Evolution in sedges (Carex, Cyperaceae)

A. A. REZNICEK University of Michigan Herbarium, North University Building, Ann Arbor, MI 48/09, U.S.A. Received January 2, 1990

REZNICEK,A. A. 1990. Evolution in sedges (Carex, Cyperaceae). Can. J. Bot. 68: 1409-1432. Carex is the largest and most widespread genus of Cyperaceae, but evolutionary relationships within it are poorly under- stood. Subgenus Primocarex was generally thought to be artificial and derived from diverse multispicate species. Relation- ships of rachilla-bearing species of subgenus Primocarex, however, were disputed, with some authors suggesting derivation from other genera, and others believing them to be primitive. Subgenus Indocarex, with compounded inflorescence units, was thought to be primitive, with subgenera Carex and Vignea reduced and derived. However, occurrence of rachillas is not confined to a few unispicate species, as previously thought, but is widespread. The often suggested connection between Uncinia and unispicate Carex is shown, based on rachilla morphology, to be founded on incorrect interpretation OF homology. Uncinia kingii, the alleged connecting link, is, in fact, a Carex. Unispicate Carex without close multispicate relatives probably originated from independent, ancient reductions of primitive, rachilla-bearing, multispicate Carex. The highly compounded inflorescences occumng in subgenus Vignea are hypothesized to represent a primitive state in Carex, and the more specialized inflorescences in subgenus Carex derived from inflorescences of this type. The relationships of subgenus Indocurex, with its unique perigynium-like inflorescence prophylls, remain unclear.

REZNICEK,A. A. 1990. Evolution in sedges (Carex, Cyperaceae). Can. J. Bot. 68 : 1409-1432. Le Carex est le genre le plus irilportant et le plus rCpandu des Cyperaceae, mais les affinites Cvolutives a I'intCrieur de ce genre sont ma1 connues. Le sous-genre Primocarex Ctait gCnCralement considCrC commeztant artificiel et dCrivC d'une gamme d'espbces a plusieurs Cpis. Les affinitCs des espbces a rachColes du sous-genre Primocarex Ctaient, cependant, mises en question, certains auteurs invoquant la possibilitC d'une dirivation d'autres genres, d'autres les supposant primitives. Le sous-genre Indocarex, avec ses unitCs d'inflorescences glomCrulaires, Ctait considCrC comme primitif, et les sous-genres Carex et Vignea rkduits et dCrivCs. Cependant, la prCsence de rachColes ne se limite pas a quelques espbces a Cpi solitaire, comme on le croyait; elle est en fait trbs rkpandue. Le lien prCsumC entre le Uncinia et les Carex a Cpi solitaire est identifii, d'aprks la morphologie des rachkoles, comme Ctant fond6 sur une interpretation incorrecte de I'homologie. Le Uncinia-kingii, le chainon de liaison prCsumC, est en fait un Carex. Les Carex a Cpi solitaire sans proches parents a plusieurs Cpis sont probablement issus de forrnes rkduites, anciennes et indkpendantes de Carex primitifs a rachColes et a plusieurs Cpis. Les inflorescences hauternent glornCrulaires prCsentes dans le sous-genre Vignea representent, par hypothese, un Ctat primitif chez le Carex, et les inflorescences plus spCcialisCes dans le sous-genre Carex sont dCrivCes d'inflorescences de ce type. Les affinitCs du sous-genre Indocarex, avec ses prophylles d'inflorescences ressemblant a des pCrigynes, demeurent incertaines. [Traduit par la revue]

Introduction Systematic position of Carex within the Cyperaceae Carex L., with about 2000 species, is by far the largest Although this paper is concerned with evolution within genus in the Cyperaceae and one of the most widespread and Carex, a brief diversion to establish the context within which ecologically important genera of vascular plants. Remarkably, Carex fits is necessary. Hypotheses of evolution within Carex evolutionary trends within the genus are poorly understood, invariably become entangled with the problems of generic lim- and published work on this topic is sketchy. This paper will its within the tribe Cariceae Kunth ex Dumort.' and frequently both briefly review past work on evolution in Carex and pre- also invoke the phytogeography of other genera in the tribe. sent new or overlooked information that may shed light on past As well, discussions of inflorescence structure in Carex rely problems or suggest alternative hypotheses for certain aspects heavily on the character states displayed by some of the other of the evolution of sedges. The review of past work will con- genera in the Cariceae. centrate primarily on more recent work that actually proposed With about 4500 to 5000 species in 100 to 105 genera (Goetghebeur 1987), the Cyperaceae are probably the seventh hypotheses rather than enumerate the schemes of the numerous largest family of vascular plants. The subfamily Caricoideae authors that merely developed an arrangement convenient for Pax, with one tribe, Cariceae, to which Carex belongs, com- listing the species in their local region. Since this review is prises five genera and about 2100 species, thus constituting concerned with evolution within Carex, it will also not touch nearly half the family. Almost all the species in the subfamily on the extensive literature concerning topics such as the der- belong to the enormous genus Carex. ivation of the perigynium and the controversy over whether or The tribe Cariceae is characterized by uniformly unisexual not male flowers are true flowers or synanthia. However, a flowers with the female subtended by a partially or wholly brief discussion of terminology, especially as it relates to Carex closed perigynium of prophyllar origin (Blaser 1944). The five inflorescences, is included, since varied and conflicting ter- genera here recognized are Carex, Cymophyllus Mackenzie, minology is in use in recent literature. While developing a Kobresia Willd., Schoenoxiphium Nees, and Utzcinia Pers. AS natural classification of Carex reaching to the species or sec- noted by Mora-Osejo (1982) and Wheeler (1989), the South tional level is not yet feasible, comments are brought to bear American segregate genus Vesicarex (Steyermark 1951; Cleef on the classification at the subgeneric level. A few previously 1982) is referable to Carex sect. Abditispicae G. Wheeler. The proposed hypotheses are discussed and rejected, but unfortu- nately, more questions are raised than are answered. 'Authors for suprageneric names follow Goetghebeur (1985).

Printed in Canada i Imprim6 au Canada 1410 CAN. J. BOT. VOL. 68, 1990 peculiar collarlike appendage noted by Steyermark (195 1) may tiated inflorescences, Schoenoxiphium has often been regarded be an artifact caused by separation of the pericarp and seed as the most primitive genus in the Cariceae. coat, which sometimes occurs when immature material of A fundamental blurring of generic limits may occur between Carex is softened too long in water before dissection. The Carex aand Kobresia, with certain species traditional!^ put in small, South American genus Bisboeckelera Kuntze (Hoppia Carex, but very similar to Kobresia except for the more Nees non Sprengel), sometimes cited as belonging in the tribe reduced rachilla (Nelmes 1952). Some of these, such as Cariceae (Jermy et. al. 1982), is here considered to belong in C. bucharica Kiikenth., C. hepburnii Boott, and C. nardina the subfamily Sclerioideae C. B. Clarke, tribe Bisboeckeler- Fries were, in fact, transferred to Kobresia by Ivanova (1939). eae Pax ex L. Eiten (Koyama 1965; Smith and Faulkner 1976; In addition, the considerable similarity of Schoenoxiphium to Meert and Goetghebeur 1979). members of Carex subgenus Indocarex has not gone unno- Except for Cymophyllus, which has unique, flat, sheathless ticed, and may indicate a close relationship (Haines and Lye leaves without a midvein or ligule, the genera are all essen- 1983). Some species of Schoenoxiphium have also been placed tially indistinguishable vegetatively (Metcalfe 1969, 1971). in Carex in the past by Clarke (1898, 1908). The delimitation of the genera based on reproductive charac- Carex is cosmopolitan in distribution, but with the great ters is also difficult, with apparently transitional species link- majority of the species in the north and south temperate re,'OIO~S ing all the genera except Cymophyllus. The brief outline fol- and the montane tropics. Uncinia, with perhaps 50 species lowing is, of necessity, inadequate since the topic of generic (Kukkonen 1967a), is predominantly Southern Hemisphere, limits is beyond the scope of this work. with occurrences north of the Equator from northern South Carex, Cymophyllus, and Uncinia are characterized by a America to Mexico and the West Indies, and in Hawaii. completely closed perigynium. Only in Carex subgenus Vig- Kobresia, with about 40 species (Ivanova 1939), is best devel- nea (P. Beauv. ex Lestib. f.) Peterm. does the perigynium oped in the Himalayas and adjacent regions of central Asia, have an abaxial false suture (Mackenzie 193 1; Nannfeldt 1977) with a few boreal species and one in alpine Sumatra (Kern that may be a vestige of a partially open perigynium. As well, 1958). Schoenoxip&um, with 17 species now known in all three genera, there is a distinction between spikes and (Kukkonen 1983, 1986a), is restricted to southern and eastern one-flowered spikelets, as defined by Smith and Faulkner Africa and Madagascar. 'The monotypic Cymophyllus is a (1976). In Uncitzia, the rachilla (the vestigial continuation of southern Appalachian Mountain endemic, occurring from the axis that bears the female flower laterally) is smooth, stiff, Pennsylvania to Georgia (Clarkson 1962). exserted beyond the orifice of the perigynium, and tipped by a retrorse, inrolled scale to form a hook. Also, the orientation of the perigynia of Uncinia with respect to their axes is said Inflorescence terminology to be slightly different than in Carex (Meert and Goetghebeur Before any discussion of morphology and infrageneric clas- 1979). Alone among Cyperologists, Koyama (1961) suggested sification in Carex can be entertained, terminology must be submerging Uncinia within Carex, believing it sprang from clarified. It is not possible to review here the complex and within Carex and that its rachilla was not a generic character, often disputed field of terminology of Cyperaceae infloresc- since rachillas also occur in perigynia of Carex. However, in ences or even Carex inflorescences. Rather, Fig. 1 illustrates Carex and Cymophyllus, the rachilla, when present, is almost basic terminology used in this paper. The term spike is used never exserted and never tipped by a retrorse scale. Uncinia here for the penultimate divisions of the inflorescence and is and Cymophyllus have as their inflorescence a solitary, andro- used merely for convenience. These structures are not true gynous spike. In Carex, the inflorescence is highly variable, spikes but rather compound structures. Also, there is doubt, but few species have only a solitary androgynous spike. Cymo- as outlined later in this paper, that the units are homologous phyllus is entomophilous (Thomas 1984), a pollination syn- drome unknown in Uncinia and very rare in Carex (Hesse among all three subgenera. The term spikelet, often used, is 1980). here abandoned, primarily because of its use sometimes for an Kobresia and Schoenoxiphium usually have a more or less entire inflorescence unit and sometimes for the unit of peri- open perigynium. Both genera also have well-developed rach- gynium and enclosed achene, and also because of the defini- illas. In those species of both genera with compound inflo- tion problems outlined by Levyns (1945) and Kukkonen rescences, there is a lack of differentiation of the inflorescence (1984). Not illustrated in Fig. 1, the term rachilla is used for into spike and spikelet (Smith and Faulkner 1976; Timonen the reduced axis, within the perigynium, found in some Carex. 1985, 1989). Branching in the inflorescence thus is both by In some species of the related genera Kobresia and Schoenox- continued development of the rachilla (spikelet axis) and by iphium (Timonen 1985, 1889), as well as in teratological additional branching of the inflorescence axis. In Carex spe- Carex, there is a continuum between the rachilla and the spike cies with compound inflorescences, branching is almost never axis. Nevertheless, the term rachilla is here used for clarity, by development of the rachilla, except in teratological forms even though it merely describes a stage in the reduction of an (but see the discussion of subgenus Indocarex Baillon). axis, rather than an independent organ. Kobresia and Schoetzoxiphium have sometimes been merged Two specific terminological items need clarification in (Koyama 1961), but differ in that the rachilla of Schoenoxi- Fig. 1. First, a single peduncled unit of an inflorescence in phium is strongly flattened and several-veined, whereas that Carex is simply called an inflorescence unit. This avoids the of Kobresia is more or less terete to sometimes flattened, but contradictory term compound spike seen in some works. Sec- one- or two-veined. Inflorescence bracts, if present in Kobre- ond, the sessile, perigynium-like prophylls at the base of spikes sia, are always small and evaginate. In Schoenoxiphium, usu- in subgenus Indocarex are here termed inflorescence prophylls ally large and leaf-like inflorescence bracts are always present to distinguish them from cladoprophylls found near the base and in many species have an elongate sheath at least on the of inflorescence unit peduncles in subgeiiera Indocarex and lowermost bract. Because of its relatively large, undifferen- Carex. FIG. 1. Subgenera in Carex. Left: subgenus Vignea, C. sparganioides (drawn from Bucks Co., Pennsylvania, Hermatln 4299, MICH). Center: subgenus Carex, C. tetanica (drawn from Norfolk Co., Ontario, Reznicek 5531, MICH). Right: subgenus Indocarex, C. standleyana (drawn from Chiapas, MCxico, Breedlove 52083, CAS). Horizontal bars = 1 mrn and vertical bars = 5 rnm. 1412 CAN. J. BOT. VOL. 68, 1990 Terminological schemes for inflorescences in Cyperaceae accepted, nor is St. John and Parker's (1925) subgenus and Carex based on the typological method of Troll and Altericarex. Weberling, as used by Kukkonen (1984. 1986), or terminol- A detailed morphological survey of the three subgenera rec- ogy based on the anthoid concept of Meeuse, as defined by ognized here is not practical. Rather, Table 1 summarizes sali- Meert and Goetghebeur (1979), offer the promise of much ent differences among them and Fig. 1 illustrates these needed stability and uniformity but are not adopted here for differences. primarily pragmatic reasons. Some of the terms are rather Subgenus Vignea, with perhaps 400 to 500 species, is gen- unfamiliar. Also, while these terminologies are based on com- erally considered to be a homogeneous, natural group. mendable uniform application over the entire family, precise Subgenus Carex, with possibly 1400 species, is very diverse homology of structures to which the same names are applied in morphology with an especially great degree of v-xiability among various tribes and subfamilies is far from established. in the sexual expression of the spikes. It may prove to be poly- The peculiar terminology espoused by Gilly (1952) specifi- phyletic, but too little is known for further subdivision at pres- cally for the Cariceae (which he recognized as the family knt. Subgenus Indocarex is by far the smallest with app;oxi- Kobresiaceae) is considered unnecessary. The homology and mately 100 species. Some people accord subgenus Indocarex evolution of inflorescence structures in Cyperaceae is a only sectional rank within subgenus Carex (Kern and research field where much work needs to be done, and a basic Nooteboom 1979). Definition of subgenus Indocarex is cer- introduction to it can be found in the following articles: tainly made difficult by species such as C. phyllostachys C. A. Holttum (1948), Kern (1958, 1962), Schultze-Motel (1959, Meyer or C. schiedeana Kunze that are apparently members 1964), Koyama (196 1, 197 l), Haines ( 1967), Kukkonen of subgenus Indocarex, but have reduced infflorescences lack- (1967b, 1984, 1986b), Haines and Lye (1972), Meeuse (1975), ing compound inflorescence units. However, in spite of Eiten (1976), Meert and Goetghebeur (1979), and Timonen Raymond's (1959) suggestion that subgenus Indocarex may be (1989). artificial, there seems to be no compelling evidence at present for dispersing it amo-$ diverse sections in subgenus Carex as was done by Koyama (1962). Subgenera in Carex Subgenus Vignea is best developed in North and South Recognition of subgenera in Carex is, of course, linked to America, but with ample representation in the temperate and the establishment of a phylogenetic classification. As such, boreal regions of Eurasia. It is very scarce in the Paleotropics. deciding on recognition of subgenera should more properly Subgenus Carex is widespread throughout the range of the come at the end, rather than the beginning, of a paper such as genus, although rare at lower elevations in the tropics. this. However, the numerous species of Carex do need to be Subgenus ~ndocarexis found primarily in the tropics and-sub- placed in some framework simply for discussion. Recognition tropics of southeast Asia, with a few species elsewhere in the of subgenera at this point should, therefore, be construed as Paleotropics, mostly in East Africa, and perhaps 10 to 15 spe- an hypothesis. Three subgenera are here recognized in Carex: cies in the Neotropics subgenus Carex,' subgenus Indocarex, and subgenus Vignea. The heterogeneity of unispicate Carex (Kreczetovicz 1936; Past work on evolution in Carex Nelmes 1952) leaves no doubt that subgenus Primocarex Kiikenth.. based only on having a single spike as the inflo- Many authors, starting with Linnaeus, arranged species of rescence, is an artificial conglomeration and it is not recog- Carex in various hierarchical classifications to facilitate iden- nized here. However, the subgeneric affinities of some anom- tification (see Kreczetovicz 1935, Egorova 1972, and alous unispicate species with no evident close relatives are Robertson 1979 for summaries). In many cases, these systems nevertheless in doubt (Egorova 1972; Chater 1980). did place closely related species together. However, the first Other subgeneric classifications such as those of Clarke avowed attempts to .develop a natural system in Carex were (1904) and Kreczetovicz ( 1935) are plagued by being, in large by Edward Tuckerman in 1843 in North America and Salomon part, artificial with heavy stress on one or a few characters, Drejer in 1844 in Europe. especially stigma num%er, and are not considered here. Savile As Bailey (1886) noted, Tuckerman's booklet was the first and Calder's (1953) subgenus K~tekenthnlia is also not professed attempt to make a natural classification of Carex with named divisions. He recognized unispicate species as sec- tion Psyllophorae (Degl.) Koch, distigmatic species with 'The Berlin Congress (1987). in accepting Proposals C and F to amend Article 8 of the lntemational Code of Botanical Nomenclature, spikes all similar as sections Vignea Koch and Leptantherae ruled that priorability of lectotypification depends on effectively pub- Tuckerman, species with compounded inflorescences as sec- lished, explicit designation of types (McNeill 1987). Lectotypifica- tion Vigneastra Tuckerman, and tristigmatic species with dis- tions contrary to a Recommendation (as opposed to a Rule), cannot, similar spikes as section Legitimae Koch. Within these five on that account, be rejected. Therefore, the type of Cnre.~must. sections, Tuckerman had a number of subsections and subor- unfortunately. be C. pulicnris L.. as selected by Britton (1907) and dinate groups within which he clustered putatively related followed by Mackenzie (1913) (McNeill et 01. 1987). Most authors species. have defended C. llirtn L. as the type, based on its selection by Green Drejer (1844) did not recognize subgenera or sections, nor (1929) (Hitchcock and Green 1947; Egorova 197 I; Voss 1972; Tucker did he have hierarchical groupings, Rather, he grouped species 1987). Although distigmatic and excluded by Egorova ( 1972), C. he considered to be related into "greges." Drejer discussed at pulicaris is tentatively here placed in what has traditionally been called subgenus Carex because of its putative relationship with tristigmatic length the affinities of species, producing several "tabulae unispicate species such as C. pnrccijlorn Lightf., whose subgeneric affinitatum" showing his view of relationships among species. affinities are more agreed upon. If Pritnocnrex were to be recognized Drejer was apparently the first to consider not only that uni- as a subgenus distinct from subgenus Carex, then the latter epithet spicate species of Carex constituted an artificial group, but would have to supplant Pritnocnrex. also that they were reduced from rnultispicate species, "for- TABLE1. Characters of three subgenera in Carex Character Subgen. Vinnea Subgen. Carex Subgen. lndocarex Stigma number 2, rarely 3 or mixed 2 and 3 Mostly 3, sometimes 2, very 3 rarely 4 Perigynium shape in cross section Usually biconvex or planoconvex Highly variable Trigonous or more or less terete Abaxial false suture Present Absent Absent Cladoprophylls (in multispicate species) Absent Present, very rarely absent Present Inflorescence bracts Mostly not prolonged and leaf- Mostly leaf-like, sheathing or not Mostly leaf-like, sheathing like, never sheathing Inflorescence units Lateral usually similar to terminal Lateral mostly pistillate or Lateral similar to terminal androgynous; terminal mostly staminate; occassionally with. other configurations; rarely all similar Sessile, rarely peduncled Sessile or peduncled Mostly peduncled Single or multiple at nodes Mostly single at nodes, rarely Single or multiple at nodes multiple Compound or not Mostly not compound Mostly compound Inflorescence prophy lls Absent Absent Present and more or less perigynium-like, rarely fertile Spikes Mostly androgynous, sometimes na (see text) Androgynous gynecandrous, rarely unisexual -- mae hebetatae Caricum genuinamm" (Drejer 1844, p. 8). He ently the first to distribute unispicate species, as reduced mem- even suggested the multispicate species to which he considered bers, among other putatively natural, named groups. Bailey specific unispicate Carex to be related. elaborated on this further in his synopsis of North American The middle 1800s saw a great flourish of descriptive activ- Carex (1 886). ity, culminating in Chester Dewey's Caricography (Dewey In the only existing worldwide monograph of Carex, 1866, index) and Francis Boott's (1858, 1860, 1862, 1867) Kukenthal (1909) produced a classification recognizing four monumental Illustrations of the Genus Carex. While Boott and subgenera. His classification is, in large part, still followed at Dewey astutely understood the interrelationships of sedges at present, and because of its importance, his key -to subgenera the species level, the next contribution to a natural system of is quoted here, translated from Latin. classification in Carex was by Bailey (1885), who was appar-

A. Spikelet solitary, terminal ...... subgen. I. Primocarex B. Spikelets several a. Spikelets bisexual, sessile. Cladoprophyll (i.e., the prophyll of the second- or third-order axis) for the most part not developed ...... subgen. 11. Vignea b. Spikelets bisexual or unisexual, often peduncled. Cladoprophyll always developed a. Spikelets always bisexual. Cladoprophyll in all the spikelets perigynium-like (to me a perigynium with a branch) ...... subgen. 111. Indocarex b. Spikelets for the most part of separate sex. Cladoprophyll (at least in the lower spikelets always) ocreaeform ...... subgen. IV. Eucarex

As his subgeneric name suggests, Kukenthal, unlike Drejer Heilborn (1924), in an extensive treatment of chromosome and Bailey, thought that the unispicate species he placed in numbers in Carex, produced the first phylogenetic tree dia- subgenus Primocarex were primitive. Although he evidently gram for Carex. His tree was based on Kukenthal's subgenera thought that subgenus Primocarex was a natural group, he was and sections, arranged to reflect the concept that low chro- certainly aware that some unispicate species were related to mosome numbers were ancestral and high chromosome num- multispicate ones. However, he felt the unispicate species to bers were derived. Heilborn suggested that sections Acrocystis be ancestral. He noted, for example, under the unispicate Dumort. (Morztanae (Fries) Christ) and Paniceae G. Don were C. picta Steudel "species subgeneris Eucarex ab hac derivata primitive since they had low chromosome numbers. He then est C. Baltzellii Chapm. " proposed that evolution in Carex proceeded from small, dry- Agreement by subsequent authors with Kiikenthal's subge- land species to large wetland ones. However, his limited data neric classification has been for the most part universal except for subgenus Primocarex. Authors such as Ohwi (1936) and also forced him to suggest that at least some unispicate species Mackenzie (1931, 1935), although they did not comment in were the result of reduction, since they have relatively high detail, associated unispicate species with multispicate ones, chromosome numbers. sometimes even in the same section, while otherwise following In a seminal paper, Kreczetovicz (1936) strongly argued that Kiikenthal. Ohwi (1936) also did not recognize subgenus subgenus Primocarex was not primitive, but rather a dustbin Dzdocarex. group whose members were the result of repeated, independent 1414 CAN. J. BOT. VOL. 68, 1990 reduction of multispicate species belonging to the other three rachilla-bearing unispicate Carex was to consider them to be subgenera. He hypothesized that this reduction to a unispicate reduced from other genera in the tribe Cariceae rather than condition occurred in conjunction with migration of Carex from subgenus Indocarex. He succinctly said, "the Carex ori- from the tropics into boreal, arctic, and alpine habitats. gin of many of the Primocarices may be questioned. . ." and Kreczetovicz recognized two types of reduction from a mul- without much evidence, went on to list species he felt were tispicate to a unispicate condition. He defined digressive derived from Kobresia, Schoenoxiphium, and Uncinia. The reduction as the loss of lateral spikes to produce a unispicate strongest case Nelmes presented was for the origin of inflorescence and cited C. exilis Dewey and C. ursirza Dewey C. microglochin and its relatives from species of Uncinia. Like as examples. These species occasionally produce accessory Kreczetovicz, Nelmes pointed to the small, weak-hooked 'spikes' and are closely related to multispicate species. U. kingii as an intermediate. However, Nelmes was forced to Kreczetovicz defined transmutive reduction as the reduction of weaken even his strongest argument by noting that C. pauci- members of subgenus Indocarex to a unispicate condition by flora, a species he considered very closely related to reduction of the lateral inflorescence units to a vestigial axis C. microglochin, must also be descended from Uncinia even and the concomitant conversion of the cladoprophylls into per- though its rachilla "has entirely disappeared. " Although Carex igynia. Kreczetovicz termed the resulting inflorescences pseu- is primarily a north temperate genus, the puzzling fact that domonostachyous, a condition where the terminal staminate nearly half of the 60-70 unispicate species are tropical or portion of the inflorescence is derived from the staminate por- Southern Hemisphere probably contributed much to Nelmes' tion of the terminal spike, and the perigynia with their scales beliefs, since Uncinia and Schoenoxiphium are also primarily are each derived from an inflorescence unit reduced to one tropical or Southern Hemisphere. perigynium. Kreczetovicz cited species such as C. willdenowii In addition to his ideas on the phylogeny of unispicate Willd. as demonstrating this type of reduction, noting the leafy, Carex, Nelmes (195 1, 1952, 1955) also believed that the bract-like pistillate scales and the peculiar, stalked staminate broad-leaved Paleotropical species of subgenus Indocarex were portion of the inflorescence as clear evidence of this primitive. However, he presented no evidence in support of transformation. this idea. .-- Since he considered unispicate species derived rather than Nelmes' proposals, if correct, would not only have shattered primitive, Kreczetovicz was the first author for whom the fact the integrity of the genus Carex, but would also have made that many unispicate Carex have rachillas could present an distinguishing genera in the tribe Cariceae impossible. For- evolutionary paradox: if unispicate Carex are reduced from tunately, he did not formalize his schemes nomenclaturally. multispicate species, which allegedly no longer have rachillas, Savile and Calder (1953), using smut infection data as a then unispicate species should also not have rachillas, assum- biochemical tool for establishing relationships, mustered evi- ing they did not reevolve them. If rachillas are present in unis- dence that Carex was "a reasonably natural genus" and not picate species, then presumably these unispicate species should composed of such diverse elements as Nelmes (1952) sug- be primitive, as Kiikenthal (1909) believed. Kreczetovicz dealt gested. They outlined, in some detail, hypothesized lines of with this paradox by hypothesizing that those unispicate spe- evolution in Carex, which they considered directly descended cies with rachillas were produced by transmutive reduction from within Kobresia, and produced an evolutionary tree dia- from species of subgenus Indocarex, and their rachillas are a gram somewhat similar to Heilborn's ( 1924). A new subgenus, vestige of a Carex spike axis rather than a vestige of a Kobre- Kuekenthalia, was described that consisted of those members sia-type rachilla (spikelet axis). of subgenus Carex with more or less inflated perigynia and Kreczetovicz (1936) was also the first to clearly hypothesize persistent styles. Unfortunately, this subgenus included the that some species treated as unispicate Carex have their origins type of the genus Carex, C. pulicaris. It should therefore have outside the genus. He postulated the direct origin of C. microg- been called Carex subgenus Carex and the name Carex lochin Wahlenb. and its allies from reduced species of Unci- subgenus Kuekenthalia was thus not validly published. Savile nia, citing U. kingii Boott in J. D. Hook. as a reduced Uncinia and Calder's subgenus was, in fact, recognized as a natural and even suggested that it was closer to Carex than to the section long ago by Bailey (1886) and was also recognized as remainder of ~ncinia. a "natural phylum" by Heilborn ( 1924). Even though tlie ;achilla paradox was not completely Like Kreczetovicz (1936) and Nelmes (1952), Savile and resolved, Kreczetovicz's paper marked a turning point. Few Calder placed great emphasis on the presence or absence of a people since have considered unispicate species to be primi- rachilla within the perigynium, noting that "characters such tive, and authors who continued to recognize subgenus Pri- as the possession of a rachilla . . . seem to be of fundamental mocarex did so largely for convenience (Chater 1980; Jermy importance." However, rather than consider unispicate Carex et al. 1982). The only exceptions have been some workers who with rachillas derived from other genera or other Carex, Savile studied evolution in Carex from the viewpoint of smut infec- and Calder considered them to be primitive. They retained tion data, e.g., Savile and Calder (1953). them in subgenus Primocarex, which they considered a natural Kreczetovicz's ideas were greatly elaborated by Nelmes subgenus after transferring those unispicate species without (1952), who developed the concept that subgenus Primocarex rachillas into other subgenera. They dealt with the rachilla par- was artificial to its zenith. Nelmes placed great stress on the adox by postulating extinct multispicate and rachilla-bearing evolutionary significance of the presence of a rachilla only in species of subgenus Primocarex as ancestral to the unispicate, certain unispicate species. He noted, "The whole of its [Carex] rachilla-bearing species of subgenus Primocarex from which 2500 worldwide species, if much of Kiikenthal's subgenus Pri- most of the remainder of Carex evolved. Subgenus Indocarex mocarex he excluded, have a completely closed utricle [peri- was hardly mentioned, but was evidently regarded as a pri- gynium] and an invariably absent rachilla." Nelmes consid- mitive, early offshoot of the line leading from Kobresia to ered unispicate Carex lacking rachillas to be indeed reduced subgenus Primocarex. Within subgenus Garex, they consid- from multispicate Carex. His hypothesis for the origin of the ered the unispicate section Obtusatae (Tuckerman) Mackenzie and section Acrocystis (Montanae) primitive, and section have arisen from either subgenus Indocarex or "advanced Phacocystis Dumort. (Acutae (Fries) Christ) the most derived. Kobresia or Schoerzoxiphium." Carex subgenus Vignea and Within subgenus Vignea, C. sabulosa Kunth (C. leiophylla subgenus Carex were both regarded as independently derived Mackenzie) was considered the most primitive, with the unis- by reduction from subgenus Indocarex. They viewed subgenus picate sections Ursinae Kiikenth and Physoglochin Dumort. Prirnocarex as reduced and polyphyletic, with the rachilla (Dioicae (Tuckerman) Pax) the most derived. In their subgenus being a reduced inflorescence axis, as suggested previously by Kuekenthalia, the unispicate C. pauc@ora and its allies were Kreczetovicz (1936) and Smith (1966). considered primitive and section Vesicariae (0. Lang) Christ All of these treatments commenting on evolution in Carex advanced. presented only incomplete sketches lacking satisfactory con- Several other features of Savile and Calder's (1953) system clusions supported by diverse evidence. As noted by Raymond were noteworthy. Where Nelmes (1952) hypothesized that (1959), Kreczetovicz's (1936) ideas were formulated with lit- Carex rnicroglochin and its allies were reduced from Uncinia, tle knowledge of tropical species. Nelmes (1951, 1952) knew Savile and Calder suggested essentially the reverse, that tropical Carex, but his hypotheses were colored by precon- C. microglochin and allies were an offshoot of the evolution- ceived notions about primitive groups in Carex and by his ary line leading to Uncinia and that this evolutionary line orig- treatment of speculation as fact (Kern 1958). Savile and Calder inated from within Carex. Based on their concept of relation- (1953) again approached the problem from the narrow per- ships of smut fungi, they considered their subgenus spective of primarily boreal species. As well, some of the smut Kuekenthalia to be more closely related to subgenus Vignea phylogeny they used has undergone revision since they pub- than to the remainder of Carex. lished their paper (Savile 1979). Much past work also empha- Koyama (1962) followed Ohwi (1936) in formally recog- sized anomalous features of a few species, or even one, rather nizing only two subgenera in Carex, subgenus Carex (includ- than considering the genus as a whole. Examples include ing subgenus Indocarex), which he considered to be primitive, Kreczetovicz's and Nelmes' concentration on a few unispicate and the derived subgenus Vignea. Koyama considered section Carex such as C, rnicroglochin and on Uncinia kingii, and Glareosae G. Don (Canescentes (Fries) Christ, Heleonastes Savile and Calder'sxmphasis on C. sabulosa (C. leiophylla). (Kunth) Mackenzie) in subgenus Vignea and section Atratae The detailed and thorough discussions of Smith and Faulkner (Fries) Christ in subgenus Carex to have "a close affinity." (1976) offered the best treatment of the subject of evolution Koyama (1957) earlier noted that the broad-leaved sections in Carex and they generated a numer of specific hypotheses. Hemiscaposae C. B. Clarke and Mapaniifoliae Nelmes & Airy However, their work was framed in the context of a review of Shaw "have more primitive characters than any other Indo- the inflorescence of Carex, and it too dealt primarily with tem- carices," although he did not specify what these were. He also perate species. suggested that C. tricephala Boeckeler was a reduced Indo- In spite of these problems, there has been, except for work carex and that it showed "where we may look for a solution based on smuts, broad agreement for the last half-century on to the evolution of Indocarices into Eucarices." several overall trends of evolution in Carex. Briefly summa- Kukkonen (1963), briefly commenting on evolution within rized, Carex is derived from Kobresia- or Schoenoxiphium- the tribe Cariceae in a treatment of the smut genus Anthra- like ancestors that did not have a sharp distinction between coidea Bref., generally supported Savile and Calder's (1953) spikes and spikelets in their branched inflorescences. The ideas. He suggested two possible evolutionary lines, one lead- subgenus of Carex in which are found the most primitive fea- ing from Kobresia through species in subgenus Primocarex tures is Indocarex, with normally compounded inflorescences such as sections Filifoliae (Tuckerman) Mackenzie and and unique, perigynium-like and occasionally fertile prophylls Nardinae Kiikenth. to subgenus Vignea, and another directly in the inflorescence. From subgenus Indocarex, subgenus connecting Kobresia to section Acrocystis (Montanae) in Carex could be derived by reduction in branching and subgenus subgenus Carex. Vignea by reduction in branching and stigma number, and Nannfeldt (1977), working primarily with Anthracoidea suppression of peduncles and cladoprophylls. Subgenus Pri- smuts on subgenus Vignea, briefly reviewed the systematics mocarex is polyphyletic and derived by reduction from the of Carex as a whole. He agreed that subgenus Indocarex was other three subgenera or possibly even other genera of the tribe primitive and probably a natural group, although smuts on Cariceae. The rachilla of some unispicate species may be a C. distachya Desf., presumed to be a reduced Indocarex, sug- reduced inflorescence axis. gested that it was not. He considered subgenus Vignea a mon- This broad agreement is, in large part, a reflection of there ophyletic and derived group "whose connections with the rest being establishable polarities for at least a few inflorescence of the genus are obscure." He also agreed that Savile and characters. Compound inflorescence units are primitive, as are Calder's (1953) subgenus Kuekenthalia was, in part, a natural uniform and numerous androgynous spikes, lateral infloresc- group, but that it was not deserving of subgeneric rank and ence units similar to the terminal (Kukkonen 1984), and that the "lower" species in Savile and Calder's treatment were peduncled inflorescence units. Regardless of whether one unrelated to the remainder. Nannfeldt considered subgenus selected the genus Schoenoxiphium or one of the most fre- Primocarex to be a "very heterogenous assemblage of species quently suggested related tribes, Hypolytreae J. S. & C. Presl having only one common character, viz. the single uni- or bi- ex Fenzl, Rhynchosporeae Nees ex Fenzl, or Sclerieae Kunth sexual terminal spike. " ex Fenzl, as the outgroup for Carex (Koyama 1961; Haines Smith and Faulkner (1976), in a thorough and detailed and Lye 1972; Smith and Faulkner 1976), the polarities of review of the inflorescence of Carex, presented a clear treat- these inflorescence features remain the same. However, if the ment of past hypotheses of inflorescence evolution in Carex. Cariceae indeed "sprang from very close to Trichophorum" They considered that subgenus Indocarex had the least spe- as stated by Savile (1979) and alluded to by Kukkonen and cialized inflorescence morphology and could be derived from Timonen (1979) based on smut data, a rethinking of all the "primitive Kobresia-Schoenoxiphium stock. " Uncinia could above concepts would be in order. Most notably, the latter case 1416 CAN. J. BOT. VOL. 68. 1990 would require that the unisexual flowers of the tribe Cariceae The rachilla in Carex be derived from immediate bisexual ancestors. There is considerable literature about the occurrence of rach- illas in multispicate as well as unispicate Carex that was gen- Remaining evolutionary problems in Carex erally not cited by previous workers considering evolution in Carex. Presumably, this is because they assumed rachillas in Although the rachilla paradox has not been totally resolved, multispicate Carex to be merely teratological. Snell (1936) the concepts summarized in the preceding section otherwise briefly reviewed previous work reporting rachillas in Carex. make good sense on superficial examination and have rarely He noted that abnormal development of rachillas to bear fur- been questioned. Nevertheless, on close examination, diffi- ther flowers is common in Carex, but also that "nearly every culties in addition to the rachilla paradox come into focus, as species of Carex shows some remnant of the spikelet axis enumerated below. [rachilla] within the perigynium at flowering time and even (1) The most compounded inflorescences in Carex occur not later. " This observation is devastating to evolutionary schemes in subgenus Indocarex, but in section Fecurzdae Kukenth. of such as those of Nelmes (1952) and Savile and Calder (1953), subgenus Carex and in species of several sections in subgenus which were built on the foundation that rachillas occurred only Vignea, for example, C. canariensis Kukenth., C. crus-corvi in certain unispicate Carex. Snell's statement was supported Kunze, and C. decomposita Muhl. by Barnard (1957), Schultze-Motel (1959), Le Cohu (1968), (2) Cladoprophylls are absent from some highly branched and Svenson (1972) and is, in fact, readily demonstrable. species in subgenus Carex, for example, C. fecunda Steudel The largest rachillas in Carex are often found in unispicate and its relatives. species such as C. obtusata Lilj. (Figs. 2 to 4), which is reg- (3) Subgenus Carex usually has highly developed sexual ularly reported (e.g., Wheeler and Ownbey 1984) as being one specialization within its inflorescences, most commonly with of the few species whose rachilla has a vestigial scale at the the terminal inflorescence unit staminate and the lateral inflo- apex. The degree of development of this scale is, however, rescence units pistillate or androgynous. This is in contrast to variable, as shown inFig. 2 and 4, and other species such as the more or less uniformly androgynous spikes and similar C. filifolia Nutt. (Fig. 5) also often have a vestigial scale at terminal and lateral inflorescence units of both the putatively the apex. These large rachillas are more or les flattened, sca- primitive subgenus Irldocarex and the supposedly derived brous-margined, and two-veined. subgenus Vignea. As noted by Snell(1936) and Svenson (1972), many species (4) If the abaxial false suture of the perigynia of subgenus of multispicate Carex in all three subgenera also normally have Vignea is a vestige of a formerly open perigynium, this would rachillas. Figures 6 and 7 show two rachillas from the same be an anomaly for a derived subgenus supposedly descended inflorescence of C. backii Boott, one with two veins and from ancestors with a fully closed perigynium. There was another smaller and with only one faint vein. In many species, essentially no discussion of this potentially significant feature the rachillas are merely tiny nubs without vascular tissue. in the recent literature, except for the brief mention of it by Figures 8 and 9 show the rachillas of C. capillaris L. and Nannfeldt (1977). However, it is in the anatomically correct C. debilis Michaux, respectively. Figures 2-9 are from position to be a vestige of an open perigynium. subgenus Carex. Figures 10 and 11 show rachillas of C. exilis (5) The perigynium-like inflorescence prophylls in the com- and C. stipata Willd. in subgenus Vignea, respectively. Rach- pounded inflorescences of subgenus Indocarex are unique in illas of C. schiedeana and C. satsumensis Franch. & Savat. Carex and appear to have been derived from fully formed per- in subgenus Irldocarex are illustrated in Figs. 12 and 13. igynia. This suggests that the perigynium may have evolved Rachillas of varying size are characteristic of many species to its present highly developed state before additional branch- throughout Carex. These rachillas are simply a function of the ing occurred, contrary to the hypothesis that subgenus Indo- fact that the perigynium-achene unit is really a condensed axis carex is descended from primitive, highly branched Kobresia- with an axillary female flower. The presence of a rachilla in or Schoenoxiphizim-like ancestors, which presumably would unispicate species is neither unique nor of fundamental phy- not have had a fully developed Carex-like perigynium. logenetic importance. It does not indicate that those species In addition to highlighting gaps in our knowledge, all these are ancestral. Elaborate schemes such as those of Nelmes above points also suggest the need for a closer examination of (1952) or hypothetical taxa such as those proposed by Savile the relationships of the subgenera of Carex and a careful study and Calder (1953) are not necessary to explain rachillas in of homology of inflorescence units and spikes among the unispicate Carex. subgenera. Although its phylogenetic importance is minimized by the While some of the points raised by previous work and this information discussed above, the rachilla paradox still exists brief review cannot be clarified without a great deal more basic in a modified form. Some of the large and presumably least data, a few aspects that convey considerable phylogenetic reduced rachillas do occur in unispicate and presumably highly information can be resolved further. The two aspects most cru- reduced Carex. The simplest hypothesis accounting for this cial to discussion of evolution in Care,x are the distribution fact may be that many unispicate species are relictual and orig- within Carex of the rachilla and the morphology of the rachilla. inated from early, independent reductions to a unispicate con- and inflorescence. Insofar as it is possible, these two topics dition of primitive multispicate ancestors with large rachillas. are pursued further here. As well, ever since Kreczetovicz's This hypothesis is similar to that proposed by Savile and Calder (1936) paper, C. microglochin, with its exserted rachilla, and (1953) but cast in a quite different evolutionary framework, U. kingii, with its weak hooks, were frequently cited as dem- with unispicate species not considered ancestral to the remain- onstrating a clear evolutionary link between Carex and Unci- der of Carex. While none of the preceding discussions abso- nia. This important example, around which much speculation lutely preclude the possibility that occasional unispicate spe- has revolved, is also closely examined. cies of Carex are descended from Kobresia or the possibility FIGS. 2-5. Rachillas of Carex. Fig. 2. C. obtusata with large terminal scale. Fig. 3. C. obtusata rachilla base showing two veins. Fig. 4. C. obtusata variation in terminal scale size (Pennington Co., South Dakota, Reznicek 7280 MICH). Fig. 5. C. filifolia rachilla with large terminal scale (Custer Co., Montana, Calder et al. 20565, MICH). Bars = 0.2 rnrn.

that some large rachillas may indeed be vestigial Carex spike for a number of reasons. Rachillas can be of direct taxonomic axes, as discussed in more detail by Smith and Faulkner significance. Kukenthal(1909) used rachilla characters in some (1976), there is no clear evidence to support adoption of these sections, e.g., section Junciforrnes (Boeckeler) Kukenth. Lim- more complex hypotheses. ited anatomical evidence can also be gleaned from rachilla Since they were thought by systematists not to exist in most positions. Since the rachilla in C. debilis diverges at the base Carex, the occurrence and morphological diversity of rachillas of the stalk of the achene (Fig. 9), the stalk is probably com- are poorly known. Further study of rachillas is highly desirable posed entirely of achene tissue. The stalk of the achene (more CAN. J. BOT. VOL. 68. 1990

bud scale

rachill

rachilla

bud

14 -5 REZNICEK 1419 or less fused with spongy tissue of the perigynium base) in U. kingii, in fact, is not homologous with hooks in all other C. pauc@ora, on the other hand, is presumably axis tissue, Uncinia species, but has evolved independently. As noted by since the tiny rachilla diverges at the apex of the stalk Kukkonen (1967~)and visible in Fig. 24, the hook of U. kingii (Fig. 14). is often quite irregular and twisted and sometimes hardly hook- like. Kukkonen (1967~)also noted that U. kingii is the only species of Uncinia with a single vascular strand in the rachilla. The relationships of Carex microglochin and allies in The rachilla of U. kingii is also terete, smooth, and with a light of rachilla morphology thick layer of spongy parenchyma surrounding the central vas- More important than the direct taxonomic utility of rachillas cular strand (Figs. 22 and 23). The rachilla of U. kingii is not only strikingly different from the rest of uncinia, it is, except Or the gleaned from them, is the phy- for its somewhat larger size and ilregular hook, essentially logenetic information obtainable from comparative studies of identical in structure to that of Carex microglochin, as rachillas. While many more species need to be studied before of Figs. 17 and 23 will show. detailed conclusions can be drawn, there are generalizations iff^^^^^^^ between U. kingii and the remainder of Uncinia that can alrehdy be made concerning rachillas. Most of the are not confined to the rachilla. Uncinja species are usually largest rachillas in Carex are, with the exception of C. micro- large (>2 dm) plants with ovoid to elongate-cylindric, many- glochin and its allies, two-veined, more or less flattened, and flowered inflorescences whereas U. kingii is a small (<] dm) often scabrous-margined. Some even have a reduced scale sub- plant with a capitate, few-flowered inflorescence (~i~.24). tending a tiny bud (Figs. 2 and 5). These character states tor- he inflorescence of U. kingii is similar in structure to that of respond to those of smaller rachillas in Kobresia (Fig. 15) and Carex microglochin, especially before the perigynia of the lat- also Schoenoxiphium, although well-developed rachillas in the ter fully reflex at maturity, as shown in Fig. 25. Uncinia kingii latter are typically more than two-veined. In addition, even the also differs in leaf and culm anatomy from the remainder of highly specialized, hooked rachillas of Uncinia are two-veined Uncinia. It has terete culms and leaves without a keel whereas (Kukkonen 1967~)and may be slightly flattened, although most other Uncinia secies have obtusely trigonous to trique- smooth. The unispicate, monotypic Cymophyllus also often trous culms (except U. erinacea (Cav.) Pers.) and all have possesses a rachilla (Tucker 19871, but it is mostly small and keeled leaves (Kukkonen 1967~).In these anatomical features, without a vascular strand. Based on comparison with the rest U. kingii again closely resembles C. microglochin and its of the tribe Cariceae, large, two-veined, more or less flattened allies, which also have terete culms and leaves without a keel rachillas are primitive in Carex. (Kukkonen 1970). As well, in Uncinia, the achene, the rach- A significant element of the hypotheses of both Nelmes illa, and one nerve of the perigynium are rotated, with respect (1952) and Savile and Calder (1953) was the relationship of to their axis, 60" from the position shown in Carex, except for Carex microglochin and allies to Uncinia, especially U. kingii U. kingii, which is apparently oriented as in Carex (Meert and from temperate South America. Figures 16 and 17 show the Goetghebeur 1979). rachilla of C. mi~ro~lochin.~It is clearly in striking contrast There can be no doubt that U. kingii does not share an to the other Carex rachillas illustrated in Figs. 2-14. It is very immediate common ancestry with the rest of Uncinia. It is not large, but nevertheless only one-veined. It is terete and smooth. a species of remarkable ph~logeneticinterest that together with Surrounding the single central vascular strand is a thick layer Carex microglochin, bridges the gap between Carex and Unci- of spongy parenchyma, making the rachilla a relatively soft, nia. Uncinia kingii is simply a Carex very closely related to more or less easily bent structure. C. microglochin and its allies, but with the rachilla more dif- Uncinia by contrast has, in addition to the distinctive hook ferentiated and somewhat larger, reflecting a hooked dispersal formed entirely by an inrolled, reflexed scale, as illustrated by mechanism as opposed to the putative ''clinging" dispersal the New Zealand U. divaricata Boott (Fig. 18), a stiff, slightly mechanism found in C. microglochin and presumably flattened, two-veined rachilla with considerable sclerenchyma C. camptoglochin ". Kreczt and C. parva Nees (Fernald 1926; (Fig, 19). The morphology of the hook and rachilla in Uncinia Savile 1972) or the' ''catapult" mechanism of C. pauclflora is uniformly as described above (~~kk~~~~1967c) with one (Hutton 1976). It has nothing in common with Uncinia except exception. The exception is the species touted as demonstrat- the su~erficiall~ but ing the evolutionary connection between Carex and Uncinia, reflect this relationship, U. kingii is here transferred into U. kingii. Carex. This transfer does not affect in any way the definition The hook of I/. kingii is strikingly different from all other the genus as given On page 1410, since the unique species of Uncinia. It is formed not by a reflexed scale, but character of Uncinia is the rachilla tipped by a retrorse, by a bend in the rachilla (Figs. 20 and 21). The hook of Carex kingii (Boott) Reznicek, comb. nov. 3Note in ~i~,16 the tiny, hyaline scale subtending the slightly Uncinia kingii Boott in J.D. Hook., F1. Antarct. 2: 370. stalked ovary. This rare abnormality in Carex is evidently not a pro- 1846 phyll, since it faces the main axis, but is probably a lateral scale The removal of Carex kingii makes Uncinia a much more axillating the branch bearing the ovary. uniform and presumably natural genus. With the number of

FIGS.6-15. Rachillas of Carex and Kobresia. Figs. 6 and 7. C. backii (Pennington Co., South Dakota, Reznicek 7303, MICH). Fig. 8. C. cupillaris (Victoria Co., New Brunswick, Reznicek5043, MICH). Fig. 9. C. debilis (Blount Co., Tennessee, Camp 1877, MICH). Fig. 10. C. exilis (Atlantic Co., New Jersey, Hermann 3055a, MICH). Fig. l I. C. stipata (Benton Co., Washington, Smith in 1929, MICH). Fig. 12. C. schiedeana (San Luis Potosi, Mexico, Anderson 13313, MICH). Fig. 13. C. satsumensis (cultivated, Hermann garden No. 15541 in 1935, MICH). Fig. 14. C. pauciflora (Thunder Bay Dist., Ontario, Garton 18392, MICH). Fig. 15. K. rnyosuroides (Fjaeldmark, Greenland, Erlan- son 2562, MICH). Bars = 0.2 mm. CAN. J. BOT. VOL. 68. 1990

rachi

scale differences now known between Carex and Uncinia, enum- inflorescence of this species may have as many as 2500 andro- erated in part above, the origin of Uncinia from within Carex gynous spikes and up to about 10 000 to 12 000 perigynia. as suggested by Savile and Calder (1953) or the origin of some Each spike is identical in basic structure to the spikes in the unispicate Carex from Uncinia as suggested by Nelmes (1952) smaller inflorescences of species like C. sparganioides Willd. is improbable. Although Uncinia and some Carex both have as shown in Fig. I. The inflorescence illustrated in Fig. 30 is rachillas, the rachillas are different in structure. In fact, the noteworthy in one other respect. It has peduncled inflorescence very close relationship of Carex and Uncinia assumed by most units, a character state rare in subgenus Vignea. previous workers needs careful examination, as also suggested By loss of peduncles and reduction in branch size and num- by Timonen (1985). The two genera are certainly not conge- ber, it is conceptually simple to derive inflorescences similar neric, as stated by Koyama (1961). in structure to most smaller species of subgenus Vignea from The repeated, independent evolution of a dispersal mecha- inflorescences of the type illustrated in Fig. 30 and 3 1. Some nism as efficient as a hook is certainly plausible. In fact, this highly branched members of subgenus Vignea have peduncles mechanism has evolved at least three times in the tribe Cari- that while still clearly evident, are fused with the inflorescence ceae. Carex collinsii Nutt. from eastern North America also axis for some distance above the node. This suggests the addi- possesses this dispersal adaptation (Fig. 26), but utilizing per- tional possibility that the sessile inflorescence units character- igynium beak teeth rather than rachilla apices. istic of almost all of subgenus Vignea may also have origi- The unusual structure of the rachillas in C. microglochin nated, at least in part, through fusion of peduncles to the main and its allies, terete and smooth with a single central vascular axis. The derivation, however, of the rather uniform group- - of strand and ample parenchyma contrasted with large size, is not sections with exclusively gynecandrous spikes (section Ovales consistent with these rachillas being primitive. Also, since (Kunth) Christ and related sections Glareosae (Canescentes, structures similar to the rachillas of C. microglochin and allies Heleonastes), Deweyanae Mackenzie, Stellulatae (Kunth) occur nowhere else in the tribe Cariceae, it seems unlikely Christ, etc.), is unclear. that they represent mere stages in the process of reduction of The precise homology of inflorescences of subgenus Carex caricoid rachillas. Rather, it suggests that these rachillas have and subgenus Vignezpresents a difficult puzzle since the two either secondarily reevolved after their loss or that C. micro- subgenera have such different inflorescences, as outlined in glochin and its allies have an origin completely different than Table 1. The suggestions of previous authors for the origin of the remainder of the tribe Cariceae. There are no obvious pos- subgenus Vignea from ancestors resembling either subgenus sibilities for the latter suggestion (however, see the discussion lndocarex or subgenus Carex by loss of cladoprophylls and in Kukkonen and Timonen 1979), but the former is here pre- reduction in stigma number do not address the issue of how sented as a hypothesis for the origin of the unusual rachillas one might bridge the substantial discontinuity composed of of C. camptoglochin, C. kingii, C. microglochin, and many characters (Table 1). C. parva. The following hypothesis, derived from studies of highly In some Carex, e.g., C. schweinitzii Schw., the rachilla is branched inflorescences in subgenus Carex, will hopefully cast usually a tiny, unvascularized knob (Fig. 27). However, even the problem of homology of inflorescences within a testable within the same inflorescence, the rachilla is sometimes slen- framework. Figure 32 shows a portion of an inflorescence der and elongate (Fig. 28) and occasionally very long and even branch of the Andean C. fecunda, in section Fecundae. This exceeding the perigynium in length (Fig. 29), although usually species and its closest relatives have some of the most com- coiled within the perigynium. The occasional elongate rach- pounded inflorescences in Carex. In some ways, their inflo- illas produced by C. sckweinitzii are slender, parenchymatous rescence structure is intermediate between subgenus Carex and structures quite unlike other large rachillas in Carex, and are subgenus Vignea. The inflorescence branches have no clado- probably secondarily proliferated after having been lost or prophylls, the spikes are all androgynous, and the female flow- highly reduced. Since this proliferation may be under simple ers are distigmatic, as in subgenus Vignea. However, both the hormonal control (Smith 1967, 1969), it may be quite feasible first-order infloreskence branches and the second-order for the rachilla to be reevolved and subsequently fixed genet- branches are peduncled, and the ultimate portions of the inflo- ically. As a final note, C. pauciflora, often cited as a close rescences have long ranks of perigynia or staminate flowers, relative of C. microglochin but lacking a rachilla (Nelmes as in subgenus Carex. 1952), does not lack a rachilla. it clearly has, as demonstrated Closer examination of inflorescence branches of Carex previously, a small, unvascularized vestige (Fig. 14). fecunda shows that the terminal portions of inflorescence branches are essentially identical in morphology to the unbranched inflorescence units typical of subgenus Carex. Inflorescence structure in Carex Moving farther back from the apex of an inflorescence branch, One of the difficulties noted with existing hypotheses of the ranks of single perigynia grade into ranks of increasingly evolution in Carex is that many species of the supposedly larger sessile androgynous spikes similar to the structure of a derived subgenus Vignea have highly compounded infloresc- compound inflorescence unit of subgenus Vignea, as shown in ences. In fact, the most complex inflorescences in Carex occur Fig. 31. Still farther back from the apex, peduncled branches in subgenus Vignea, as demonstrated by the undescribed dis- occur that repeat the above-described structure. In C.fecunda, tigmatic Andean species illustrated in Figs. 30 and 31. A large the lateral inflorescence units are similar to the terminal and

FIGS. 16-19. Rachillas of Carex and Uncinia. Fig. 16. Base of rachilla of C. microglochin (Banff National Park, Alberta, Hermann 12861, MICH). Fig. 17. Middle portion of rachilla of C. microglochin (same as Fig. 16). Fig. 18. Hook at apex of rachilla of-U. divaricata (South Island, New Zealand, Detwyler NZ F58, MICH). Fig. 19. Lower portion of rachilla of U. divaricata (same as Fig. 18). Bars = 0.2 mm. -. 1422 CAN. J. BOT. VOL. 68, 1990

FIGS. 20-23. Rachillas of Uncinia kingii. Figs. 20 and 21. Hooks at apices of rachillas. Fig. 22. Base of rachilla. Fig. 23. Middle portion of rachilla (Tierra del Fuego, Argentina, Goodall 1393, MICH). Bars = 0.2 rnrn. REZNICEK 1423

FIGS. 24 and 25. Inflorescences of Carex and Uncinia. Fig. 24. Inflorescence of U. kingii (Tierra del Fuego, Argentina, Goodall 3497, MICH). Fig. 25. Inflorescence of C. microglochirl (Banff National Park, Alberta, Hermann 12861, MICH). Bars = 2 cm.

there is no sharp differentiation between inflorescence units and single spikes. Some close relatives of C. fecunda are sim- ilar to it in inflorescence morphology, but with longer ranks of perigynia in the terminal portions of the inflorescence branches, as in Fig. 33, an undescribed Andean species. In yet other species in section Fecundae, such as C. chordalis Liebm. from Mexico (Fig. 34), the inflorescence units are similar to those of subgenus Carex, e.g., section Phacocystis Dumort., although still all androgynous. From the infloresc- ence of C. chordalis, it is conceptually simple, by reduction in number of inflorescence units at a node to one and devel- opment of sexual specialization by conversion of female flower nodes to male flower nodes, to achieve an inflorescence typical of the majority of subgenus Carex, with a terminal staminate inflorescence unit And unbranched lateral pistillate or andro- gynous inflorescence units borne singly at the nodes. Unusual terminal spikes that are pistillate in the middle but staminate proximally and distally, which are characteristic of some trop- ical species (Reznicek 1986), may represent intermediate mor- phological stages in the evolution of staminate or even gyne- candrous spikes from androgynous ones. Concisely stated, the hypothesis proposed here is that the primitive type of inflorescence in Carex is exemplified by the highly compound inflorescence of subgenus Vignea shown in Fig. 30. Inflorescence units of subgenus Carex, with long ranks of perigynia, are derived from an entire inflorescence unit similar to that shown in Fig. 31 by reduction of long ranks of androgynous spikes along an axis to single perigynia. This reduction could occur either by reducing the many-flowered spikes to one-flowered spikes or by conversion of each node from an inflorescence node to a female flower node (Smith and Faulkner 1976). Thus, a single inflorescence unit of subgenus Carex (Fig. 1) would not be homologous with a sin- FIG.26. Hooks formed by beak teeth of Carex collinsii (Prince gle spike of subgenus Vignea, but rather-.with an entire, com- Georges Co., Maryland, Herrnatztz 14413, MICH). Bar = 0.2 mm. pound inflorescence unit. This is not to suggest that subgenus CAN. J. BOT. VOL. 68, 1990

FIGS.27-29. Rachillas of Carex schweinitzii. Fig. 27. Small, knob-like rachilla. Fig. 28. Elongate rachilla extending beyond basal knob. Fig. 29. Elongate and coiled rachilla extending beyond basal knob (Simcoe Co., Ontario, Reznicek 6362, MICH). Bars = 0.2 mm.

Carex originated from extant members of subgenus Vignea, or doprophylls in subgenus Carex, e.g., Le Cohu (1968), sessile even from species that, if extant, would be referable to perigynium-like cladoprophylls probably originated by devel- subgenus Vigtzea. Rather, the hypothesis is strictly one of opment of a perigynium at the node combined with prolifer- homology. ation of the rachilla, rather than development of an infloresc- The inflorescence branching of the very few members of ence unit peduncle. Peduncles with scalelike, nonbasal subgenus Vignea with peduncled inflorescence units (e.g., cladoprophylls are more interesting. In fact, in many, if not Fig. 30) is worth noting. Most peduncles have no evidence of most, species of subgenus Carex (and subgenus Indocarex), a cladoprophyll. On some, however, a partly sheathing scale cladoprophylls are not basal on the peduncle (Fig. 1). In many is present above the base. Also, some peduncles that are partly species of section Fecundae, as well as in many other species fused with inflorescence axes have a scale, oriented with its in subgenus Carex, inflorescence units are loosely flowered at back to the main axis, at the point where the peduncle diverges. the base, often with widely scattered empty pistillate scales. Empty scales at the bases of lateral inflorescence units have These empty scales appear to grade, in some species of section been documented in temperate members of subgenus Vignea Fecundae in particular, into cladoprophylls, and I hypothesize as well (Alexeev 1978; Kukkonen 19860). Whether or not that cladoprophylls in subgenus Carex originated from empty these scales are homologous with the cladoprophylls of pistillate scales at the bases of inflorescence units. As well, in subgenus Car-ex is not certain. If they are, then perhaps species in which the pistillate scales are radically different in subgenus care,^ and subgenus Vigrzea diverged from common color from the perigynia, such as the beautiful C. macrochaeta ancestors before cladoprophylls were lost in the line leading C. A. Meyer from the Pacific Coast of Alaska, British to subgenus Vignea, and cladoprophylls are homologous with Columbia, and northeastern Asia, with black pistillate scales prophylls at the bases of inflorescence branches in other but pale green perigynia, the cladoprophylls are black, iden- Cyperaceae. If they are not, then both these rare scales in tical to the pistillate scales. subgenus Vignea and cladoprophylls in subgenus Car-ex may This hypothesis is contrary to the generally accepted concept have an origin different from the prophylls in inflorescences that cladoprophylls and perigynia are homologous (Holm 1896; of other Cyperaceae. The fact that some members of subgenus Smith and Faulkner 1976) and also suggests that absence of Care-x do not have cladoprophylls may suggest that the latter cladoprophylls may be primitive in Carex. It does not explain choice is correct. Indeed, in some members of section Fecun- how or why the tubular prophylls so widely present in inf- dae, e.g., C. chordalis, each of several peduncles occurring loresences of putative relatives were lost in the ancestors of at a node may differ in the presence, absence, or morphology Carex. of the cladoprophyll. Some peduncles may have no cladopro- The origin of the inflorescence units in subgenus Care-x by phyll, others may have more or less sheathing cladoprophylls varying degrees of reduction from compound inflorescence borne several millimetres above the base of the peduncle, and branches may explain both the variability of inflorescence form yet others may have sessile, perigynium-like (and sometimes characteristic of subgenus Carex and the occurrence of inflo- even fertile) cladoprophylls. Like other reports of fertile cla- rescence units in the position of perigynia and vice versa that FIGS.30 and 31. Inflorescence of Carex sp.nov. Fig. 30. Complete inflorescence. Fig. 31. Terminal portion of lowermost inflorescence unit from Fig. 30 (Loja Prov., Ecuador, Hitchcock 21556, NY). Bars = 2 cm. is also sometimes seen in subgenus Carex. This hypothesis The position of subgenus Indocarex provides a mechanism to explain the specialized nature of the The relationships of subgenus Indocarex within Carex are inflorescence in subgenus Carex, with usually elongate, sim- in some doubt. The subgenus has some apparently primitive ple inflorescence units and sexual specialization, with the lat- features and it has generally been regarded as a primitive group era1 inflorescence units usually different from the terminal. within Carex (Holttum 1948; Nelmes 1955; Savile and Calder This explains point 3 on page 1416. The hypothesis is also 1953; Koyama 1962). Nevertheless, it also shows some sim- consistent with points I, 2, and 4. ilarities to some species of subgenus Carex (Raymond 1959) FIGS.32-34. Inflorescences of Carex. Fig. 32. Portion of lowermost inflorescence unit of C. fecunda (Chapare Prov., Bolivia, Davidson 5138, MO). Fig. 33. Portion of lowermost inflorescence unit of Carex sp.nov. (Dept. Santa Cruz, Bolivia, Steinbach 8390, F). Fig. 34. Inflorescence of C. chordalis (Oaxaca, Mix- ico, Anderson 13175, MICH). Bars = 2 cm. FIGS.35 and 36. Inflorescence and spikes of Carex baalensis. Fig. 35. Inflorescence of C. baldensis. Fig. 36. Two spikes of C. baldensis with perigynium-like inflorescence prophylls (a) (cultivated, Reznicek 8250, MICH). Bars = 2 mm.

or may even be related to Schoenoxiphium, as suggested by as an inflorescence branch. If this interpretation is correct, cla- Haines and Lye (1983). It may be significant that no Anthra- doprophylls and inflorescence prophylls are not strictly homol- coidea smuts are known, as yet, in either Schoenoxiphium or ogous. Inflorescence prophylls are certainly homologous with in subgenus Indocarex (Kukkonen and Timonen 1979). Inter- perigynia and are sometimes even fertile. Cladoprophylls may estingly, the complanate, narrowly ovate rachillas of some be homologous with empty pistillate scales, as noted earlier. species of subgenus Indocarex, such as C. satsumensis The fact that in subgenus Indocarex, apparent transitional (Fig. l3), also resemble Schoenoxiphium rachillas in stages occur between cladoprophylls and fertile perigynia may miniature. have prompted a more global view of the homology of these If subgenus Indocarex is derived from Schoenoxiphium-like two structures than may be warranted. ancestors, the similarity of some members of subgenus Carex Examples of compounding of inflorescences by proliferation to subgenus Indocarex may suggest that subgenus Carex (as of rachillas are frequent (Smith and Faulkner 1976), although well as the genus Carex) is polyphyletic. Possibly, species of mostly teratological in origin. However, the unusual Indo- subgenus Carex that are putatively related to section Fecun- carex-like structure of C. baldensis L. of southern Europe is due, such as sections Limosae (0.Lang) Christ, Phacocystis likely derived from a unispicate ancestor by continued devel- (Acutae), Temnemis (Raf.) V. Krecz. (Cryptocarpae opment of rachillas, probably associated with development (L. Bailey) Mackenzie), etc., represent one evolutionary line towards an insect pollination syndrome (Hesse 1980). Carex and other sections another evolutionary line linked to baldensis has a few sheathless, leafy, horizontal bracts sub- Indocarex. tending a capitate inflorescence with striking, white perigynia Conversely, subgenus Indocarex may be derived from and scales (Fig. 35). The lowest nodes of the inflorescence ancestors similar to subgenus Carex by proliferation of rach- have bracts subtending often fertile perigynia with a sessile, illas to produce compound inflorescence units, this change per- androgynous spike whose axis is the rachilla of that perigyn- haps being correlated with occupation of tropical lowland hab- ium, as in subgenus Indocarex (Fig. 36). The middle inflo- itats. This possibility is suggested, as noted in point 5 on page rescence nodes have pistillate scales subtending perigynia with 14 16, by the perigynium-like inflorescence prophylls in short, included rachillas. The upper nodes have staminate subgenus Indocarex and supported further by the position of scales subtending male flowers. Carex baldensis differs from these prophylls. Inflorescence prophylls are sessile, like per- subgenus Indocarex in having a single capitate inflorescence igynia but unlike many cladoprophylls in the subgenera Carex and in occupying a temperate alpine habitat. Its Indocarex-like and Indocarex, which may be situated several millimetres features are almost certainly independently evolved. However, above the node, presumably because of intercalary growth it does provide an analogy for the possible derivation of the (Fisher and French 1978). This suggests that the inflorescence inflorescence units in subgenus Indocarex. The very different prophylls of subgenus Indocarex originate by a node devel- inflorescence of the peculiar, tristigmatic, Asian C. gibba oping first as a sessile female flower node, with the rachilla Wahlenb. in subgenus Vignea may also have originated in this then proliferating, rather than the node initially differentiating way, as it too has sessile, ~eri~ynium-likecladoprophylls at 1428 CAN. J. BOT. VOL. 68, 1990 the base of the spikes. Koyama (1957) also has suggested that is suggested. Carex baldensis, of uncertain subgeneric affinity certain other species with Indocarex-like morphology origi- (Nannfeldt 1977; Chater 1980), is suggested as an analogy, as nated through proliferation of rachillas. it has presumably independently derived Indocarex-like fea- Too little is presently known about diversity within subgenus tures. However, the possibility that subgenus Indocarex rep- Indocarex to do more than suggest possible evolutionary scen- resents an independent evolutionary line, perhaps related to arios. Much more detailed information is needed, especially Schoetzoxiphium, cannot be eliminated. about derivation of inflorescence structures, before firm con- The polarity of certain inflorescence features outlined on clusions are possible page 1415 is supported by further analysis. Compound inflo- rescence units, uniform and numerous androgynous spikes, lateral inflorescence units similar to the terminal, and pedun- Conclusions cled inflorescence units are primitive. Closer analysis of hom- Although many of the problems of previous hypotheses of ologies of vestigial structures and states of other characters evolution in Carex were outlined, and a few of these problems associated with known primitive states suggests that the fol- discussed in detail, much work remains before a coherent phy- lowing character states are also likely primitive: perigynia with logenetic classification of the genus is possible. Indeed, while an abaxial false suture, leafy but sheathless bracts, and per- a few character states in ~are~canbe polarized, a major block haps, inflorescence units multiple at the nodes and absence of to more detailed phylogenetic studies is still a lack of under- cladoprophylls. In both subgenus Carex and subgenus Vignea, standing of what are the primitive groups in Carex. In a very the features noted above occur in distigmatic as well as tris- real sense, the conclusions to this paper are of two kinds. One tigmatic species, so that polarity of stigma number is unclear. is dealt with briefly in this section on what we can conclude The presence of three stigmas, however, is certainly- primitive- from the present state of our knowledge, and the other is dis- in thk Cyperaceae as a whole. cussed in the next section and outlines what future work is More than in any other major group of flowering plants, needed to develop a natural classification of Carex and the evidence from fungal parasitism was used to derive phylogen- tribe Cariceae. ies in Carex. These prylogenies are not strongly based on mor- Rachillas, which have played a central role in some recent phology and thus are difficult to compare to other hypotheses. hypotheses of evolution in Carex, are more or less uniformly The grafting of putative smut evolutionary relationships onto present throughout Carex. Hypotheses of evolution in Carex Carex promoted extensive speculation designed to bridge that hinge on the occurrence of rachillas only in certain uni- offending gaps in morphology, rather than consideration of as spicate species are false. While the presence of a rachilla is wide a body of evidence as possible. As well, studies using not of fundamental evolutionary significance within Carex, the smuts must fulfill two vital criteria before use of smut data to morphology and anatomy of rachillas do offer data of system- construct sedge phylogenies is admissable: (i) demonstration atic and evolutionary importance. Based on rachilla morphol- of a detailed phylogeny of smuts based on cladistic method- ogy, C. kitzgii (formerly Uncinia kingii) is unrelated to Uncinia ology; (ii) demonstration that smut evolution and sedge evo- and closely related to C. ttzicroglochin. Hypotheses suggesting lution are inextricably linked in parallel. Neither of these points the evolution of some species of Carex from Uncinin, or all was convincingly presented in papers published to date. Smut of the genus Uncinia from Carex, based on the transitional data, however, are very useful in suggesting affinities between nature of C. kingii and C. rnicroglochin, are false. In fact, the major evolutionary lines and in the detection of convergence supposedly close relationship of Cnrex and Uncinia needs to and reduction. be reevaluated. Putatively primitive rachillas in Cnrex are rel- The relationship of chromosome numbers to evolution in atively large, two-veined: and more or less flattened. The large Carex is still problematic. Heilbom's (1924) view that low rachillas of C. microglochir~and its allies, however, are quite chromosome numbers must be primitive undoubtedly played different in structure from those in the remainder of Carex and a role in various suggestions that section Acrocytis (Montanae) may be secondary in origin. Even excluding C. microglochin is primitive. In the absence of a detailed phylogeny, these sug- and its allies, unispicate Carex are a diverse group and do not gestions could not be easily challenged. However, it now constitute a natural.group, as was widely recognized. Remov- seems clear that the tiny plants in section Acrocystis (Motz- ing those species without a rachilla into other subgenera, as tanae), with xeromorphic adaptations and reduced infloresc- suggested by Savile and Calder (1953), does not help. The ences, are highly specialized, as are probably also the wide- sharing of a primitive feature, i.e., large rachillas, is not evi- leaved forest understory species that have low chromosome dence of close relationship. Many, though not all, extant uni- numbers (Tanaka 1940; Hoshino 1981). Very few chromo- spicate Carex likely originated from a series of independent, some numbers are known for tropical species; however, the early reductions to a unispicate condition, this accounting for only one known for a species of section ~ecundae(2n = 80- the seeming impossibility of associating many unispicate spe- 82, in C. pichinchensis Kunth (Huynh 1965)) is relatively high. cies with extant multis~icateCarex. If the hypotheses proposed here are even partially correct, The most compounded inflorescences in Cnrex are found reduction in chromosome numbers must also be a frequent not in subgenus Indocarex, as often stated, but in subgenus phenomenon in Carex, as well as an increase in numbers. This

Vignea. The inflorescence units of subgenus Carex are hypoth- reduction in chromosome number is ,~ostulated - to be associated esized to have been derived by the reduction of long ranks of with fixing of favorable allelic combinations related to migra- small androgynous spikes, as found in some compound inflo- tion into specialized habitats, as suggested by Stebbins (1974) rescences of subgenus Vignen, to long ranks of single peri- for a number of other genera. In fact, the same cytological gynia. Cladoprophylls, widely present in subgenera Cares and peculiarities of Carex (Wahl 1940) that allow fragmentation Indocnrex, may be derived from empty scales at the bases of presumably could also allow aggregation. The low chromo- inflorescence unit axes, as found in subgenus Vignea. some numbers of some putatively reduced and specialized sec- The origin and relationships of subgenus Indoc~zrexremain tions, and the essentially normal distribution of chromosome unresolved. The possibility that it may be secondarily derived numbers in the genus as a whole and in large sections (Wahl 1940), support this conclusion. As well, the lower chromo- resolve the developmental and morphological stages involved some numbers in Carex are among the lowest known in the in the evolution of the various organs. Cyperaceae, even though Carex is considered derived within With about 2000 species, probably the greatest barrier to the family. Furthermore, it is not known at present what the work on more detailed evolutionary relationships within the difference is in DNA content between, for example, C. tri- genus Carex is the lack of the most basic knowledge about chocarpa Willd. with 2n = 110 (Wahl 1940) and C. peniz- many species. Hundreds of species, especially tropical ones, sylvanica Lam. with 2n = 18 (Crins and Ball 1983), although are very poorly collected. Many of these poorly known species data are accumulating in this area (Nishikawa et al. 1984). were never illustrated and are sometimes known only from the This factor also renders dubious the use of simple chromosome original descriptions, which almost invariably are incomplete number data in evolutionary studies of Carex. Of course, chro- and are sometimes erroneous. Monographic work to develop mosome numbers may still be valuable in systematic studies a satisfactory disposition of species into natural sections and at the species and sectional level. to treat sections of natural groups of species over all or most On a broader scale, other ramifications are clear from the of their ranges is probably- the single most vital need. This information discussed. The occurrence of the putatively most work is the foundation upon which other work in systematics primitive segment of subgenus Vignea in South America, com- (and many other fields) is based, as eloquently stated by Wilson bined with the Southern Hemisphere distribution of Uncinia, (1985). Within small, well-defined natural groups of species, the primarily South African distribution of Schoenoxiphium, the difficulties of applying cladistic methodologies also may and the Himalayan and central Asian distribution of Kobresia, be reduced (Crins and Ball 1988; Crins 1990). Especially for suggests a Gondwanaland origin for both the tribe Cariceae the tropical species, this need should override all others and the genus Carex. The time of origin of the tribe Cariceae because of habitat destruction. Even in this paper, two of the and probably the genus Carex, given this distribution, must tropical species figured (Figs. 30, 31, and 34) are as yet unde- date back at least to the early Tertiary (Raven and Axelrod scribed. Descriptions of species should not only include the 1974). Carex may have evolved in sunny, montane tropical or bare minimum needed for diagnostic purposes, but also subtropical wetlands and subsequently occupied temperate and describe the entire'gant, paying close attention to the inflo- boreal regions, as well as radiating into shaded, alpine, and rescence structures and also including, as much as possible, xeric habitats. This also suggests a complex origin for the vegetative organs (Reznicek and Catling 1986; Tang and Liang South American Carex flora, for example. Many South Amer- 1987). Illustrations are also highly desirable. ican species, such as some members of sections Glareosae More work is needed to establish the limits of subgenus" (Canescentes, Heleonastes) and Ovales are identical or very Indocarex and to achieve a clear morphological definition of similar to boreal species (Moore and Chater 1971) and prob- the subgenus (if possible), especially with regard to ambigu- ably have relatively recently migrated into South America. ous, reduced members. Information about inflorescence mor- Other species, such as members of section Fecundae, are prob- phology and development may be particularly useful in assess- ably autochthonous and presumably diversified as the Andes ing whether subgenus Indocarex is polyphyletic or not, and if were uplifted (Van der Hammen 1988). If the hypothesis that not, whether it is primitive or secondarily derived. The genus many unispicate species are descendants of early reductions Schoenoxiphium should be examined in conjunction with these from more primitive multispicate ancestors with large rachillas studies. is correct, this area of origin of Carex is also consistent with While unispicate Carex are well known not to be a natural the disproportionate representation of unispicate species, many group, the precise relationships of the majority of unispicate with well-developed rachillas, in the tropics and Southern species are as uncertain today as they were in Kreczetovicz's Hemisphere. time. The extreme reduction and specialization they have The distribution of the genera of the Cariceae as a whole, undergone makes understanding their affinities perhaps the together with the large, highly compounded inflorescences of most difficult problem in Carex. Even in the genus-as a whole, the putatively primitive species of Carex, also suggests that reduced flowers and relatively uniform vegetative morphology Carex and Kobresia diverged from common ancestors far in mean few macromorphological characters are available. The the past. Carex probably did not evolve from within Kobresia, repeated occurrence of parallelisms and reversals in Carex as suggested by some previous authors, since species with complicates the situation. Thus, search for more characters of inflorescences much more elaborated than any Kobresia occur systematic value is also a high priority. Surveys of rachilla in all three subgenera of Carex. morphology and position, and other micromorphological and anatomical characters such as stomate type and leaf anatomical Future work characters, need to be conducted. Essentially nothing is known about these features for most species of Carex even though When there is so much to do, it is difficult to assign prior- their study has been greatly facilitated by SEM (Shepherd ities to future work. Indeed, we know so little that it is sorne- 1977, Standley 1990). These data and the classifications what difficult to predict what will repay future investigations derived from them are vital for determination of those species the most. Testing of the hypotheses presented in this paper is and groups of species that will most repay use of powerful probably possible, on a broad scale, with our present knowl- tools such as isozyme and DNA analyses to achieve a finer edge and using techniques such as chloroplast DNA restriction level of resolution of phylogeny in Carex. site analysis (Manhart 1990). Results produced with this tech- Chromosome numbers are now known for many hundreds nique are not plagued by high levels of homoplasy. This type of species and should continue to be gathered, especially for of large scale work will bring about a clearer understanding of tropical species. However, before chromosome number data the homology of different structures in Carex and the tribe can play a major role in reconstruction of evolution in Carex, Cariceae. It will likely allow a cladistic analysis of the tribe the relationship between chromosome numbers and DNA con- Cariceae at the generic and subgeneric levels, which would tent must be established. Work is also needed to establish the be a major breakthrough. However, it will not necessarily role of reduction in chromosome number in Carex and the 1430 CAN. J. BOT. \ overall significance of agmatoploidy to speciation (Whitkus BRITON, N. L. 1907. The sedges of Jamaica. Bull. Dep. Agric. 19871. Jamaica, 5 (Suppl. 1): 1-19. Testing the hypotheses of inflorescence evolution proposed CHATER,A. 0. 1980. Carex. In Flora Europaea. Vol. 5. Edited by here should also be a high priority. This should involve devel- T. G. Tutin, V. H. Heywood, N. A. Burges, D. M. Moore, S. M. Walters, and D. A. Webb. Cambridge University Press, opmental studies of those species showing characteristics inter- Cambridge. pp. 290-323. mediate between subgenera. Polarization of more features of CLARKE,C. B. 1898. Carex. In Flora Capensis. VII. Edited by W. T. inflorescences may also be a by-product of these investiga- Thiselton-Dyer. Lovell Reeve & Co. Ltd., London. pp. 299- tions. Study of the ontogeny and development of the abaxial 310. false suture in subgenus Vignea to determine whether or not 1904. List of the Carices of Malaya. J. Linn. Soc. Bot. 37: it is a vestige of an open perigynium is essential. As well, the 1-16. inflorescences of species such as C. willdenowii and its allies 1908. New genera and species of Cyperaceae. Kew Bull. should be studied further to elucidate their peculiar structures. Misc. Inf. Addit. Ser. 8. Developmental and anatomical studies of C. kingii, C. micro- CLARKSON,R. B. 1962. Fraser's sedge, Cymophyllus fraseri glochin, and their allies, aimed at testing the hypothesis that (Andrews) Mackenzie. Castanea, 26: 129-136. the rachillas of these species are secondary, are also needed CLEEF,A. M. 1982. Distribucion y ecologia de Vesicarex colluman- thus Steyermark (Cyperac.) Acta Biol. Colombiana, 1: 4349. and may provide clues about the true relationships of these CRINS,W. J. 1990. Phylogenetic considerations below the sectional remarkable species. Finally, comparative developmental and level in Carex. Can. J. Bot. 68: 1433-1440. anatomical studies, concentrating on the orientation and ven- CRINS,W. J., and BALL,P. W. 1983. The taxonomy of the Carex ation of inflorescence prophylls and cladoprophylls, are needed pennsylvanica complex (Cyperaceae) in North America. Can. J. to determine their homologies. Bot. 61: 1692-1717. The end point, a phylogenetic classification of Carex to the 1988. Sectional limits and phylogenetic considerations in species level, still remains an ideal to strive for rather than a Carex section Ceratocystis (Cyperaceae). Brittonia, 40: 38-47. goal in sight. The first step in this direction, a phylogenetic DEWEY,C. 1866. Caricography: index to species. Am. J. Sci. Arts classification at the subgeneric and possibly the sectional level, Ser. 2, 42: 1-10.- though not yet possible, is an attainable goal. DREJER,S. 1844. Symbolae Caricologicae ad synonymiam Caricum extricandam stabiliendamque et affinitates naturales emendas. Bianco Luno, Halfniae. Acknowledgments EITEN,L. T. 1976. Inflorescence units in the Cyperaceae. Ann. Mo. Bot. Gard. 63: 81-112. My thanks go to Martha Perkins, who typed the messy first EGOROVA,T. V. 1971. Notes on the nomenclature of the genus Carex draft of this manuscript, and to my wife Susan, who did the L. [In Russian with English translation, Amerind Publishing Co., drawing for Fig. 1 and helped a great deal with editorial work. New Delhi, 1978.1 Novit.Syst. 8: 88-89. I am particularly grateful to Jim Manhart and Lisa Standley 1972. Classification and summary of sedges of the subgenus for the discussions on evolution in Carex that we had. Sheila Carex in the European part of the USSR. [In Russian with Eng- Hooper and David Simpson at Kew sent a copy of H.K. Airy lish translation, Amerind Publishing Co., New Delhi, 1978.1 Shaw's manuscript translation of Kreczetovicz's (1936) clas- Novit. Syst. 9: 64-95. sical article and Paul Catling at DAO sent titles concerning FERNALD,M. L. 1926. Two summers of botanizing in Newfound- smuts and sedge evolution. Edward Voss provided much land. Rhodora, 28: 49-63. FISHER,J. B., and FRENCH,J. C. 1978. Internodal meristems of mon- needed nomenclatural advice and Gerald Wheeler provided ocotyledons: further studies and a general taxonomic summary. helpful information on the identity of the genus Vesicarex and Ann. Bot. (London), 42: 41-50. other useful comments on the manuscript. Jeremy Bruhl also GILLY,C. L. 1952. Phylogenetic development of the inflorescence commented on the manuscript. Finally, an anonymous and generic relationships in the Kobresiaceae. Iowa State Coll. reviewer provided numerous helpful and insightful J. Sci. 26: 210-212. suggestions. GOETGHEBEUR,P. 1985. Studies in Cyperaceae 6. Nomenclature of the suprageneric taxa in the Cy peraceae. Taxon, 34: 6 17-632. ALEXEEV,Y. E. 1978. Morphological elements of inflorescence and 1987. A holosystematic approach of the family Cyperaceae. problems of the evolution of sedges of the subgenera Primocarex In Abstracts of the 14th International Botanical Congress, Berlin, Kukenth. and Vignea (Beauv.) Peterm. [In Russian with English July 24 to August I, 1987. Edited by W. Greuter, B. Zimmer, summary.] Bjull. Moskovsk. Obsc. Insp. Prir. Otd. Biol. 83: and H.-D. Behnke. Botanical Museum Berlin-Dahlem, Berlin. 84-93. p. 276. (Abstr.) BAILEY,L. H. 1885. Carex. In Manual of the botany of the Rocky GREEN,M. L. 1929. Cares. In Standard-species of Linnean genera Mountain region. Edited by J. M. Coulter. Ivison, Blakeman, of Phanerogamae.(1753-54). By A. S. Hitchcock and M. L. Taylor and Co., New York. pp. 370-397. Green. p. 187. In International Botanical Congress, Cambridge 1886. A preliminary synopsis of North American Carices. (England) 1930. Nomenclature proposals by British botanists. Proc. Am. Acad. Arts, 22: 59-157. Edited by T. A. Sprague, J. Ramsbottom, E. M. Wakefield, and BARNARD,C. 1957. Floral histogenesis in the monocotyledons. 11. A. J. Wilmot. Wyman & Sons, London. pp. 111-199. The Cyperaceae. Aust. J. Bot. 5: 115-128. HAINES,R. W. 1967. Prophylls and branching in Cyperaceae. J. East. BLASER,H. W. 1944. Studies in the morphology of the Cyperaceae. Afr. Nat. Hist. Soc. Nat. Mus. 26: 51-70. 11. The prophyll. Am. J. Bot. 31: 53-64. HAINES,R. W., and LYE,K. A. 1972. Studies in African Cypera- BOOT, F. 1858. Illustrations of the genus Carex. Part I. William ceae. VII. Panicle morphology and possible relationships in Pamplin, London. pp. 1-74, P1. 1-200. Sclerieae and Cariceae. Bot. Not. 125: 331-343. 1860. Illustrations of the genus Carex. Part 11. William Pam- 1983. The sedges and rushes of East Africa. East African plin, London, pp. 75-104, P1. 201-310. Natural History Society, Nairobi. 1862. lllustrations of the genus Carex. Part 111. William Pam- HEILBORN,0. 1924. Chromosome numbers, species formation, and plin, London. pp. 105-126, P1. 31 1-41 1. phylogeny in Cares. Hereditas, 5: 129-216. 1867. lllustrations of the genus Carex. Part IV. L. Reeve and HESSE,M. 1980. Entwicklungsgeschichte und Ultrastmktur von pol- Co., London. pp. 127-234, P1. 412-600. lenkitt und exine bei nahe verwandten entomophilen und ane- mophilen Angiospermensippen der Alismataceae, Liliaceae, Africa chiefly from Natal: XIII. Edited by 0. M. Hilliard and Juncaceae, Cyperaceae, Poaceae und Araceae. Plant Syst. Evol. B. L. Burtt. Notes R. Bot. Gard. Edinb. 43: 345-405. 134: 229-267. 19868. Special features of the inflorescence structure in the HITCHCOCK,A. S., and GREEN,M. L. 1947. Species lectotypicae family Cyperaceae. Ann. Bot. Fenn. 23: 107-120. generum Linnaei. Brittonia, 6: 1 14- 1 18. KUKKONEN,I., and TIMONEN,T. 1979. Species of Ustilaginales, HOLM,T. 1896. Studies upon the Cyperaceae 11. The clado- and especially of the genus Anthracoidea, as tools in plant taxonomy. antho-prophyllon in the genus Carex. Am. J. Sci. 2: 214-220. Symb. Bot. Ups. 22: 166-176. HOLTTUM,R. E. 1948. The spikelet in Cyperaceae. Bot. Rev. 14: LE COHU,M.-C. 1968. Remarques sur I'inflorescence femelle des 525-54 1. Carex interprktation de faits tkratologiques. Bot. Rhedonica Ser. HOSHINO,T. 1981. Karyomorphological and cytological studies on A, 5: 37-45. aneuploidy in Carex. J. Sci. Hiroshima Univ. Ser. B Div. 2 LEVYNS,M. R. 1945. A comparative study of the inflorescence in (Bot.), 17: 155-238. four species of Schoenoxiphium and its significance in relation HUTTON,E. E. 1976. Dissemination of perigynia in Carex pauci- to Carex and its allies. J. S. Afr. Bot. 11: 79-89. flora. Castanea, 41: 346-348. MACKENZIE,K. K. 1913. Carex. In An Illustrated flora of the north- HUYNH,K. L. 1965. Contribution a I'itude caryologique et embry- western United States and Canada and the British possessions. ologique des Phanirogarnes du PCrou. Denkschr. Schweiz. 2nd ed. Vol. 1. By N. L. Britton and A. Brown. Charles Scrib- Naturforsch. Ges. 85: 1-178. [Not seen.] ner's Sons, New York. pp. 352-441. IVANOVA,N. A. 1939. The genus Kobresia Willd., its morphology 1931. Cyperaceae - Cariceae. N. Am. Flora, 18 (1-3): 1- and systematics. [In Russian with English summary.] Bot. Zh. 168. SSSR, 24: 455-503. 1935. Cyperaceae - Cariceae. N. Am. Flora, 18 (4-7): JERMY,A. C., CHATER,A. O., and DAVID,R. W. 1982. Sedges of 169-47 8. the British Isles. Botanical Society of the British Isles, Handbook MANHART,J. R. 1990. Chemotaxonomy of the genus Carex. Can. No. 1. 2nd ed. Botanical Society of the British Isles, London. J. Bot. 68: 1457-1472. KERN,J. H. 1958. Florae Malesianae precursores. XXI. Notes on MCNEILL,J. 1987. 14th International Botanical Congress: mail vote Malaysian and some S. E. Asian Cyperaceae. VII. Acta Bot. and final congress action on nomenclature proposals. Taxon, 36: Neerl. 7: 786-800. 858-868. - 1962. New look at some Cyperaceae mainly from the tropical MCNEILL,J., ODELL,E. A,, CONSAUL,L. L., and KATZ,D. S. 1987. standpoint. Adv. Sci. 19: 14 1-148. American code and later lectotypifications of Linnaean generic KERN,J. H., and NOOTEBOOM,H. P. 1979. Cyperaceae. 11. Flora names dating from 1753: a case study of discrepancies. Taxon, Malesiana Ser. I, 9: 107-187. 36: 350-40 1. KOYAMA,T. 1957. An enumeration of Hayata's Indo-Chinese col- MEERT,M., and GOETGHEBEUR,P. 1979. Comparative floral mor- lection of Cyperaceae. Contrib. Inst. Bot. Univ. Montreal, 70: phology of Bisboeckelereae and Cariceae (Cyperaceae) on the 5-64. basis of the anthoid concept. Bull. Soc. R. Bot. Belg. 112: 128- 1961. Classification of the family Cyperaceae (1). J. Fac. 142. Sci. Univ. Tokyo Sect. 3 Bot. 8: 37-148. MEEUSE,A. D. J. 1975. Interpretative floral morphology of the 1962. Classification of the family Cyperaceaze (2). J. Fac. Cyperaceae on the basis of the anthoid concept. Acta Bot. Neerl. Sci. Univ. Tokyo Sect. 3 Bot. 8: 149-278. 24: 29 1-304. 1965. Interrelationships between the tribes Lagenocarpeae METCALFE,C. R. 1969. Anatomy as an aid to classifying the Cyper- and Sclerieae (Cyperaceae). Bull Torry Bot. Club, 92: 250-265. aceae. Am. J. Bot. 56: 782-790. 197 1. Systematic interrelationships among Sclerieae, Lagen- 197 1. Anatomy of the monocotyledons. V. Cyperaceae. Clar- ocarpeae and Mapanieae (Cyperaceae). Mitt. Bot. Staatssamml. endon Press, Oxford. Muench. 10: 604-617. MOORE,D. M., and CHATER,A. 0. 1971. Studies of bipolar disjunct KRECZETOVICZ,V. I. 1935. Carex. In Flora URSS. Vol. 3. Edited species I. Carex. Bot. Not. 124: 317-334. by V. L. Komarov. [In Russian with English translation by N. MORA-OSEJO,L. E. 1982. Consideraciones sobre la morfologia, ana- Landau, Jerusalem, 1964.1 Academiae Scientiarum URSS, Len- tomia y posicion sistematica de Vesicarex Steyermark (Cypera- ingrad. pp. 1 11-464. ceae). Acta Biol. Colombiana, 1: 31-41. 1936. Are the sedges of subgenus Primocarex Kiik. primi- NANNFELDT,J. A. 1977. The species of Anthracoidea (Ustilaginales) tive? [In Russian with English summary.] Bot. Zh. SSSR, 21: on Carex subgen. Vignea with special regard to the Nordic spe- 395-425. cies. Bot. Not. 130: 351-375. KUKENTHAL,G. 1909.-Cyperaceae - Caricoideae. In Das Pflan- NELMES,E. 195 1. The genus Carex in Malaysia. Reinwardtia, 1: zenreich. IV. Vol. 20, Heft 38. Edited by A. Engler. Wilhelm 221-450. Englemann, Leipzig. pp. 1-824. 1952. Facts and speculations on phylogeny in the tribe Car- KUKKONEN,I. 1963. Taxonomic studies on the genus Anthracoidea iceae of the Cyperaceae. Kew Bull. 195 l: 427436. (Ustilaginales). Ann. Bot. Soc. Zool. Bot. Fenn. 'Vanamo', 1955. The genus Carex in Indo-China, including Thailand 34(3): 1-122. and Lower Burma. Mem. Mus. Natl. Hist. Nat. Ser. B Bot. 4: 1967a. Vegetative anatomy of Uncinia (Cyperaceae). Ann. 83-182. Bot. (London), 31: 523-544. NISHIKAWA,K., FURUTA,Y., and ISH~TOB~,K. 1984. Chromosomal 19678. Gedanken und Probleme zur Systematik der Familie evolution in genus Carex as viewed from nuclear DNA content, Cyperaceae. Eine Zusammenfassung. Aquilo Ser. Bot. 6: 18- with special reference to its aneuploidy. Jpn. J. Genet. 59: 465- 42. 472. 1967c. Spikelet morphology and anatomy of Uncinia Pers. OHWI,J. 1936. Cyperaceae Japonicae. I. A synopsis of the Caricoi- (Cyperaceae). Kew Bull. 21: 93-97. deae of Japan, including the Kuriles, Saghalin, Korea, and For- 1970. Vegetative anatomy of Carex microglochin Wahlenb. mosa. Mem. Coll. Sci. Kyoto Imp. Univ. Ser. B. Biol. 11: 229- and Carex camptoglochin Krech. New research in plant anatomy. 530. J. Linn. Soc. Lond. Bot. 63 (Suppl. 1): 137-145. RAVEN,P. H., and AXELROD,D. L. 1974. Angiosperm biogeography 1983. The genus Schoenoxiphium (Cyperaceae). A prelimi- and past continental movements. Ann. Mo. Bot. Gard. 61: 539- nary account. Bothalia, 14: 819-823. 673. 1984. On the inflorescence structure in the family Cypera- RAYMOND,M. 1959. Carices Indochinenses necnon Siamenses. Mem. ceae. Ann. Bot. Fenn. 21: 257-264. Jard. Bot. Montreal, 53: 1-125. 1986~.Schoenoxiphium. In Notes on some plants of southern REZNICEK,A. A. 1986. The taxonomy . of Carex. sect. Hymenoch- 1432 CAN. J. BOT. VOL. 68, 1990

laenae (Cyperaceae) in Mexico and Central America. Syst. Bot. STEYERMARK,J. A. 1951. Botanical exploration in Venezuela. I. 11: 56-87. Fieldiana Bot. 28(1): 1-242. REZNICEK,A. A,, and CATLING,P. M. 1986. Vegetative shoots in SVENSON,H. K. 1972. The rachilla in Cape Cod species of Carex the taxonomy of sedges (Carex, Cyperaceae). Taxon, 35: 495- with notes on the history of the perigynium and rachilla. Rho- 501. dora, 74: 321-330. ROBERTSON,A. 1979. History of the classification of the genus Carex. TANAKA,N. 1940. Chromosome studies in Cyperaceae VIII. Meiosis Taxon, 28: 535-548. in diploid and tetraploid forms of Carex siderosticta Hance. ST. JOHN,H., and PARKER,C. S. 1925. A tetramerous species, sec- Cytologia, 11: 282-310. tion, and subgenus of Carex. Am. J. Bot. 12: 63-68. TANG,Y., and LIANG,S. 1987. On Carex siderosticta with special SAVILE,D. B. 0. 1972. Arctic adaptations inplants. Can. Dep. Agric. reference to the term "stem lateral" in caricology. Acta Bot. Monogr. No. 6. Yunnanica, 9: 309-3 14. 1979. Fungi as aids in higher plant classification. Bot. Rev. THOMAS,W. W. 1984. Insect pollination of Cymophyllus fraseri 45: 377-503. (Andrews) Mackenzie. Castanea, 49: 94-95. SAVILE,D. B. O., and CALDER,J. A. 1953. Phylogeny of Carex in TIMONEN,T. 1985. Synflorescence morphology and anatomy in the light of parasitism by the smut fungi. Can. J. Bot. 31: 1% Kobresia laxa (Cyperaceae). Ann. Bot. Fenn. 22: 153-171. 174. 1989. Synflorescence structure of Schoenoxiphium latzceum SCHULTZE-MOTEL,W. 1959. Entwicklungsgeschichtliche und ver- (Cyperaceae). Ann. Bot. Fenn. 26: 319-342. gleichend-morphologische Untersuchungen im Bliitenbereich der TUCKER,G. C. 1987. The genera of Cyperaceae in the southeastern Cyperaceae. Bot. Jahrb. Syst. Pflanzengesh. Pflanzengeogrf. 78: United States. J. Arnold Arbor. Harv. Univ. 68: 361-445. 129-170. TUCKERMAN,E. 1843. Enumeratio methodica caricum quarundam. 1964. Cyperales. In Engler's Syllabus der Pflanzenfamilien. Isaacus Riggs, Schenectady. 12th ed. Vol. 2. Edited by H. Melchior. Gebriider Borntraeger, VANDER HAMMEN,T. 1988. The tropical flora in historical perspec- Berlin. pp. 602-607. tive. Taxon, 37: 5 15-5 18. SHEPHERD,G. J. 1977. The use of anatomical characters in the intra- Voss, E. G. 1972. Additional nomenclatural and other notes on generic classification of Carex (Cyperaceae). Hoehnea, 6: 33- Michigan monocots and gymnosperms. Mich. Bot. 11: 26-37. 54. WAHL,H. A. 1940. dfiromosome numbers and meiosis in the genus SMITH,D. L. 1966. Development of the inflorescence in Carex. Ann. Carex. Am. J. Bot. 27: 458-470. Bot. (London), 30: 475-486. WHEELER,G. A. 1989. A new species of Carex sect. Abditispicae 1967. The experimental control of inflorescence development (Cyperaceae) from northern Argentina and the status of Vesicarex in Carex. Ann. Bot. (London), 31: 19-30. collumanthus in South America. Syst. Bot. 14: 37-42. 1969. The role of leaves and roots in the control of inflo- WHEELER,G. A,, and OWNBEY,G. B 1984. Annotated list of Min- rescence development in Carex. Ann. Bot. (London), 33: 505- nesota Carices with phytogeographical and ecological notes. 5 14. Rhodora, 86: 151-23 1. SMITH,D. L., and FAULKNER,J. S. 1976. The inflorescence of Carex WHITKUS,R. 1987. Mechanism of chromosome number change in and related genera. Bot. Rev. 42: 53-81. Carex, a genus exhibiting agmatoploidy. Am. J. Bot. 74: 1765. SNELL,R. S. 1936. Anatomy of the spikelets and flowers of Carex, (Abstr.) Kobresia, and Utlcitlia. Bull. Torrey Bot. Club, 63: 277-295. WILSON,E. 0. 1985. Time to revive systematics. Science (Wash- STANDLEY,L. A. 1990. Anatomical aspects of the taxonomy of sedges ington, D.C.), 230: 1227. (Carex, Cyperaecae). Can. J. Bot. 68: 1449-1456. EBBIN IN^, G. L. 1974. Flowering plants, evolution above the species level. Belknap Press. Harvard University, Cambridge, MA.