade Lip appendage Lip callus Thecae Connective Rostellum Stigmas Ceratandra sect. Evota Ceratandra sect. Venosa Evo tella rubiginosa W. Cape fynbos roots linear, cauline lax large minute obscure deltoid/lunate deeply 2-lobed present pendent minute bulge flat inner portion of rostellum W. Cape fynbos roots narrow, cauline dense medium-sized significant obscure orbicular deeply 2-lobed present pendent wide strap flat inner portion of rostellum W. Cape, S. Cape fynbos tuber linear, cauline dense medium-sized missing significant triangular apex bifid missing semipendent wide strap strap-like terminal port. of rostellum placed it in Pterygodium, but ROLFE (1913) and STEWART 8L al. (1982) included it in Corycium. The tuberoid shape of the underground organs is similar to that found in the Pterygodium-Corycium group, while the foliage is remarkably similar to that found in Ceratandra sect. Evota. Comparatively significant basal anterior petal lobes (Char. 16) occur in Evotella rubiginosa as well as in Ceratandra. This and other affinities between Ceratandra and E. rubiginosa are shown in Table 5. Striking similarities to Ceratandra can also be found in the lip structure (flat basal claw, Char. 18) and the triangular lip blade of Evotellarubiginosa may have been primitive to (or derived from) an anchor-shaped lip-blade as found in Ceratandra. However, it may also be derived from a three-lobed lip as found in Pterygodium volucris. The architecture of the gynostemium of the species reveals affinities to Pterygodium (elongate strap-shaped and untwisted connective). The situation of the stigmas is very peculiar as they are found on the terminal portions of the rostellum. This character, however, is also found in a few other Coryciinae s. str. The gynostemium of the species exhibits significant vestiges of lateral carpel apices, which are obviously primitive features. E. rubiginosa is probably an isolated species close to the ancestral condition of the Coryciinae s. str. The cladogram shows three characters linking Evotella rubiginosa and Ceratandra: the cauline leaves all of the same size (Char. 3); the flat lip blade with a wide claw (Char. 18); and the 4- anchor-shaped lip blade (Chr. 26). If E. rubiginosa is inserted at the base of the Pterygodium clade, the tree is three steps longer, and if it is placed as a polychotomy at the base of the cladogram, it is also three steps longer. As the most parsimonious position is basal to the Ceratandra clade rather than the Pterygodium clade, this is where E. rubiginosa is placed in this study. Pterygodium and Corycium.SCHLECHTER (1898) defined the Pterygodium-Cor- Phylogeny of the Coryciinae ~ 201 o UNDER 2000 F% 2000-4000 ,, 4000-6000 ,, OVER 6000 100 I i I 200 ~ .,,..,/-.. ~ . ~ " ~ I MILES ~ " • " . ' , ' "~" , ' . ", " . '~ ~[["'-".',,~'~ i i!ii . . . . . . . . . . . . . . . . ., ..,,. •..,..... , Fig. 21. Centres in southern Africa, as delimited by the numerical analysis. Area I includes Namaqualand, area 2 Cape Floristic Region and area 3 the Drakensberg. Area 4 is in southern Tanzania, and includes only a single species ycium group by the presence of tubers and by the lack of a lip callus. We were able to find two synapomorphies for the group: the narrow strap-like rostellum (Char. 52; there may be some doubt about the polarisation of this character), and the single fused lip appendage (Char. 32). Pterygodium can be based on the possession of unfused lateral sepals and an elongate, simple lip appendage (Char. 33). This may well be a weak character, but provides the basic structure from which the extant lip appendage structures may have diversified. Corycium is defined by globose or semi-globose flowers (Char. 9). The lateral sepals are fused or occasionally unfused in the genus. In both genera, the flowers may be resupinate or non-resupinate. However, SCHLECRTER (1898) placed all species with non-resupinate flowers in a separate genus, Anochilus. He also considered the degree of lateral sepal fusion as an important character to subdivide the Corycium-Pterygodium group, and he transferred the two Corycium species with free lateral sepals (C. carnosum, C. tricuspidatum) to Pterygodium sect. Eleutero-Corycium. Pterygodium magnum (P. sect. Magnum). This species was placed previously in Corycium (RoLFE 1913, STEWART& al. 1982). The robust habit of the species (many -4- dense, imbricate and lanceolate leaves, dense inflorescence), which is more 202 H. KURZWEIL& al.: common in Corycium than in Pterygodium spp., is difficult to interpret. It may be the primitive habit of the Coryciinae s. str. and thus does not necessarily indicate a close relationship. The general flower appearance (large open cup) is a definitive affinity to Pterygodium, but this is also plesiomorphic. The unlobed sessile lip with its entire and sessile appendage supports the inclusion of the species in Pterygodium. This is also supported by the structure of the gynostemium with its wide and untwisted connective. The most parsimonious arrangement places Pterygodium magnum as the sister group to the Pterygodium volucris clade. The two are linked by the joint possession of reflexed floral bracts (Char. 6). Placing Pterygodium magnum in a basal position in Pterygodium is one step longer, and placing it at the base of the Corycium clade is two steps longer. Nevertheless, the species reveals a number of curious autapomorphic features: fringed petals and lip, and odd-shaped and apically bifid lip appendage, an ascending anther, strongly convex stigmas in the terminal rostellum portions (Char. 53), and holes in the Periclinal walls of the seed-coat. The laceration of the petals and the lip is superficially similar to what is found in the genus Huttonaea. This all supports the concept of this species as being a rather isolated taxon within the Pterygodium clade. Pterygodium volucrisclade (P. sect. Ommatodium). The group, consisting of P. volucris, P. schelpei, P. cooperi, and P. ukingense, has a large set of synapomorphies: an erect or suberect anther (Char. 50), a three-lobed sessile lip (Char. 19) with fleshy or calliferous side lobes, a funnel-shaped lip appendage (Char. 39), a narrowly wall-like connective and the stigmas on bulges in the inner portion of the rostellum and facing forward (Char. 54). The most obvious feature among them is the erect or suberect anther. LINDLEY (1833) separated Pterygodium volucris from Pterygodium into a new genus Ommatodium, an approach also followed by ROLFE (1913). However, later authors emphasized the affinities to Pterygodium (few lax leaves, open cup-like flowers; elongate and sessile lip appendage, sessile lip blade) and consequently placed the species in Pterygodium. Two subgroups can dearly be recognized: a subtropical or tropical group consisting of P. cooperi and P. ukingense, with lax inflorescences made up of few large flowers (Char. 7) and lips with fleshy sidelobes (Char. 23); and a western Cape group consisting of the two very similar species P. volucris and P. schelpei, with semi-dense inflorescences made up of medium-sized flowers, petals which are apically two-lobed (Char. 15) and a flat + three-lobed lip with fleshy horn-like calli on the side lobes (Char. 24; LtNDER 1988 a). Pterygoeh'um inversum clade (P. sect. Anoc~Tus). The lip with an elongate, solid and sessile appendage (Char. 37), the strap-shaped and untwisted connective and rostellum and stigmas in the inner rostellum portion and facing upwards link this group (Pterygodium inversum, P. hallii) with the rest of Pterygodium. The two species are very similar in appearance and the group is defined by the short teeth at the base of the lip appendage (Char. 38), the hyperresupinate flowers (Char. 11) and the lip blades with distinct flat and wide claws (Char. 18). ROLFE (1913) grouped two species originally described as Pterygodium inversum and P. flanaganii in a new genus, Anochilus, due to their having non-resupinate flowers. A third species, Anochilus hallii, was later added. Both A. hallii and A. Phylogeny of the Coryciinae 203 inversus share a number of advanced characters with other Pterygodium species, while the third species, A. flanaganii, is quite distinct and reveals affinities to some Corycium species. Pterygodium catholicum clade (P. sect. Pterygodium). This group comprises the majority of Pterygodium spp., and may thus be referred to as the "typical" Pterygodiums. The species of the section are recognized by the erect bracts, the few and large open cup-like flowers borne in lax racemes, the sessile and 2- or unlobed lip, the sessile elongate lip appendage, the horizontal to semi-pendent thecae, the untwisted strap-like rostellum and connective and the stigmas almost always in the inner rostellum portion and facing upwards. The group is a sister group of the Pterygodium inversum clade with which it shares the elongate, solid, and sessile lip appendage (Char. 37). There are no unique synapomorphies for the group, and the only evidence for its monophyly is the reduced habit, with plants with only a few leaves (Char. 2) and inflorescences which are generally laxly few-flowered (Char. 7). The group is rather uniform, and the main diagnostic features of the species are found in details of the lip shape. Consequently, the cladistic history of the group is as yet poorly understood. Several subgroups can be recognized: The first group, with P. caffrum, P. alaturn, and P. pentherianum, have bilobed lip blades spanning the width of the flower (Chars. 21 and 25; the lip blade is thus 3 - 4 times wider than the lip appendage). P. alatum has a rudimentary lip midlobe, and this is also associated with the presence of many cauline leaves (Char. 2). The second group, which includes P. catholicum, P. acutifolium, P. newdigatae, and P. platypetalum, has a saccate median sepal (Char. 13). A phyletic trend in the group is the successive reflexion of the elongate and apically inflexed lip appendage, which is developed to its highest degree in P. platypetalurn. Another trend is in the further reduction of the plants from three to two to a single leaf and a co-ordinate reduction in the inflorescence size. P. catholicum is the most robust member of the group and P. platypetalum the most reduced. A cleistogamic Pterygodium with an abnormal gynostemium is recorded from the southern Cape and this is usually considered as being a form of P. newdigatae (P. newdigatae var. cleistogarnurn). The third group consists of only two species, P. cruciferum and P. connivens. Here the lip is reduced to a linear lobe. An apomorphy of these two species is the cross-shaped lip appendage (Char. 40). An abnormal gynostemium is found in P. connivens. The connivent perianth of this species may be interpreted as being the initial stage in the formation of a cleistogamic perianth. This suggests that this species is in fact only a very localised form of P. cruciferum on the way to cleistogamy. The two remaining species are rather peculiar. They appear to be well placed in this group, but are clearly isolated. P. hastaturn has a small ovate, fringed lip blade and a shallowly trifid lip appendage with a big cavity on the front face (Char. 39). The shape of the lip appendage almost certainly does not indicate an affinity to the P. volucris group, as not many other similarities exist. The weakly developed sac on the median sepal (Char. 13) suggests a relationship to P. catholicum. P. leucanthum has a very much reduced lip blade consisting of two short round lobes (Char. 21), which appears to be an affinity to P. caffrurn. However, the lip blade of the species is much narrower than in P. caffrum (as wide as the lip appendage 204 H. KURZWEIL& al.: only). The reflexed and apically strongly inflexed lip appendage (Char. 35) represents a similarity to P. catholicum. The narrow lateral rostellum lobe is evidently an apomorphy of the species. Corycium microglossumclade (C. sect. Microglossum). This group, containing Corycium microglossum, C. bifidum, and C. flanaganii, is very heterogenous. C. microglossum and C. bifidium were in the past placed together with the "typical" Corycium species due to their semi-globose perianth and the strong fusion of the lateral sepals (ScI-ILECnTER 1898, ROLVE 1913). The third species, C. flanaganii, is very odd in many characters, but is here considered to be distantly related to C. microglossum. The morphological similarity between these two species was already noted by BoLus (1893 - 1896) and KRXNZLIN (1897). The sessile and elongate lip appendage is superficially similar to what is found in Pterygodium. However, it is here interpreted as the ancestral condition of the genus Corycium, and may be understood to have evolved into the shield-like lip appendage found in the C. bicolorum group. A relationship to Corycium is also supported by the fusion of the lateral sepals and the semi-globose flowers. The latter feature is, however, not seen in C. flanaganii. The gynostemia of the three species are strikingly similar to Pterygodium (untwisted strap-like very wide connective and rostellum). As this is perhaps the primitive condition in Corycium and Pterygodium, this does not necessarily indicate a relationship to Pterygodium. The elongate undivided lip appendage with its fleshy median keel (Char. 36) and the sepals being dry at anthesis (Char. 12) are the synapomorphies for the group. Within the group, Coryciumflanaganii is rather isolated. The habit is robust with equal-sized, imbricate and cauline leaves and large leaf-like bracts. The flowers are not resupinated (Char. 10) and the median sepal and petals are not adnate. The latter was probably lost, as there is still some fusion at the beginning of anthesis (BoLus 1893- 1896). The lip claw is flat and wide (Char. 18). Corycium microglossum and C. bifidium are more closely related, sharing elongate lip appendages with dilated and inflexed apical parts (Char. 42). Coryciumcarnosum (C. sect. Carnosum). C. carnosum is placed in the genus Coycium mainly on account of its foliage, the globose flowers (Char. 9) and the two-lobed lip. The exact placement of the species was much disputed in the past, as it is intermediate between Pterygodium and Corycium for several characters. The species was placed by SCnLECHTER (1898) in Pterygodium sect. Eleutero-Corycium, which only contains such "problem species". The feature which apparently caused the greatest deal of confusion in the past is the complete absence of any lateral sepal fusion. The occurrence of the stigmas in the terminal rostellum portion (Char. 53) is not taxonomically valuable, since this character evolved independently several times in the Coryciinae s. str. The occurrence of an obscure calyculus, which is apparently a primitive feature in orchids in general (see also DRESSLER 1983), is another odd feature of the species, emphasizing its phylogenetically isolated position. Very surprisingly, several affinities to Ceratandra are found here. The anticlinal walls of the seed coat are strongly undulate (Char. 55) in all three species of Ceratandra sect. Ceratandra and in Corycium carnosum, and the lip of C. carnosum is similar to the anchor-shaped lip found in Ceratandra. The lip of C. carnosum has a very obscure laminal callus which is reminiscent of the situation found in Phylogeny of the Coryciinae 205 Ceratandra (see also VOOEL 1959: 164). The lip callus of C. carnosum is, however, very insignificant. If C. carnosum is placed ancestral to the Pterygodium clade, the cladogram is only two steps longer, similarly placing it ancestral to the Ceratandra clade increases the length by two steps. This indicates that the position of C. carnosum is not very certain. Corycium bicolorum elade (C. sect. Corycium). The species included here can be recognized by their robust habit, their globose- or semi-globose flowers (Char. 9), their almost always fused lateral sepals (Char. 14), and the lip which is longitudinally differentiated into a slightly stalked shield-like lip appendage with lateral processes, a narrow pendent claw and the usually two-lobed lip blade (Chars. 20, 41, 43, and 44). In the past, there was little doubt about the unity of the group, and the species - with the exception of C. tricuspidatum - were generally placed together, either in the genus Corycium (BoLus 1888, KRaNZLIN 1897, ROLFE 1913), or in Pterygodium sect. Corycium (ScHLECHTER 1898). However, several apparently not closely related species were added in the above classifications. The group is remarkably uniform, and differs mainly in the degree of lateral sepal fusion and in the shape of the lip. Consequently, it is difficult to fully resolve the cladogram to the species. The clade containing Corycium tricuspidatum, C. nigrescens, C. dracomontanum, and C. alticola is defined by several synapomorphies. The lateral sepals are fused only up to V2 of their length (Char. 14), and in C. tricuspidatum this fusion is lost. It is interesting to note that in C. nigrescens the free apices of the lateral sepals are postgenitally adnate, while they remain free in C. dracomontanum and C. alticola. The lip appendage with its generally decurved and inrolled lateral processes is fused to the gynostemium (Chars. 46 and 48), and the lateral horns of the lip appendage are expanded apically (Char. 47). Within the clade, C. nigrescens and C. dracomontanum are linked by the verrucose-bacculate pollen walls (Chars. 56). Corycium bicolorum, C. excisum, C. crispum, C. deflexum, C. ingeanum, and C. orobanchoides can be recognized by their strongly fused lateral sepals (more than 2/3 of their length), and by their throughout bilobed lip blade. The lip appendage is free from the gynostemium, and has variously shaped lateral processes. A phylogenetic trend can be observed in the successive twisting of connective and rostellum. The data are not adequate to determine the relationships between the species, but C. ingeanum and C. orobanchoides can be grouped on the basis of the slender elongate lip appendage horns which extend into pouches on the petals (Chars. 17 and 45). Phytogeography The basic data for determining the phytogeographical patterns in the Coryciinae s. str. were taken from the manuscript account of the Orchidaceae for the Flora of Southern Africa. The distribution data for this account were compiled from the specimens held at BOL, NBG, STE, PRE, and NU (Herbarium abbreviations in accordance with HOLM~REN & al. 1981). The distributions were mapped according to the quarter-degree system of EDWARDS & LEISTNER (1971), and the detailed maps will be published in the Flora. 206 H. KURZWEIL& al.: 12 ° 14° 16 ° 18 ° 20 ° 22 ° 24 ° 26 ° 28 ° 30 ° 32 ° 34 ° 18 ° 20° 22 ° 24 ° 26° 28 ° ]0 ° 32 ° ~4° 10 ° 1 1 12 ° 14° 1 I 16 ° 1 )rr/~'PU/~li~ 18 ° 20 ° 22 ° " 24 ° / 26 ° 28 ° 30 ° 32 ° 34 ° 36 ° Fig. 22. Species richness per half-degree square in southern Africa. The figures indicate the total number of species of the Coryciinae s.str, found in each half-degree square The Coryciinae s. str. are widespread in the more temperate parts of southern Africa, with a single species occurring in the uplands of southern Tanzania. The greatest species diversity is found in the south-western corner of the Cape Province, with a second centre of diversity in the Drakensberg of Natal. The number of species per degree square for southern Africa is indicated in Fig. 22. This pattern is typical for the temperate flora of sub-Saharan Africa (LINDER 1990, WroTE 1983). In order to determine whether any centres of endemism can be delimited, two approaches were followed. The distribution data at a degree-square scale were used to determine the relationships among the degree squares determined by analysing the data set using NTSYS-pc (RoHLV 1988). The dissimilarity coefficients were calculated using the Jaccard coefficient as this does not take mutual absences into account. From the dissimilarity table a phenogram was prepared using UPGMA. From the phenogram it is evident that three large groups can be recognized: the Karoo margins, the south-western Cape Province, and the Drakensberg. As it was found that degree squares with single species present led to distortions in the result, these were eliminated from the data set for the analysis. In the second approach the species maps were compared, and those species showing similar patterns were 207 Phylogeny of the Coryciinae Table 6. The centres of endemism in the Coryciinae s. str. Drakensberg Pterygodium magnum P. cooperi P. hastatum P. leucanthum CotTcium flanaganii C. tricuspidatum C. nigrescens C. dracomontanum C. alticola Central African Centre Pterygodium ukingense Karoo-margins species P. schelpei P. hallii Corycium deflexum C. ingeanum Cape Floristic Centre Ubiquitous Southern group Northern group SW. Cape W. Cape SW.- and S. Cape Ceratandra atrata C. globosa Pterygodium volucris P. alarum P. catholicum P. acutifolium P. caffrum Ceratandra grandiJTora Pterygodium newdigatae Ceratandra harveyana C. venosa Evotella rubiginosa Pterygodium cruciferum P. connivens Corycium microglossum C. bifidum Ceratandra bicolor Pterygodium inversum P. platypetalum P. pentherianum CotTcium excisum C. orobanchoides C. crispum C. bicolorum C. carnosum grouped together. These, as expected, are largely similar to the patterns found in the numerical analysis. However, overlaps due to interdigitating biotic patterns are more easily analysed by the visual approach, as it does not have the coarseness of the stricter areal approach. The following groups can be recognized (see also Table 6): 208 H. KURZWEIL8¢ al.: The Drakensberg endemic centre includes nine species. All but one of these are found in the central Natal Drakensberg. Coryciumflanaganii is found only in the southern Drakensberg, but as this species is collected relatively rarely it may well be found towards the central Natal Drakensberg. Some species are restricted to the central Natal Drakensberg, while others range north to the Houtboschberg near Pietersburg and south to Knysna. To the south the species overlap with species from the Cape centre. This pattern has been amply demonstrated for the Disinae, particularly in sect. Micranthae (LINDER 1981 d). There are no subcentres within this Drakensberg centre. The centre has been well documented for other groups (PHILLIPS1917, HILLIARD & BURTT 1987, LINDER 1990), and has been called the "Eastern Mountain Region" by PHILLIPS (1917). The Central African centre is represented by a single species, Pterygodium ukingense, which is endemic to southern Tanzania. Although WHITE (1983) and LINDER (1990) include the Drakensberg in the Afromontane phytochorion, there is no species level overlap in the Coryciinae s. str. between the two regions. LINDER (1983) showed this area, in combination with the Nyika Plateau in Northern Malawi, is an important centre for the related Disinae. The Karoo-margins species are found along the escarpment from Springbok in the north to Sutherland in the south, and penetrate into the arid central valleys of the Cape Floristic Region. There are four species in this group. The overlap with the species showing a Cape Floristic Centre distribution is extensive, especially in the Calvinia area, where the Cape species are found on the edge of the escarpment at Nieuwoudtville, on sandstone derived soils, while the Karoo species occur away from the escarpment, generally on heavier soils. The centre lacks the pattern evident in the Drakensberg centre, with a central "core" area and with decreasing species richness away from this core. This lack of pattern may indicate that the Karoo mountains do not represent a true centre, but rather a border area of the Cape centre. However, the small number of species may stochastically obscure the effect of a core with decreasing margins. The Cape Floristic Centre (CFR) (sensu GOLDBLATT 1978) has long been recognized as being very distinct floristically, with a high level of endemism (Good 1974, TAKHTAJAN 1986, WEIMARCK 1941). WEIMARCK (1941) and OLIVER • al. (1983) sought to define centres and intervals within the Cape Floristic Region. The numerical analysis was probably too coarse to recognize such centres from the data on the Coryciinae, consequently the manual analysis needs to be referred to. The species can be grouped as follows: there are seven species ubiquitous to the centre and the remaining species can be divided into the following groups: a) A southern group, with two endemic species; b) A northern group, with sixteen species. Of these, all but one (Pterygodiurn pentherianum) occur on the Cape Peninsula. Some seven species are found only in the very south-western corner of the area (Cape Peninsula and SW. Cape northwards up to Worcester), while seven extend northwards to the margins of Namaqualand (W. Cape west of approximately 20°; Cape Agulhas to Clanwilliam), and two extend eastwards along the south coast. The overlaps between these groups in the CFR are so large that it does not appear feasible to recognize distinct groups, as WEIMARCK (1941) and OLIVER & al. (1983) suggest. The patterns of species distribution correspond remarkably with those shown for the Drakensberg centre. Of the 25 species found in this centre, only three (Ceratandra grandiflora, Pterygodiurn pentherianurn, P. newdigatae) do Phylogeny of the Coryciinae 209 Disperis 3 E.rubignosa Ce.venosa 2 (~e.bicolor 2 Ce.~trat~ 2 Ce.grandiflora 2 , Ce.globosa 2 Oo.dracomontanum 3 Co.nigrescens 3 Co.ingeanum 1 Co.orobanchoides Co.crlspum 2 2 1 Co.bicolorum 2 Co.flanaganii 3 Oo.bifidum 2 C 3 P.volucris 2 P.sohelpei 1 P.magnum 1 2 3 1 P.inversum 2 P.leucanthum 3 P_cruciferum 2 P.connivens P.alatum 2 2 P.caffrum 3 2 ~ P.pentherianum P.hastatum P.platypetal = ~ 4 e.oooperi - 3 ~/ ~ P.ukingense - 4 2 C:o. c a r n o s u m __ 2 ~ Co.miGroglossum - B 1 Co.excisum - / 2 Oo.deflexum g 2 3 Co=tricuspidatum p.hallii 3 3 Co-alticola C 2 2 Ce.harveyana A 1 2 2 / 2 3 um 2 p.acutifolium 2 p.~at:holicum 2 P.newdigatae 2 Fig. 23. Cladistic biogeographic analysis for the Coryciinae s. str. The areas are numbered according to Figs. 21 and 24, and the reduced area cladograms are given on the right-hand side of the full area cladogram. The capital letters on the reduced area cladograms refer to the parts of the full area cladogram from which they are derived 210 H. KURZWEIL& al.: not occur around the shores of False Bay. Of these, two (Ceratandra grandiflora, Pterygodium newdigatae) are found along the southern Cape Coast, and one (P. pentherianum) is centred in the arid portions of the northern parts of the Cape Floristic Region and may be better assigned to the Karoo margins group. Historical patterns In order to analyze the historical relationships among the centres, the distribution centres are added to the species on the cladogram (Figs. 23 and 24) to make an area cladogram (NELSON & PLATNICK 1981, HUMPHRIES • PARENTI 1986). The distributions of the '"ancestors" are determined by summing the areas of the species to the nodes on the cladogram. An inspection of the area cladogram reveals that there are numerous components with two areas, one component with three areas and one with four areas. Two area components are consistent with all possible patterns and can therefore not be used either to corroborate or to falsify any patterns. The three-area and four-area components are consistent with each other, and suggest the relationships [(CFR + Karoo), (Drakensberg + Southern Tanzania)]. This pattern is also consistent with the general "impression" from the area cladogram. In all cases the sister species for Karoo-margin taxa are found in the CFR. As the four Karoo-margin species are taxonomically rather isolated from each other, it would appear as if they are derived from the Cape centre. In three of the four cases (Pterygodium volucris/schelpei; Pterygodium hallii/inversum; Corycium orobanchoides/ingeanum) the species in the pairs are very closely related, and F Equator Zapricorn -- AFRICA Statute Mites ;"2;o 6o 6;o 8;0 ,obo Projection: Sanson-Flarnsteed's Sinusoidal West from 20 Greenwmh lO Copyright 1951, George Philip and Son, Limited, 0 10 20 30 40 East from 50 Greenwich The London Geographical Institute. PRINTED IN GREAT BRITAIN BY GEORGE PHILIP A N D SON, LIMITED, L O N D O N 0tA Fig. 24. Reduced area cladogram mapped onto the geographical distribution of the Coryciinae s. str. in Africa. The sister-area relationships between Namaqualand and the Cape, and between the Drakensberg and southern Tanzania, are clearly demonstrated Phylogeny of the Coryciinae 211 in two of them (Pterygodium volucris/schelpei and Corycium orobanchoides/ingeanum) it would appear that differences in soil types between the two centres may be a better explanation for the existence of the species rather than historical events. Curiously, the one Cape species that occurs to the north of the Cape Peninsula, but does not reach the Peninsula (Pterygodium pentherianum), has a similar ecological relationship with its sister taxon. The species from the Drakensberg centre do not appear to be closely related to the Cape species. Four of the species (Corycium JTanaganii, Pterygodium magnum, P. leucanthum, and P. hastatum) are taxonomically isolated, another four (Corycium dracomontanum, C. nigrescens, C. alticola, and C. tricuspidatum) form a rather closeknit clade, while Pterygodium cooperihas its sister species in southern Tanzania. It suggests a rather closer relationship with central Africa than with the Cape, and further supports the hypothesis that the group has had a long history in the Drakensberg. This is consistent with the patterns described for the Disinae by LINDEF. (1983) and for Lotononis by LINDER & VAN WYK (unpubl.). The geo-climatic implications of this pattern are still not clear. More area cladograms would be needed, especially from other groups, to indicate whether this is a generalized pattern, and to attempt to correlate it to historical geomorphological and palaeoclimatic events. Speciation The patterns of speciation in the group can be analyzed by contrasting sister species. For more distantly related clades, secondary dispersal and adaptation may have obscured the original factors that led to the speciation. If geographical isolation were the dominant factor in speciation it could be expected that the species pairs would be allopatric: If the species are sympatric the isolating factor may be ecological or phenological. The environmental factors for the ten species pairs are summarized in Table 7. The environmental data are compiled from herbarium labels, personal experience during field work, and inference from the distribution patterns. No direct data collection was done. The distributional data are categorized into allopatric pairs in which the distributions are geographically isolated, parapatric pairs in which there is a marginal overlap in the distributions, and sympatric pairs in which the species widely overlap or have identical distribution areas. The environmental data are separated into four categories: where the species are separated altitudinally, where they occur on different soil types, where they are found under different rainfall regimes and, finally, where they are found together in the same habitat. Reproductive data are unfortunately not available, but where species are separated by different flowering times this is indicated. The distributional patterns show that allopatry is rare, occurring in only three pairs. These are inter-centre pairs. Curiously, there are two cases of complete sympatry, one in the CFR and one in the Drakensberg. In both cases there is also no detectable habitat difference between the sister species. The second author has collected both the members of both these species pairs in the same locality, on the same day, in the same type of habitat. It remains a puzzle what caused the speciation in these taxa. Parapatry is by far the most common pattern. Here the sister species occupy different habitats in the same general area leading to an interdigitation of their distribution areas. 212 H. KURZWHL& al.: Table 7. Species pairs, with their environmental correlates. Distributions: 1 allopatric, 2 parapatric, 3 sympatric. Environmental differences: 4 altitudinal, 5 edaphic, 6 rainfall, 7 none (species co-occur). Breeding systems: 8 flowering time differences Species pair Ceratandra harveyana/bicolor Ceratandra globosa/grandiflora Pterygodium caffrum/pentherianum P. volucris/scheIpei P. ukingense/ cooperi P. cruciferum/connivens Pterygodium hallii/inversum Corycium dracomontanum/nigrescens C. orobanchoides/ingeanum C. microgIossum/bifidurn 1 2 3 4 5 6 * * * * 7 8 * * * * * * * * * ? ? * * * * * * * ? * ? The analysis of the environmental factors associated with the species pairs reveals some curious patterns. The two sympatric species have no ecological differentiation. Two pairs show a clear altitudinal separation. In the southern Cape Province Ceratandra grandiflora is found on the coastal platform and on the foothills of the Outeniqua mountains, rarely reaching altitudes of 400m. Its sister species, C. globosa, is widespread in the mountains of the CFR, and is found in the southern Cape along the higher mountain ranges, generally above 1 000 m. Similarly, in the western Cape Province C o r y c i u m microglossum is endemic to the sandy flats from the Cape Peninsula north-wards, rarely reaching 100 m s.m., while its sister species C. bifidum is found in the mountains on the Cape Peninsula and in the Caledon area. A more common difference is in the substrates. There are four, possibly five, species pairs in which there is a substrate difference between the sister species. In one pair, the C. m i c r o g l o s s u m / b i f i d u m pair discussed above, the species on the flats occurs in recent, aeolian sands, while C. bifidum occurs in acid sands derived from Table Mountain Sandstones. The other three pairs all seem to be separated with one species occurring on heavier clayey soils associated with the Karoo Sediments and with Malmesbury Shales, while the sister taxon is found on sandier soils derived from Table Mountain Sandstone and granites (LINDER 1988 a). Associated with most of these substrate and altitude differences would also be a rainfall difference, but this has not been analysed in detail. Phenological differences may account for the C o r y c i u m nigrescens/dracomontanum pair. This is not evident if all the collections are studied, but is visible during fieldwork. This would suggest that geographical isolating is not very important in speciation in the Coryciinae, but that edaphic and altitudinal gradients may play a more significant role. This could be expected in the CFR, due to the extremely steep environmental gradients, and the relatively small area in which a remarkably rich flora occurs (LINDER 1985). Phylogeny of the Coryciinae 213 Classification Coryciinae s. str. P l a n t s herbaceous, geophytic, slender to robust; perennating organ an undivided globular to elongate tuber or rarely a cluster of cylindrical thickened roots; stems solitary, glabrous, sterile shoots always missing, stem bases very rarely with a prominent thickening; leaves cauline, the basal ones frequently reduced to membranaceous scales; lower foliage leaves fiequently basally 4- clustered but never rosette-like, few to many, linear to lanceolate and oblong, lax to imbricate, almost always fresh at flowering time, conduplicate or flat, glabrous and usually flat but rarely marginally ciliate or with crisp or undulate margins, apically grading into the similar floral bracts; basal clusters of linear radical leaves occasionally present. I n f l o r e s c e n c e dense or lax, racemose or rarely corymbose, with few to many small to large resupinate or non-resupinate flowers. F l o w e r s fiat to cup-shaped or globose with narrow flower entrance; colour green, yellow or brown and rarely red or white; bracts fresh but occasionally with dry apices, mostly shorter to somewhat longer than the ovaries, usually lanceolate-acute, adpressed to the ovaries or rarely strongly reflexed; ovaries variously slender. Sepals usually fresh but occasionally dry and dark at flowering time; median sepal usually postgenitally adnate to the petals forming a galea, galea hood-like or shallowly concave, erect or reflexed; median sepal variously lanceolate, apically sub-acute, very rarely with a sac-like spur; lateral sepals spreading or reflexed, free or fused, lanceolate to oblong, mostly flat but occasionally naviculate, sub-acute. P e t a l s with various lobes and sacs, adnate to the median sepal, extremely asymmetrical with enormous anterior portions and very narrow posterior portions; distinct extraordinarily thickened petal nerves present. Lip lanceolate, linear, square, three-lobed or bi-lobed; basally fused to the gynostemium, sessile or stalked, unspurred, almost always with an enormous appendage, occasionally also with a laminal callus; very rarely marginally strongly undulate or lacerate; appendage either sessile at the lip base or somewhat stalked, usually elongate with a terminal callus or cavity or shield-like with various shaped lateral processes, very rarely bipartite. G yn o s t e m i u m short and rarely stalked. A n the r with erect, horizontally reflexed, semi-pendent or pendent thecae; thecae separated and at least their anther canals diverging, outer anther wall usually with an enlarged lobe. R o s t e l l u m either strap-shaped or flat and covering the dorsal side of the thecae, bilobed or obscurely three-lobed, lateral parts frequently variously curved and twisted, bearing the stigmas; terminal portion bearing the viscidia which are exposed to the dorsal side. S t i g m a s two, slightly convex cushions, mostly in the inner portion of the rostellum or rarely in the terminal rostellum portions, stigmas generally derived from the median carpel apex, lateral carpel apices usually not visible. Seeds fusiform, mostly 0.3 - 0 . 6 mm long, seed-coat consisting of one row of parallel large oblong cells in the median seed portion and one or two rows on each end, with straight to undulate anticlinal walls, periclinal walls usually slightly wavy. P o l l i n i a two, sectile, with long caudicles; pollen tetrads linear. A well defined group of 39 species in fynbos, grassland and open shrub in South Africa, with one species endemic to Tanzania. 1. CeratandraEc~LoN ex BAUERin Illustr. Orch. P1.4: t. 16 (1837). Type species: C. chloroleuca ECKLON ex BAUER = C . atrata (L.) DUR. ~; SCHINZ. 214 H. KURZWEIL& al.: P l a n t s perennating with cylindrical thickened roots; leaves few or many, cauline, linear or narrowly lanceolate, dense, sometimes marginally ciliate or coarsely denticulate; basal clusters of radical leaves mostly present; inflorescence lax or dense, racemose or corymbose, with few to many small to medium-sized nonresupinate or resupinate flowers; bracts shorter to slightly longer than the ovaries, adpressed to the ovaries. F l o w e r s flat to slightly cup-shaped, yellow or pink and white; ovaries slender; sepals fresh at the time of flowering; median sepal + reflexed, broadly lanceolate; lateral sepals spreading, free, lanceolate, slightly naviculate; petals broadly lanceolate with obscure or rarely prominent basal anterior lobes and basal claws; lip anchor-shaped to semi-ovate, with flat and wide claw, with or without appendage, frequently with a laminal callus. G y n o s t e mi u m sometimes stalked; anther with pendent thecae which are mostly clearly differentiated into anther canals and wide anther sacs, thecae separated and at least their anther canals diverging, lateral rostellum portions flat and covering the dorsal side of the thecae, carrying the stigmas; stigmas convex cushions in the inner portion of the rostellum. 1 a. Ceratandra ECKLON ex BAUERsect. Ceratandra. P 1a n t s slender to robust; leaves numerous, cauline, linear; frequently with basalcluster of linear radical leaves; inflorescence moderately dense, racemose or corymbose, with many medium-sized non-resupinate flowers. F l o w e r s flat to cup-shaped, yellow or white and pink; lateral sepals spreading and free; petals with obscure to prominent anterior lobes; spreading and free; petals with obscure to prominent anterior lobes; lip 4- anchorshaped, with a pendent flat claw, frequently with a laminal callus; lip appendage insignificant or missing. Gy n o s t e m i u m shortly stalked; anther with pendent thecae, thecae close to one another and only the anther canals diverging; canals stout but narrower than the anther sacs, round, elongate, projecting beyond the point of lip insertion; rostellum flat and covering the thecae at their dorsal sides, bearing the stigmas in their inner portions. Three species in marshes in the southwestern Cape. 1. Ceratandra atrata (L.) DuR. & ScniNz in Conspect. F1. Aft. 5:123 (1892). 2. Ceratandra grandiflora LINDL. in Gen. Sp. Orch. PI.: 364 (1838). 3. Ceratandra globosa LINDL. in Gen. Sp. Orch. PI.: 364 (1838). 1 b. Ceratandra ECKLON ex BAUERsect. Evota LINDL., Gen. Sp. Orch. PI.: 363 (1838). Type species: Ceratandra harveyana LINDL. P l a n t s slender; leaves few, cauline, linear, subterete; frequently with basal cluster of linear radical leaves; inflorescence lax, with few large resupinate flowers. F 1o w e r s cup-shaped, yellow; lateral sepals spreading and free; petals clawed; lip + anchor-shaped, with an obscure pendent flat claw, with a laminal callus; lip appendage prominent, deeply two-lobed. G y n o s t e m i u m : anther with pendent thecae, thecae close to one another and only the anther canals diverging; canals laterally flattened, elongate, projecting beyond the point of lip insertion; rostellum flat and covering the thecae at their dorsal sides, bearing the stigmas in their inner portions. Two rare species in moderately wet places in fynbos in the Southwestern Cape. 4. Ceratandra bicolor SOND. ex H. BOL. in J. Linn. Soc. 20:487 (1884). 5. Ceratandra harveyana LINDL. in Gen. Spec. Orch. PI.: 365 (1838). I c. C e r a t a n d r a EC~:LONex BAUERsect. Venosa KURZWr~L & L~NDra, sect. nov.; Ceratandrae sect. Ceratandrae et sect. Evotae affinis tuberibus palmatis, lamina labio ecallosa et appendice labio biloba, sed differt foliis linearibus basalibus nullis et connectivo lorato. Type species: Ceratandra venosa (LINDL.) SCMLTR. P l a n t s Phylogeny of the Coryciinae 215 slender; leaves numerous, cauline, narrowly lanceolate; inflorescence dense, racemose, with many small resupinate flowers. F l o w e r s rather open; green, pink and white; sepals lanceolate, all three of them similar in shape, free; petals stalked, the upper margin irregularly toothed; lip orbicular, crenulate, with laminal callus; appendage deeply bilobed. G y n o s t e m i u m: anther with pendent thecae separated at the far corners of an elongate connective, without significant anther canals; terminal rostellum portions flat and covering the thecae at their dorsal sides, inner rostellum portion strap-like; stigmas borne where the two portions meet. A single rare species in fynbos in the Southwestern Cape. 6. Ceratandra v e n o s a (LINDL.)SCHLTR. in Bot. Jahrb. 24:433 (1897). 2. Evotella KVRZWrIL & LINDER, gen. nov. A Ceratandrae tuberibus duo, ovoideis, antherarum connectivo ligulato, a Pterygodio foliis linearibus, ab ambobus stigmatibus rostellis apicibus insidens recedit. Type species: Evotel[a rubiginosa (SoND.) KuRzwE~L & LINDER. P la n t s with few cauline and linear leaves; inflorescence usually dense and manyflowered, racemose, with medium-sized resupinate flowers; bracts ovate-acute, as long as and adpressed to the ovary. F1 o w e r s cup-shaped with red petals and white lip; median sepal lanceolate, apically sub-acute; lateral sepals spreading, free; lanceolate, naviculate, subacute; petals with a basal anterior lobe; lip triangular-hastate, sub-acute, with a flat, wide and pendent claw; appendage sessile at the lip base, erect, elongate, apex bilobulate with pendent process. G y n o s t e mi u m: anther with semi-pendent thecae, terminal rostellum portions bearing the stigmas and viscidia. A single rare species in the southwestern and southern Cape. 1. EvoteIla rubiginosa (SoND. ex H. BoL.) Kt~Rzwrn~ & L~NDrR, comb. nov. Basionym: Pterygodium rubiginosum SoND. ex H. BoL., J. Linn. Soc. 20:486 (1884). Type: Cape Province, Rivier Zonder Einde, ZEY~ER 3946 (?S, holo.; K!-BOL, photo.!; SAM!). 3. Pterygodium SWARTZ in Kungl. Svens. Vetenskap. Handl. ser. 2, 21:217 (1800). Type species: P. alatum (L. f.) SWA~TZ. P l a n t s slender to robust; perennating with globular or elongate tubers; stems solitary, glabrous; leaves few to many, cauline but frequently basally :t: clustered, lanceolate to oblong, lax to imbricate, glabrous; inflorescence a dense or lax raceme, with few to many medium-sized to large resupinate or rarely non-resupinate flowers; bracts fresh, almost of the same length as the ovaries and only very rarely longer, lanceolate-acute, adpressed to the ovaries or strongly reflexed. F l o w e r s cupshaped, mostly greenish yellow or brown; sepals fresh at flowering time; median sepal variously lanceolate, rarely with a sac-like spur; lateral sepals spreading and free, lanceolate, flat to naviculate, sub-acute; petals elliptic to ovate, usually unlobed; lip three-lobed, bi-lobed, ovate, lanceolate or linear, sessile or stalked, rarely with undulate or lacerate margins; lip appendage entire, sessile, elongate with a terminal callosity or cavity, erect or reflexed. G y n o s t e mi u m unstalked; anther with erect, horizontal or semi-pendent thecae situated at the far corners of a long horse-shoe shaped untwisted connective; thecae not clearly differentiated into anther canals and anther sacs; rostellum untwisted and strap-shaped adjacent to the connective; stigmas convex cushions in the central portion of the rostellum, rarely in the terminal rostellum portions. A heterogenous group of 18 species occurring in South Africa with one species endemic to Tanzania. 3 a. Pterygodium SWARTZ sect. Magnum KuRzwriL & LINDER,sect. nov., a sect. Pterygodib habitu robusto, a sect. Ommatodio labi appendice solida, a sect. Anochilo 216 H. KURZWEIL& al.: floribus resupinatus differt. Type species: Pterygodium magnum REICnB. f. P 1a n t s very robust, perennating with tubers; leaves many, cauline, imbricate, lanceolate, grading into the floral bracts; inflorescence dense, racemose, with many large resupinate flowers; bracts linear-lanceolate, considerably exceeding the flowers, strongly reflexed. F l o w e r s cup-shaped, whitish and brownish yellow with purple markings; sepals entire, the laterals spreading and free; petals ovate, strongly lacerate; lip semi-ovate, lacerate; appendage sessile at the lip base, elongate with bifid apical portion. G y n o s t e m i u m : anther with ascending thecae, terminal portions bearing the viscidia and club-shaped stigmas. A single apparently isolated species in marshy grassland and river margins of the eastern Cape, Transkei, Orange Free State, Natal and Transvaal. 1. Pterygodium magnum REICHB. f. in Flora 50:117 (1867). 3b. Pterygodium SWARTZ sect. Ommatodium (LINDL.)KURZWEIL & LINDER, comb. nov. Basionym: Ommatodium LINDL. in Gen. Sp. Orch. PI.: 365 (1838). Type: O. volucris (L. f.) LINDL. P l a n t s with 1 - 4 cauline and lanceolate to elliptic foliage leaves, the lowermost the largest; inflorescence dense or lax, racemose, with few to many medium-sized small to large resupinate flowers; bracts strongly reflexed. Flowers cup-shaped, pale green or white; sepals fresh at flowering time; lateral sepals spreading, free; petals with or without significant apical posterior lobes; lip three-lobed, side lobes with callus or callose throughout, sessile; appendage sessile, cup- or funnel-shaped, erect or reflexed. G y n o s t emium: anther with erect thecae situated at the far corners of a long horse-shoe shaped connective; rostellum strapshaped adjacent to the connective and bearing the stigmas; stigmas comparatively small convex cushions facing to the front, situated in the inner portion of the rostellum, lateral carpel apices usually not visible. A well defined group of four apparently closely related species with an interesting geographical distribution. While two species (P. volucris, P. schelpei) are found in the Western Cape, one species (P. cooperi) occurs in South Africa's summer-rainfall area and the fourth species (P. ukingense) is endemic to southern Tanzania. 2. Pterygodium ukingense SCHLTR.in Bot. Jahrb. 53:546 (1915). 3. Pterygodium cooperi ROLFE in F1. Cap. 5, 3:278 (1913). 4. Pterygodium volucris (L. f.) SWARTZin Kungl. Svens. Vetenskap. Handl. ser. 2, 21: 218 (1800). 5. Pterygodium schelpei LINDERin S. African J. Bot. 54:496 - 500 (1988). 3c. Pterygodium SWARTZ sect. AnochflusScnLTR. in Bull. Herb. Boiss. 1, 6:854 (1898). Type species: Pterygodium inversum (TnuNB.) SWARTZ. P la n t s very robust, perennating with tubers; leaves few, cauline, lanceolate, imbricate; inflorescence a dense raceme, with many medium-sized to large non-resupinate flowers; bracts erect and slightly longer than the ovary. F l o w e r s shallowly cup-shaped, somewhat nodding, green with purple markings on the lip; sepals narrowly lanceolate, reflexed, naviculate; lateral sepals free; petals broadly oblong, shallowly concave; lip ascending, semi-ovate, with a flat crenulate claw, apex occasionally slightly emarginate; appendage solid, sessile at the lip base, deflexed and broadly tongue-shaped, unlobed or bilobed, fleshy. G y n o s t e m i u m : anther with horizontal or semi-pendent thecae, outer anther wall with an enlarged lobe; rostellum bearing the stigmas in the inner portion, terminal portions bearing the viscidia; stigmas facing upwards; lateral carpel apices visible or not. Two very similar species in dry open scrub in the western and northwestern Cape. Phylogeny of the Coryciinae 217 6. Pterygodium inversum (THUNB.)SWARTZin Kungl. Svens. Vetenskap. Handl. ser. 2, 21:218 (1800). 7. PterygodiumhalIii (ScnELPE) KURZWEIL& LINDER,comb. nov. Basionym: Anochilus hallii SCHELPEin Contr. Bol. Herb. 10:156 (1982). Type: Cape Province, Calvinia, near Middelpos, H. HALL 4391 (BOL, holo.!). 3 d. Pterygodium SWARTZ sect. Pterygodium. P 1a n t s slender to robust; foliage leaves 1 - 6, cauline, lax, narrowly tanceolate to elliptically oblong; inflorescence lax, racemose, with few large resupinate flowers; bracts fresh, slightly shorter to somewhat longer than the ovary, lanceolate-acute, adpressed to the ovary. F 1o w e r s cup-shaped, greenish- or bright yellow, rarely white; hood broadly galeate, median sepal occasionally with a sac-like spur; lateral sepals spreading or reflexed, free, lanceolate, flat to naviculate; lip ovate, lanceolate, linear or hi-lobed, sessile, rarely with undulate margins; lip appendage erect or reflexed, sessile, elongate and solid with a terminal callus, apex frequently inflexed. G y n o s t e mi u m very wide; anther with horizontal or semi-pendent thecae situated at the far corners of a long horseshoe-shaped connective, outer anther wall usually with an enlarged lobe; rostellum strap-shaped and adjacent to the connective, bearing the stigmas; stigmas convex cushions facing upwards, almost always in the inner portion of the rostellum, very rarely in the terminal rostellum portions; lateral carpel apices visible or not. A homogenous group of 11 species, mainly distributed in the southwestern and western Cape with two species (P. hastatum, P. leucanthum) occurring in South Africa's summer-rainfall area. 8. Pterygodium cruciferum SOND. in Linnaea 19:109 (1847). 9. Pterygodium connivens SCHELPEin Contr. Bol. Herb. 10:161 (1982). 10. Pterygodium acutifolium LINDL. Gen. Sp. Orch. Pl.: 366 (1838). 11. Pterygodium catholicum (L.) SWARTZin Kungl. Svens. Vetenskap. Handl. set. 2, 21:218 (1800). 12. Pterygodiurn newdigatae H. BOL. in Ic. Orch. Austro-Afric. Extratrop. 1: t. 99a (1896). 13. Pterygodium platypetalum LINDL. in Gen. Sp. Orch. PI.: 366 (1838). 14. Pterygodium hastatum H. BOL. in J. Linn. Soc. Bot. (London) 25: 177, t. 14 (1889). 15. Pterygodiurn leucanthum H. BOL. in Trans. S. African Phil. Soc. 16:150 (1907). 16. Pterygodium caffrum (L.) SWARTZin Kungl. Svens. Vetenskap. Handl. set. 2, 21:218 (1800). 17. Pterygodium pentherianum SCHLTR. in Bot. Jahrb. 24:432 (1897). 18. Pterygodium alarum (THuNB.) SWARTZin Kungl. Svensk. Vetenskap. Handl. ser. 2, 21:218 (1800). 4. Coryeium SWARTZin Kungl. Svensk. Vetenskap. Handl. set. 2, 21:220 (1800). Type species: C. orobanchoides (L. f.) SWARTZ (lectotype). P l a n t s slender to robust; perennating with tubers; leaves cauline, few to many, basally + clustered but never rosette-like, linear to narrowly lanceolate, imbricate, non-petiolate, rarely with crisp or undulate margins; inflorescence dense, racemose, with many small resupinate or very rarely non-resupinate flowers; bracts fresh but occasionally with dry apices, shorter to somewhat longer than the ovaries, lanceolate-acute, adpressed to the ovary. F 1o w e r s usually semi-globose or globose with narrow flower entrance; green, yellow or brown, very rarely pink; sepals usually fresh but occasionally dry and dark at flowering time; median sepal broadly to narrowly lanceolate, apically sub-acute; lateral sepals lanceolate, flat, apically subacute, more or less fused and only rarely spreading and free; petals frequently with 218 H. KURZWEIL& al.: various anterior and posterior sacs; lip usually bi-lobed, rarely narrowly lanceolate, square or apically trifid, usually stalked; appendage shield-like with variously shaped lateral processes and clearly stalked or rarely elongate or hood-like. G yn o s t emium: anther with horizontal or semi-pendent thecae situated at the far corners of a long horse-shoe shaped and frequently twisted connective; rostellum strap-shaped adjacent to the connective, frequently variously twisted and contorted, bearing the stigmas; stigmas convex cushions in the inner portion of the rostellum, facing upwards, to the sides or downwards; lateral carpel apices not visible. 14 species in South Africa. 4 a. Corycium SWARTZsect. Carnosum (LINDL.)KURZWEIL& LINDER,sect. nov., habitu Corycio sect. Corycio et sect. Microglosso similis, praecipue differt sepalis lateralibus liberis et appendix labii non scutiformi. Type species: Corycium carnosum (LINDL.) ROLFE. P l a n t s slender to robust; leaves cauline but basally 4- clustered, narrowly lanceolate, imbricate; ovary apex inconspicuously calyculate; inflorescence dense, racemose, with many medium-sized resupinate flowers; bracts fresh, lanceolate-acute, adpressed to the ovary. F l o w e r s semi-globose, pink; lateral sepals spreading and free, lanceolate, fiat or somewhat naviculate; lip bi-lobed with semiovate lobes and narrow base, sessile; appendage sessile at the lip base, forming a beaked hood. G y n o s t e m i u m : anther with horizontal thecae, connective untwisted; rostellum untwisted and broadly strap-shaped; terminal portion bearing the viscidia and the stigmas. One species in montane fynbos and marshes in the southwestern and southern Cape. 1. Corycium carnosum (LINDL.) ROLFE in F1. Cap. 5, 3:284 (1913). 4 b. Corycium SWARTZ sect. Microglossum KURZWEn~& LINDER, sect. nov., differt a sect. Corycio appendice labii sine unguibus superis et sine processibus lateralibus. Type species: Corycium microglossum L~NDL. P l a n t s slender to robust; leaves cauline and frequently basally + clustered, narrowly lanceolate, imbricate, occasionally withered at flowering time; inflorescence dense with many small resupinate or non-resupinate flowers; bracts fresh but membranaceous and frequently with dry apices, narrowly to broadly lanceolate-acute, adpressed to the ovary. F 1o w e r s semi-globose to globose with narrow flower entrance or with secondarily spreading perianth lobes, sepals fresh or dry; lateral sepals basally or strongly fused; lip blade narrowly lanceolate or oblong-obtuse, sessile; lip appendage sessile at the lip base, elongate oblong-obtuse, erect or apically inflexed, with a median keel. G y n o s t e m i u m : anther with horizontal to semi-pendent thecae, connective untwisted; rostellum untwisted, narrowly or broadly strap-shaped adjacent to the connective, bearing the stigmas; stigmas convex cushions in the inner portion of the rostellum and facing upwards. A rather heterogenous and partly poorly known group of three apparently isolated species, with two rare species (C. microglossum, C. bifidum) recorded from the Southwestern Cape, and C. flanaganii occurring in the mountains of the Eastern Cape. 2. Corycium microglossum LINDL. Gen. Sp. Orch. PI.: 369 (1839). 3. Corycium bifidum So~D. in Linnaea 19:111 (1847). 4. Corycium flanaganii (H. BOL.) KURZWEIL& LINDER,comb. nov. Basionym: Pterygodiumflanaganii H. BoL. in Ic. Orch. Austr. Aft. I: t. 98 (1896), asflanagani; Type: Cape Province, Broughton near Molteno, FLANAGAN1639 (BOL, holo.!; K!; PRE). 4c. Corycium SWARTZ sect. Corycium. P l a n t s slender to robust; perennating with tubers; leaves cauline, basally 4- clustered, narrowly to broadly lanceolate, Phylogeny of the Coryciinae 219 imbricate, fresh at flowering time, grading into the floral bracts; inflorescence dense, racemose, with m a n y small resupinate flowers; bracts in most species shorter to somewhat longer than the ovaries, usually lanceolate-acute, adpressed to the ovaries. F l o w e r s globose with narrow flower entrance and only rarely semi-globose; with yellow, brown or white colours; sepals usually fresh but occasionally dry and then black at flowering time; lateral sepals fused for half their length or more, very rarely completely free, apical portions occasionally postgenitally adnate; petals with various anterior and posterior sacs; lip bilobed, rarely square or apically trifid, stalked; lip appendage with a minute claw, shield-like with variously shaped lateral processes. G y n o s t e mi u m: anther with semi-pendent thecae, connective frequently twisted; rostellum strap-shaped and adjacent to the connective, frequently variously curved and twisted; stigmas convex cushions in the inner portion of the rostellum, twisted like the rostellum; lateral carpel apices not visible. A very homogenous group of 10 species distributed in the summer-rainfall area of South Africa (species 5 - 8) and in the western Cape (species 9 - 14). 5. Corycium tricuspidatum H. BOL. in J. Linn. Soc., Bot. 25: 176, t. 13 (1889). 6. Corycium nigrescens SOND. in Linnaea 19:110 (1847). 7. Corycium dracomontanum PARKMAN & SCHELPE in SCHELPE, in Contr. Bol. Herb. 10:158 (1982). 8. Corycium alticola PARKMAN & SCHELPE in SCHELPE, in Contr. Bol. Herb. 10:157 (1982). 9. Corycium bicolorurn (THUNB.) SWARTZin Kungl. Svensk. Vetenskap. Handl. ser. 2, 21:220 (1800). 10. Corycium excisum L~NDL. in Gen. Spec. Orch. PI.: 368 (1839). 11. Coryciurn orobanchoides (L. f.) SWARTZin Kungl. Svensk. Vetenskap. Handl. ser. 2, 21:222 (1800). 12. Corycium ingeanum OLiVeR in S. African J. Bot. 52:256-260 (1986). 13. Coryeium deflexum (H. BOL.) ROLFE in F1. Cap. 5, 3:288 (1913). 14. Corycium crispum (THuNB.) SWARTZin Kungl. Svensk. Vetenskap. Handl. ser. 2, 21:222 (1800). We are grateful to the FRD and the Smuts Memorial Fund for funding. We would like to thank the staff of the Bolus Herbarium and the Director of the Herbarium, Kew for the use of material, and the various authorities (Cape Department of Nature and Environment Conservation, Divisional Council of the Cape, Cape Town City Council, Natal Parks Board) for permission to collect live material. We also thank various plant enthusiasts and botanists for their help in the field and for providing specimens, in particular to P. DREWE, Prof. P. JACKSON,Dr K. STEINER,and J. VLOK. For the use of SEM- and darkroom-facilities at the Electron Microscope Unit of the University of Cape Town we express our thanks to D. GERNEKE and his staff. Special thanks are also due to Dr J. MANNINGfor making most of the line drawings. References ANTHONY, N. A., SCHELPE, E. A., 1982: New species and new combinations in African Pteridophyta and Orchidaceae. - Contr. Bol. Herb. 10: 143- 164. BARTHLOTT, W., 1976: Morphologie der Samen von Orchideen in Hinblick auf taxonomische und funktionelle Aspekte. - Proc. Eight World Orchid Conference, 444- 453. Frankfurt: Deutsche Orchideengesellschaft. -- ZIEGLER,B., 1981: Mikromorphologie der Samenschalen als systematisches Merkmal bei Orchideen. - Ber. Deutsch. Bot. Ges. 94: 267-273. 220 H. KURZWEIL & al.: BENTHAM, G., HOOKER,J. D., 1883: Genera plantarum 3. - London: Reeve & Co. BoLus, H., 1888: The orchids of the Cape Peninsula. - Trans. S. African Phil. Soc. 5: 1 - 201. - 1 8 9 3 - 1896: Icones Orchidearum Austro-Africanarum Extratropicarum 1. - London: Wesley & Sons. - 1911: Icones Orchidearum Austro-Africanarum Extratropicarum 2. - London. Wesley & Sons. - 1913: Icones Orchidearum Austra-Africanarum Extratropicarum 3. - London. Wesley & Sons. - 1918: The orchids of the Cape Peninsula. 2nd edn. - Cape Town: Darter Bros. & Co. BOND, P., GOEDBLATT,P., 1984: Plants of the Cape Flora. A descriptive catalogue. - J. S. African Bot., Suppl. 13. BREMER, K., 1976: The genus R e l h a n i a ( C o m p o s i t a e ) . - Opera Bot. 40. CARPENTER, J. M., 1988: Choosing among multiple equally parsimonious cladograms. Cladistics 4:291 - 296. CHESSELET,P., 1989: Systematic implications of leaf anatomy and palynology in the D i s i n a e and C o r y c i i n a e ( O r c h i d a c e a e ) . - Masters thesis, unpublished; University of Cape Town. CLIFFORD, H. T., SMITH, W. K., 1969: Seed morphology and classification of the O r c h i d a c e a e . - Phytomorphology 19: 1 3 3 - 139. DAVIS, P. H., HEYWOOD, V. H., 1963: Principles of angiosperm taxonomy. - Edinburgh, London: Oliver & Boyd. DRESSLER, R. L., 1981: The o r c h i d s - n a t u r a l history and classification. - Cambridge, Mass., London: Harvard University Press. - 1983: Classification of the O r c h i d a c e a e and their probable origin. - Telopea 2: 413 - 4 2 4 . EDWARDS, D., LEISTNER,O. A., 1971: A degree reference system for citing biological records in Southern Africa. - Mitt. bot. Staatssamml. Miinchen 10: 5 0 1 - 509. FARRIS, J. S., 1 9 8 8 : H E N N I G 8 6 Reference. - Published privately. GARAY, L. A., 1960: On the origin of the O r c h i d a c e a e . - Bot. Museum Leaflets 19: 5 7 - 95. GERSTBERGER, P., LEINS, P., 1978: REM-Untersuchungen an Bliitenknospen von P h y s a l i s p h i l a d e l p h i c a ( S o l a n a c e a e ) . - Ber. Deutsch Bot. Ges. 91: 3 8 1 - 387. GIBBS-RUSSEL, G. E., 1985: Analysis of the size and composition of the southern African flora. - Bothalia 15: 6 1 3 - 629. GOLDBLATT, P., 1978: An analysis of the flora of Southern Africa: its characteristics, relationships, and origins. - Ann. Missouri Bot. Card. 65: 3 6 9 - 4 3 6 . - 1982: Systematics of F r e e s i a KLATT ( I r i d a c e a e ) . - J. S. African Bot. 48: 3 9 - 9 1 . - 1987: Systematics of the Southern African genus H e x a g l o t t i s ( I r i d a c e a e - I r i d o i d e a e ) . Ann. Missouri Bot. Card. 74: 5 4 2 - 5 6 9 . GooD, R., 1974: The geography of the flowering plants. 4th edn. - London: Longman. HILLIARD,O. M., BURTT, R. L., 1987: The botany of the southern Natal Drakensberg. Cape Town: National Botanic Gardens. HOLMGREN, K., KEUKEN, W., SCI4OFIELD, E. K., 1981: Index Herbariorum 1, 7th edn. Regnum vegetabile 106. HUMPHRIES,C. J., PARENTI,L., 1986: Cladistic biogeography. - Oxford: Clarendon Press. KRNNZLIN, F., 1897: Orchidacearum genera et species. - Berlin: Mayer & Miiller. KURZWEIL, H., 1987 a: Developmental studies in orchid flowers I: Epidendroid and vandoid species. - Nordic J. Bot. 7: 4 2 7 - 4 4 2 . - 1987 b: Developmental studies in orchid flowers II: Orchidoid species. - Nordic J. Bot. 7: 4 4 3 - 4 5 1 . - 1989: Floral morphology and ontogeny in H u t t o n a e a p u l c h r a . - Lindleyana 4:1 - 5. - 1990: Floral morphology and ontogeny in O r c h i d a c e a e subtribe D i s i n a e . - Bot. J. Linn. Soc. 102: 6 1 - 83. Phylogeny of the Coryciinae 221 199 l: The unusual structure of the gynostemium in the Orchidaceae-Coryciinae. - Bot. Jahrb. 112: 2 7 3 - 293. LEwis, O. J., 1950: Orchidaceae. - In ADAMSON, R. S., SALTER, T. M., (Eds.): Flora of the Cape Peninsula, pp. 2 6 5 - 3 0 9 . - Cape Town: Juta. LINDER, H. P., 1981 a: Taxonomic studies on the Disinae: l. A revision of the genus Brownleea LINDL. -- J. S. African Bot. 4 7 : 1 3 - 48. - 1981 b: Taxonomic studies on the Disinae: II. A revision of the genus S c h i z o d i u m LINDL. -- J. S. African Bot. 47: 3 3 9 - 3 7 1 . 1981 c: Taxonomic studies on the Disinae. III. A revision o f Disa BERG. excluding sect. M i c r a n t h a e LINDL. -- Contr. Bol. Herb. 9:1 - 3 7 0 . - 1981 d: Taxonomic studies in the Disinae (Orchidaceae) IV. A revision of Disa BERG. sect. M i c r a n t h a e LINDL. -- Bull. Jard. Bot. Nat. Belgique 51: 2 5 5 - 3 4 6 . - 1981e: Taxonomic studies in the Disinae. V. A revision of the genus M o n a d e n i a . Bothalia 13: 3 3 9 - 3 6 3 . - 1981 f: Taxonomic studies in the Disinae. VI. A revision of the genus Herschelia. Bothalia 13: 3 6 5 - 3 8 8 . - 1983: The historical phytogeography of the Disinae (Orchidaceae). - Bothalia 14: 565 - 570. 1985: Gene flow, speciation, and species diversity patterns in a species-rich area: the Cape Flora. - In VRBA, E. S., (Ed.): Species and speciation. - Transvaal Museum M o n o g r a p h 4: 5 3 - 57. - 1986: Notes on the phylogeny of the Orchidoideae, with particular reference to the Diseae. - Lindleyana 1: 5 1 - 6 4 . 1988 a: Taxonomic notes on some orchids from the soutwestern Cape Province, South Africa. - S. African J. Bot. 54: 4 9 6 - 5 0 0 . 1988 b: A review of cladistics for botanists. - S. African J. Bot. 54: 2 0 8 - 2 2 0 1990: On the relationship between the vegetation and floras of the Afro-montane and the Cape regions in Africa. - Mitt. Bot. Gart. Hamburg. 23 b: 7 7 7 - 790. - KURZWEIL, H., 1990: Floral m o r p h o l o g y and phylogeny in the Orchidaceae-Disinae. Bot. J. Linn. Soc. 102: 2 8 7 - 3 0 2 . - VLOK, J. H., 1991: The morphology, taxonomy and evolution of R h o d o c o m a (Restionaceae). - P1. Syst. Evol. 175: 139-160. LINDLEY, J., 1833: Genera Ceratandra, O m m a t o d i u m , P t e r y g o d i u m , Corycium, and Disperis. In: The genera and species of orchidaceous plants. - London: Ridgways. MADDISON, W. P., DONOGHUE, M. J., MADDISON, D. R., 1984: Outgroup analysis and parsimony. - Syst. Zool. 33: 8 3 - 1 0 3 . MANNING, J. C., LINDER, H. P., 199l: Pollinators and evolution in Disperis (Orchidaceae), or why are there so m a n y species. - S. African J. Sci. (in press). NELSON, G., PLATNICK, N., 1981: Systematics and biogeography. - New York: Columbia University Press. OLIVER, E. G. H., 1986: A new species of orchid from the northwestern Cape. - S. African J. Bot. 52: 2 5 6 - 2 6 0 . LINDER, H. P., ROURKE, J. P., 1983: Geographical distribution of present-day Cape taxa and their phytogeographical significance. - Bothalia 14: 4 2 7 - 4 4 0 . PFITZER, E., 1889: Orchidaceae. - In ENGLER, A., PRANTL, K., (Eds.): Die natfirlichen Pflanzenfamilien. 2, 6, pp. 5 2 - 2 2 0 . - Leipzig: Engelmann. PHILLIPS, E. P., 1917: A contribution to the flora of the Leribe Plateau and environs. Ann. S. African Mus. 16: 1 - 379. PIJL, VAN DER, L., DODSON, C. H., 1966: Orchid flowers: their pollination and evolution. Coral Gables (Florida): University of Miami Press. PLATN~CK, N. I., 1989: An empirical comparison of microcomputer parsimony programs, II. - Cladistics 5: 1 4 5 - 161. - - - - - - - - - 222 H. KURZWEIL& al.: ROHLF, F. J., 1988: NTSYS-pc. Numerical taxonomy and multivariate analysis system. New York: Exeter Publishing. ROLFE, R. A., 1913: C o r y c i e a e . - In THISELTON-DYER,W. T., (Ed.): Flora Capensis 5.3. - London: Reeve & Co. SCnELPE, E. A., 1966: An introduction to South African Orchids. - Cape Town, Johannesburg: Purnell & Sons. SCHLECHTER,R., 1898: Monographie der D i s p e r i d e a e . - Bull. Herb. Boissier., ser. 1, 6: 800- 821,846- 859, 905- 955. - 1902: Monographie der D i s e a e . - Bot. Jahrb. Syst. 31: 134-313. - 1926: Das System der Orchideen. - Notizbl. Bot. Gart. Berlin-Dahlem 9: 563- 591. SENGHAS, K., 1973-1974. Unterfamilie O r c h i d o i d e a e . - In BRIEGER, F. C., MAATSCH, R., SENGHAS,K., (Eds.): R. SCHLECHTER,Die Orchideen. 3rd edn. - Berlin: Parey. SHARMAN~B. C., 1939: The development of the sinker in O r c h i s m a s c u l a LINN. -- Bot. J. Linn. Soc. 52: 145-158. STEINER, K. E., 1989: The pollination of D i s p e r i s ( O r c h i d a c e a e ) by oil-collecting bees in Southern Africa. - Lindleyana 4: 164- 183. STEWART,J., LINDER, H. P., SCHELPE,E. A., HALL, A. V., 1982: Wild orchids of southern Africa. - Johannesburg: Macmillan South Africa. SUMMERHAYES,V, S., 1968: O r c h i d a c e a e , 1. - In MILNE-REDHEAD, E., POLHILL,R. M., (Eds.): Flora of tropical East Africa. - London: Crown Agents for Oversea Governments and Administration. TAKHTAJAN,A!, t986: Floristic regions of the world. - Berkeley: University of California Press. VERMEULEN,Pi, 1966: The system of the O r c h i d a l e s . - Acta Bot. Neerl. 15: 224- 253. VOOEL, S., 1959: Organographie der Bliiten kapl/indischer Ophrydeen. - Akad. Wiss. Lit., Math. Naturwiss. KI. (Mainz) 6 - 7 : 1 - 268. WEIMARK,H., ! 941: Phytogeographical groups, centres and intervals within the Cape flora. - Lunds Univ. Arsskr. Avd. 2 37,5:1 - 143. WHITE, F., 1983: The vegetation of Africa. - Switzerland: Unesco. Addresses of the authors: H. KURZWEIL, H. P. L~NDER, and P. CHESSELET*, Bolus Herbarium, University of Cape Town, 7700 Rondebosch, South Africa. * Current address: Grassland Research Center, P. Bag X05, Lynn East 0039, South Africa. Accepted October 24, 1990 by A. WEBER Appendix I: List of specimens studied The following list cites all specimens which have been studied more thoroughly. The nomenclature follows the classification presented here. Collections preceeded by * are preserved in alcohol. All of the other specimens are herbarium material. The collections are deposited in the Bolus Herbarium, University of Cape Town, except the specimen of P t e r y g o d i u m u k i n g e n s e which was on loan from Kew, England. Specimens which are not precisely identified by a particular collector's number are listed together with the number attached to them (the BOLp Number indicates the number in the pickled collection, BOL is the number in the herbarium collection). (L.) DUR. & SCHINZ: *JACKSONs.n., BOLp 1314; *LINDER4473; LINDER 7476; BEAN 173; LINDER 1476; BEAN 757; TAYLOR 248. -- C. b i c o l o r SOND. ex BOL.: *STEINER s.n., BOLp 1841; LINDER4474. -- C. g l o b o s a LINDL.: *AuGUSTIN s.n., BOLp 1375; *LINDER4485; ESTERHUIZEN7299; VLOK88. -- C. g r a n d i f l o r a L~NDL.:*KuRzWEIL 1226; *STEINER s.n., BOLp 1839; MIDDLEMOST 1806; FOURCADE 975; SCHELPE 7115. Ceratandra atrata Phylogeny of the C o r y c i i n a e 223 - C. h a r v e y a n a LINDL.: OLIVERs.n., BOLp 67; *KuRzWEIL 1020. -- C. v e n o s a (LINDL.) SCHLTR.: *MoSTERT s.n., BOLe 1844; sine collector, BOL 528l. E v o t e l l a r u b i g i n o s a (SOND.) KURZWEIL& LINDER : *BOLp 299; *KuRzWEIL s.n.; *LINDER 4442; *OLIVER, s.n., BOLe 98; *BOLp 85. C o r y c i u m a l t i c o l a PARKMAN & SCHELPE: *LINDER 1027; *BATTEN 22. -- C. b i c o l o r u m (THUNB.) SWARTZ: * WALTERS 540; *EsTERHUIZEN 3412, *DucKITT s.n., BOL e 83; ESTERHU~ZEN 34453; PETERSEN & DAHLGRENs.n. -- C. b i f i d u m SOND.: BOLUS 10568; STOKOE 17854. -- C. c a r n o s u m ( L I N D L . ) ROLFE: *KuRzWEIL903; KURZWEIL996; LINDER 1689. -- C. c r i s p u m (THUNB.) SWARTZ: *JACKSON s.n., BOL e 1285; *L~NDER 4314; LINDER 1147. -- C. d e f l e x u m (BoL.) ROLFE: *ScHELPE 8088; SCHELPE 8088. -- C. dracomontanum PARKMAN & SCHELPE: *SCHELPE s.n., BOLp 303; *LINDER 4666; McLOUGHIN 200; JACOBSZ 135/5. -- C. e x c i s u m LINDL.: *KuRzWEIL 1026; LINDER 1805. -- C. f l a n a g a n i i (H. BOL.) KURZWEIL& L~NDER : *BmALKE s.n., BOLp 1393; • KuRzWHL 1318; LINDER4852; FLANAGAN1639. -- C. m i c r o g l o s s u m LINDL.: *ESTERHUIZEN 35545; CARNERON & MANCHIP 50. -- C. n i g r e s c e n s SOND.: *ScHELPE 7198; • HALL 962; *KuRzWEIL 1308; *L1NDER 4665; McLOUGHIN 172; LINDER 914; SCHELPE 6324; L1NDER 4665; HALL 841. -- C. o r o b a n c h o i d e s (L. f.) SWARTZ: *KuRzWEIL 957; • LINDER 4227; LEWIS22241; SCHELPE2726; CASSADY21. P t e r y g o d i u m a c u t i f o l i u m LINDL.: *BOLp 980; *L~NDER4449; LINDER 1679; L1NDER 4449. P . a l a t u m (THuNB.) SWARTZ: *KuRzWEIL 949; *L~NDER 4296, LEW~S 584. -- P. c a f f r u m (L.) SWARTZ: *KuRzWEIL 955; *KuRzWEIL 1217; *LINDER 4220; GIFFEN 109. P . c a t h o l i c u m (L.) SWARTZ: *BOLp 1008; KURZWEIL851; LINDER 4216; LEWIS 583; McLou~HIN 29. -- P . c o n n i v e n s SCHELPE: BOLp 1829. - P . c o o p e r i ROLFE: *LINDER 956; LINDER 1050; L~NDER993; McLOUGHINs.n., BOLp 40307. - P . c r u c i f e r u m SOND.: • DuCKITT s.n., BOLe 1002. - P . hallii (ScnELPE) KURZWEIL& LINDER: *BOLe 1817; • LINDER4227; WINTER 182; PILLANS6968. -- P . h a s t a t u m H. BOL.: *BATTEN 1; SCHELPE 7193. -- P . i n v e r s u m (THuNB.) SWARTZ: *BOLe 1830; BOL 4335; LEWYNS3272. -- P. l e u c a n t h u m H. BOL.: *LINDER954; BOARDMAN245. -- P . m a g n u m REICH,. f.: *ScHELPE 7273; *ScHELPE 7196; LINDER 879; SCHELPE 6336; LINDER 938. -- P . n e w d i g a t a e H . BOL.: ROGERS 28480. var. c l e i s t o g a m u m H. BoL.: *MANNING s.n., BOL e 984. - P. p e n t h e r i a n u m SCHLTR.:*LINDER4198; LINDER 4198. -- P . p l a t y p e t a l u m LINDL.: *KURZWEIL872; LINDER4219. -- P . u k i n g e n s e SCHLTR.:STOLZ2609. -- P . s c h e l p e i LTNDER: SCHELPE s.n. -- P . v o l u c r i s (L. f.) SWARTZ: *L~NDER 4223; *BOLe 1843; HALL 953; BOLe 947. - - - - Verleger: Springer-Verlag K G , Sachsenplatz 4-6, A-1201 Wien. - Herausgeber: Pro£ Dr. Friedrich Ehrendorfer, Institut ffir Botanik und Botanischer Garten der UniversitS.t Wien, Rennweg 14, A-1030 Wien. - Redaktion: Rennweg 14, A-1030 Wien. - Hersteller: A d o l f Holzhausens Nfg., Kandlgasse 19-21, A-1070 Wien. - Verlagsort: Wien. - Herstellungsort: Wien. - Printed in Austria.