Amer. 1. Bot. 59(4): 423-436. 1972.

A RESOLUTION OF THE - CONTROVERSY () 1

RUDOLF SCHMID Department of , The University of Michigan, Ann Arbor"

ABSTRACT Floral anatomy now provides additional, strong evidence confirming the distinctness of the mainly Eugenia s. s. and the strictly Syzygium s. I. Most significantly, of Eugenia s. s. have a transeptal vascular supply to the ovules whereas those of Syzygium s. I. have an axile one. Other features of floral histology and vasculature also support such a division. In addition, a review of the taxonomic literature revealed three hitherto neglected organographic criteria-nature of bracteoles, presence or absence of pu­ bescence, and presence or absence of pseudopedicels-that sharply distinguish between Eugenia s. s. and Syzygium s. I. An ensemble of these and other organographic criteria further demon­ strates the basic disparity of these taxa. The organography and histology of of Eugenia s. I. are described in detail, with .26 characters contrasting the Old and New World species included in a table.

THE MYRTACEAE are replete with vexing and species, have been or could be reduced to Sy­ provocative nomenclatural and systematic prob­ zygium P. Browne ex J. Gaertner (1788) (Schmid, lems. The large, protean Eugenia Linnaeus unpublished data). If in turn Syzygium is reduced (1753 ),3 named in honor of Francois Eugene de to Eugenia, as has been done by many authors Savoie-Carignan (1663-1736), prince of Savoy, (see below), these 35 or more names could in­ Austrian general, and distinguished patron of art, flate the number of generic synonyms of Eugenia science, and literature, has been one of the most s. I. to nearly 70 since more than 30 mainly difficult and controversial genera in the angio­ American genera (including the Old World genus sperms to define.' Since the time of Linnaeus Iossinia Commerson ex A. P. de Candolle, 1828) many species from both the Old and New World have been considered congeneric with Eugenia s. s. have been assigned to Eugenia. By 1938 some by various workers. Clearly the genus Eugenia 2500 species had been described as belonging to, has become unwieldy, and a very complicated or transferred to, this genus (Merrill and Perry, synonymy and difficult are involved." 1939); by 1950 another 100 binomials had been There are three schools of thought regarding published under Eugenia (Merrill, 1950b); and the disposition of Eugenia and its Old World by 1965 an additional 200 names (as listed in segregates (for historical accounts see Kausel, Index Kewensis), to which may be added 22 new 1956; Merrill and Perry, 1939): names in one recent work (McVaugh, 1969) (1) Following Wight (1841) and Bentham alone. About 35 generic names (validly and in­ and Hooker (1862-67; see Bentham, 1869, for validly published), based on various Old World their rationale), who promulgated Wight's then novel treatment of Eugenia, many authors (Bailey, 1 Received for publication 21 October 1971. This study is based on a portion of a dissertation 1930, but not 1949; Baillon, 1880; Bullock and submitted in partial fulfillment of the requirements for Harrison, 1958; Corner, 1952; Fawcett and the degree of Ph.D. at The University of Michigan, Ann Rendle, 1926; Gray, 1854; Henderson, 1949; Arbor. This research was supported in part by a grant from Merrill, initially in 1917; Ridley, 1922; Wilson, the National Science Foundation, GB-82l2, to N. G. 1957, but not 1960) have preferred to retain Hairston, The University of Michigan, for research in almost all species of both hemispheres in one Systematic and Evolutionary Biology. I thank Dr. Rogers McVaugh, Dr. Warren H. Wagner, Jr., Dr. Charles B. immense collective genus Eugenia sensu latissimo. Beck, and my wife Marvin for their critical reading of (2) Other workers (Airy Shaw, 1949, 1966), the manuscript, and the last also for doing the drawings. following Niedenzu (1893), who restricted Eugenia 2 Present address: Visiting Research Associate, De­ partment of Botany, Smithsonian Institution, Washington, to largely American species, have favored the D. C. 20560. segregation of most of the Old World species into 3 Authorities and dates of publication of genera follow a second large collective genus Syzygium sensu the Index Nominum Genericorum, which should be con­ sulted for details. 5 Merrill and Perry (1939) have given the most de­ 4 For especially good taxonomic discussions of the tailed but far from complete synonymy for Syzygium. Eugenia-Syzygium controversy see Amshoff (1958), The most extensive synonymy occurs in Airy Shaw Bentham (1869), Henderson (1949), Kausel (1956), (1966). See Kausel (1957b), McVaugh (1968), and McVaugh (1956a, b), Merrill (1950b), Merrill and Perry Niedenzu (1893), for example, for generic synonymy of (1939), and Wilson (1957). Eugenia s. s. 423 424 AMERICAN JOURNAL OF BOTANY [Vol. 59 latissimo, or, as did Niedenzu (1893), into two (1855-56), are now accepted than Old World large genera Syzygium and Jambosa Adanson ex segregates. Most recent authors tend to segre­ A. P. de Candolle (1828), at the same time deny­ gate Syzygium from Eugenia s. I. and occa­ ing the validity of most or all of the small generic sionally to accept a number of other Old World segregates proposed. segregates: most notably, , Piliocalyx, Aci­ (3) Following A. P. de Candolle (1828, 1842), calyptus, Aphanomyrtus, Cleistocalyx, Cuphean­ who accepted Eugenia, Syzygium, Caryophyllus thus Seemann (1865), Iambosa, Iossinia, and Linnaeus (1753), Acmena A. P. de Candolle Pareugenia. Kausel (1956, 1957a, b, 1960, 1966) ( 1828), Jambosa, and Jossinia, still other bot­ represents the extreme liberal view and has ac­ anists (Amshoff, 1963; Bailey, 1949; Diels, 1922; cepted some 33 generic segregates of Eugenia s. s. Guillaumin, 1938; Hosokawa, 1940; Kausel, 1956, and Syzygium s. I., 22 and 11, respectively, whereas 1957b, 1966; Merrill and Perry, see below) have Airy Shaw (1966) recognized 16 and 6 segre­ recognized Svzygium (sometimes excluding Jam­ gates, respectively. The segregates accepted by bosa; or, in fact, accepting Jambosa in place of Kausel include such Old World taxa as Acmena, Syzygium, as did Blume, 1849-51) for most of Caryophyllus, Iambosa, Aphanomyrtus, Cleisto­ the Old World species and have advocated the calyx, Pareugenia, Iossinia, Cupheanthus, Pilio­ acceptance of certain small and somewhat distinct calyx, Episyzygium Suessenguth et Ludwig (1950), segregate genera. Chloromyrtus Pierre (1898) ,Meteoromyrtus Exemplifying the third school of thought, Mer­ Gamble (1918), and Acmenosperma Kausel rill and Perry, by virtue of an extended series of (1957a). In fact, Kausel (1956, 1957b, 1966) works on the Asiatic Myrtaceae (most importantly, placed Eugenia and Syzygium in separate sub­ 1937, 1938a-c, 1939, 1942a, b; Merrill, 1937, families, his Eugenioideae and Plinioideae, re­ 1939, 1950a, b; Perry, 1950), have had the spectively, and disposed most of the Old World greatest influence on the taxonomy of Eugenia s. I. segregates listed above in the Plinioideae, a few and its Old World segregates. These authors ac­ in his subfamilies Acmenoideae and Myrtoideae. cepted the very large Old World genus Syzygium Melchior (1964) is in essential agreement, but he (including most notably Jambosa and Caryophyl­ treated Kausel's subfamilies as tribes. Ius, plus some 14 other synonyms listed in their Only fairly recently have comparative anatom­ 1939 work). In addition, they reinstated the small ical studies entered into taxonomic consideration. Old World genera Acmena [including Lomastelma Although the differences between Syzygium and Rafinesque (1838) and Xenodendron Lauterbach Eugenia may be somewhat obscure organograph­ et K. Schumann (1900)-see Merrill and Perry, ically, anatomically these genera appear to be 1938a] and Cleistocalyx Blume (1849) [including quite distinct. As discussed below, previous work Acicalyptus A. Gray (1854)-see Merrill, 1939; on wood structure (Chattaway, 1959; Dadswell Merrill and Perry, 1937]. Perry (1950) also ac­ and Ingle, 1947; Ingle and Dadswell, 1953), bark cepted the New Caledonian Piliocalyx Brongniart anatomy (Chattaway, 1959), and palynology et Gris (1865). Merrill (1937, 1950a, b), but not (Pike, 1956) strongly supports the distinction Perry (personal communication, 1970), accepted between the New World species of Eugenia s. I. Aphanomyrtus Miquel (1855) [including Pseudo­ and most of the Old World taxa which are largely eugenia Scortechini (1885)] and Iossinia. Iossinia referable to Syzygium s. I. represents a small residue of Old World species The reproductive anatomy of Eugenia s. I., more closely resembling certain of the American however, has not been investigated in a compara­ species of Eugenia than the Old World species of tive manner, but it seemed reasonable to expect Syzygium (Diels, 1922; Merrill, 1950b) and thus that such a study, particularly of the flowers, has been included in Eugenia as the series Iossinia would be of systematic value and might yield in the subgenus Eueugenia (Niedenzu, 1893).· evidence paralleling that from previous compara­ Eugenia, as so defined, "would scarcely appear tive studies. In fact, it soon became apparent in in the Old World, except for a few introduced the initial stages of this research (Schmid, 1970) species" (Merrill, 1950a, p. 329). Late in his that there are indeed essential anatomical differ­ career Merrill (19 50b), for reasons not stated, ences in the between the American species seemed on the verge of accepting additional seg­ of Eugenia s. I. and most of its extra-American regates of Syzygium: Tetraeugenia Merrill (1917), Pareugenia Turrill (1915), and even Jambosa taxa. and Caryophyllus, although earlier he and Perry had considered these taxa (except Tetraeugenia) MATERIALS AND METHODs-Methodology, to be congeneric with Syzygium. voucher information, and detailed supportive data Most workers, other than those proposing the for many of the statements below are given in various segregates, have been reluctant to accept Schmid (1971; in press b, c). Thirty-one taxa the majority of Old and New World segregate of Eugenia s. I. were examined. These may be genera of Eugenia s. I., although probably more conveniently arranged under the following desig­ American segregates, proposed mainly by Berg nations: April, 1972] SCHMlD-EUGENIA-SYZYGIUM CONTROVERSY 425

1) Eugenia s. s.-mainly New World, but in­ Perry (1939) (5 of 103 in their 1942a work), cluding the Old World segregate lossinia: Eugenia but 76 of 93 species of Eugenia s. s. in McVaugh acapulcensis Steud., E. aeruginea DC., E. biilora (1969) (35 of 46 in his 1958 work and 40 of (L.) DC., E. capuli (Schlecht. et Cham.) Berg, 47 in his 1963b work) have persistent bracteoles. E. cartagensis Berg, E. cojieiiolia DC., E. confusa DC., E. duchassaingiana Berg, E. ilavescens DC., Size of buds-Floral buds near the stage of E. florida DC., E. gregii (Sw.) Poir., E. mande­ anthesis range from very small to gigantic, being villensis Urban, E. muricata DC., E. oerstedeana up to 3 em long in (Me­ Berg, E. pleurocarpa StandI., E. salamensis Donn. Vaugh, 1963b). Sm., E. tikalana Lundell, E. (= Stenocalyx) uni­ flora L., E. venezuelensis Berg, E. whytei Sprague Pseudopedicel-Many species have the lower in Stapf, E. winzerlingii StandI., Iossinia aherniana portion of the inferior ovary gradually narrowed (C. B. Rob.) Merr., and 1. palumbis (Merr.) into a slender pedicel-like base (Fig. 24, upper Diels; right). I designate this a "pseudopedicel" since 2) Syzygium s. I.-exclusively Old World, in­ it resembles a pedicel, although actually it is part cluding such segregates as Acicalyptus, Acmena, of the flower. I believe my term is more descrip­ Caryophyllus, Cleistocalyx, and lambosa: Acmena tively restrictive than the entirely comparable smithii (Poir.) Merr. et Perry, Cleistocalyx (= "pseudostalk" of Henderson (1949) or the Acicalyptus) myrtoides (A. Gray) Merr. et Perry, "pseudostipe" of Wilson (1957). C. operculatus (Roxb.) Merr. et Perry, Svzygium The pseudopedicel can only be designated on aromaticum (L.) Merr. et Perry, S. cumini (L.) a topological basis. Its proximal part is anatom­ Skeels, S. jambos (L.) Alston, S. malaccense (L.) ically indistinguishable from the distal part of the Merr. et Perry, and S. paniculatum Gaertn. pedicel, except, of course, where traces depart to the bracteoles. Thus, proximally the pseudo­ ORGANOGRAPHY-The following description of pedicel is well separated from the pedicel only Eugenia s. I. (sensu McVaugh, 1968, and also in­ by the bracteoles and articulation (if present), cluding Syzygium s. I.) is based partly on standard but distally it merges into the rest of the flower. taxonomic works (see Bentham, 1869; Gagne­ Most species of Syzygium s. I. have a pseudo­ pain, 1917). Reference should also be made to pedicel according to Henderson (1949), who re­ Fig. 24 and Table 1 for characters contrasting ported one for 113 of 122 species. Pseudopedicels Eugenia s. s. and Syzygium s. l. Details of in­ are very rare in Eugenia s. s. (McVaugh, personal florescence structure are relegated to Fig. 24. communication, 1971), but one is present in E. stipitata (McVaugh, 1958). Pubescence-i-Syzygium s. I. is almost entirely glabrous (Henderson, 1949; Merrill, 1950b), Floral tube6- The floral tube of Syzygium s. 1. whereas Eugenia s. s. is predominantly pubescent. is often much prolonged beyond the summit of For example, only five of the 139 species of the ovary (e.g., Acmena smithii) or little or not Syzygium s. I. described in Henderson (1949) so (e.g., S. aromaticum). The floral tube of but 58 of 60 species of Eugenia s. s. in McVaugh Eugenia s. s. is little or not at all prolonged beyond (1958)-47 of 54 in his 1963b work-are pu­ the ovary (e.g., up to 1.1 mm in E. salamensis bescent. Some of the so-called "glabrous" species var. hiraeiiolia), American species of the Eu­ of Eugenia s. s. are actually very sparingly pubes­ geniinae with a prolonged floral tube are usually cent. referred to , , or Myr­ Pedicels and bracteoles-The flowers are ses­ ciaria. sile, subsessile, or usually pedicellate. An articu­ lation generally separates the pedicel (if present) Parianth parts7-The calyx and corolla are con­ and flower. McVaugh (l963b) reported pedicels sistently tetramerous in Eugenia s. s. (McVaugh, up to 20 mm long for Eugenia aeruginea and E. 1958, 1963b, 1968, 1969) but occasionally pen- winzerlingii, but most species he described in this 6 In Schmid (1971) this character was defined in a work have pedicels less than 5 mm long. Two taxonomic sense. However, in some cases the top of the (sometimes 4, Merrill and Perry, 1939) bracteoles, ovary may be depressed and the peripheral parts of the ovary may thus line part or all of the so-called "floral situated opposite the ends of the septum and tube." Therefore, in such cases, which would generally above the articulation, subtend each flower. In not be apparent from macroscopic examination, part of Syzygium s. I. bracteoles (and bracts) are mostly the so-called "prolongation of the floral tube beyond the inconspicuous and very fugacious (Henderson, ovary" may actually include some ovarian tissue. 7 I use the terms "" and "" according to 1949; Perry, personal communication, 1971), taxonomic convention and, hence, without morphological whereas in Eugenia s. s. they are generally con­ connotation. Morphologically, if one adheres to the ap­ spicuous and persistent (McVaugh, 1963b) but pendicular concept of the inferior ovary (Eames, 1961), sometimes deciduous at or before anthesis (e.g., the petals and sepals extend to the base of the epigynous flower, and distally " lobes" and " lobes" would E. salamensis, E. uniflora). For example, five of be present rather than the "petals" and "sepals" desig­ 150 species of Syzygium s. I. listed in Merrill and nated by taxonomists. 426 AMERICAN JOURNAL OF BOTANY [Vol. 59 tamerous (or other) in Syzygium s. I. Jayaweera Nectaries-These are rarely described in the (1957), for example, found 95.5 % of 3000 taxonomic literature. Brown (1938) rightly em­ flowers of S. malaccense to be tetramerous for phasized that nectaries should be studied in living both calyx and corolla. Two sepals occur at points material because they become greatly distorted in opposite the ends of the septum, the other two dried specimens. An ovarian annular type of nec­ opposite the dorsal positions of the carpels. The tary is present in Syzygium aromaticum, S. jam­ petals alternate with the sepals. The perianth bas, S. malaccense, and S. paniculatum (Schmid, parts, especially the petals, are usually conspicu­ in press b). ous, but sometimes minute (e.g., Acmena smithii), particularly in Syzygium s. I. The calyx lobes are Ovary-The ovary is inferior and very con­ distinct and usually persistent in Eugenia s. s. sistently bilocular. Some species occasionally have Those of Syrygium s. I. are obscure and deciduous trilocular flowers (e.g., Eugenia coniusa, E. win­ to distinct and persistent; these extremes represent zerlingii, Acmena smithii, Cleistocalyx operculatus, a rather shaky generic distinction between Syzy­ Syzygium malaccense) or very rarely unilocular gium s. s. and lambosa. (e.g., Iossinia aherniana) or multilocular ones In Syzygium s. I. the perianth is often calyp­ (e.g., quadrilocular in E. stipitata, McVaugh, trate. Half (69 corolline versus 1 calycine) of 1958). A prominent compitum, the canal or the species listed in Henderson (1949) have space connecting two or more ovarian loculi (Carr calyptrae. Those few species of Syzygium s. I. with and Carr, 1961), is usually present. Rastogi (1951) calycine calyptrae are sometimes segregated as confused the compitum with the phyletic begin­ Acicalyptus, Cleistocalyx, or Piliocalyx. lambosa ning of true parietal placentation. In Syzygium has free petals whereas Syzygium s. s. has the s. l. the ovary is often spongy, with tissue con­ petals coherent into calyptrae, but this presumed taining large intercellular spaces. This condition generic distinction also frequently breaks down. apparently does not occur in Eugenia s. s. Calyptrate species of Eugenia s. s. do not occur since such would invariably be referred to Caly­ Ovules and seeds-The anatropous, usually corectes. bitegmic (Davis, 1966; Mauritzon, 1939) ovules are several to many per locule, rarely as few as -Most species of Eugenia s. I. have two per locule (e.g., Eugenia coffeifolia). Placen­ numerous stamens, as many as 500 reported for tation is axile. Only one or two ovules (very the Peruvian E. scalariiormis (McVaugh, 1958). rarely more) per ovary develop into the often Those few species of Syzygium s. I. with 4-8 quite large seeds. Cotyledons are usually fused stamens are occasionally segregated as Aphano­ (pseudomonocotyledonous) in Eugenia s. s. but (Amshoff, 1963; Kausel, 1957b; Merrill, usually divided in Syzvgium s. I.-although fused 1937, 1950b) or as the probably closely related in Acmena. Tetraeugenia with four stamens (Merrill, 1917). Stamens of Eugenia s. I. are usually free. How­ Style-The apex is usually simple, very rarely ever, Pareugenia, a segregate of Syzygium s. I., bifid (e.g., the African Eugenia aschersoniana, E. was proposed for those forms with united stamens. mossambicensis, Amshoff, 1958). The style is Since there are intermediate forms linking all these frequently quite long and thus usually doubled segregates with other species in Syzygium s. I., back in the bud. these genera have generally been rejected. The stamens are borne on a flat disc surround­ HISTOLOGY-Secretory cavities-Secretory cav­ ing the base of the style or on the margins or ities ("oil glands," "oil cavities," "secretion res­ inner surface of the floral tube (if this is present). ervoirs," "oil cells," etc.) are a characteristic Syzygium s. s. is supposed to be distinguishable feature of the Myrtaceae and are responsible for from I ambosa on the basis of an inconspicuous the well-known aromatic properties of its taxa. versus conspicuous staminal disc, respectively, but Vertical secretory ducts occur in a few genera many exceptions occur. (e.g., , , and Spermolepis According to Merrill and Perry (1938b, c, of the subfamily Leptospermoideae-Carr and 1939). stamens of Eugenia s. s. are much less in­ Carr, 1969; Ingle and Dadswell, 1953). curved in bud than those of SYZYRium s. I. Anther sacs are mostly parallel and longitu­ The secretory cavities of Eugenia s. I. are dinally dehiscent but divaricate and terminally spherical to ovoid and achieve a maximum di­ dehiscent in the Old World segregate Acmena. mension of about 350 p. in flowers of E. mande­ Anther connectives each with a terminal secre­ villensis and Syzygium malaccense. Secretory tory cavity are considered typical of the Myrtaceae cavities in the Myrtaceae have been described as (Erdtman and Metcalfe, 1963; Weberling, 1966). being schizogenous, oblitoschizogenous, schizoly­ However, the anthers of a few species (e.g., Eu­ sigenous, and lysigenous (Carr and Carr, 1970). genia flavescens, Iossinia aherniana, Cleistocalyx However, a careful reinvestigation with new tech­ myrtoides) lack secretory cavities. niques is in order since these differences in inter- April, 1972] SCHMID-EUGENIA-SYZYGIUM CONTROVERSY 427

®

9-13:­ ~ 1-8:~ ® ® ® 14, 15:' all 50,u Fig. 1-15. Various types of trichomes (hairs in Fig. 1-13, papillae in Fig. 14, 15) on floral organs of Eugenia s. l. Trichomes (except Fig. 6, 8, 14, 15) oriented so that distal end of flower or appendage is toward top of page. -Fig. 1,2. E. capuli, floral tube (Fig. 1, Gentle 5044, MICH), sepal (Fig. 2, Breedlove 14407, MICH).-Fig. 3-5. E. florida (Steyermark & Gibson 95781, MICH), floral tube (Fig. 3,4), margin of sepal (Fig. 5).-Fig. 6. lossinia aherniana (Velasco s. n., US), staminal disc, all hairs from an area less than 1 mm'i-s-Fig. 7. E. muricata (Pires et al. 50887, MICH), floral tube.-Fig. 8. E. gregii (Hahn 641, MICH), floral tube; side, bottom, and top views of hairs respectively.-Fig. 9. E. mandevillensis (Proctor 19686, MICH), staminal disc.-Fig. 10, 11. E. pleurocarpa (McVaugh 15322, MICH), sepal (Fig. 10), floral tube (Fig. ll).-Fig. 12. E. bijlora (Steyermark & Wurdack 31, MICH), floral tube.-Fig. 13. E. salamensis var, salamensis (Carter & Chisaki 1209, MICH), floral tube.-Fig. 14, 15. Acmena smithii (Schmid 1968-A2, MICH), transection of petal (Fig. 14), showing papillae with tannin globules and cuticle (cu); transection of filament (Fig. 15), showing papillae filled with tannin. pretation seem partly due to technical difficulties Trichomes-A variety of trichomes is en­ in tissue preparation (Carr and Carr, 1970). countered in Eugenia s. 1. (Fig. 1-15). Usually A with a terminal secretory cavity is the hairs are simple (Fig. 1-3,5,9, 10) or weakly characteristic of the Myrtaceae, as noted above. to strongly dibrachiate (malpighian) (Fig. 4, 6, At the species or series level, the distribution of 7, 11-13). Funnel-shaped (infundibuliform) hairs oil cavities has been considered to be of taxonomic (Fig. 8) occur in E. gregii. Many hairs are at­ significance in vegetative structures of Eucalyptus tached by a small peg (Fig. 4, 6-8, 12, 13), but (Carr and Carr, 1969; Welch, 1920). However, others have the large basal part of the cell em­ this seems not to be the case in the Myrtaceae I bedded in the epidermis (Fig. 1, 2, 5). The hairs examined. The distribution and number of secre­ on a often vary considerably. Those de­ tory cavities is quite variable, even between dif­ picted in Fig. 6 were found within an area less ferent buds from the same plant, and thus of little than one mm-. Much of the apparent variability use taxonomically except in the few cases where is presumably due to different ontogenetic stages. cavities occur in unusual places (e.g., adjacent to The hairs of the Myrtaceae are reported to be the loculi in Eugenia mandevillensis). exclusively unicellular (Hummel and Staesche, Secretory cavities of various species of Eugenia 1962; Metcalfe and Chalk, 1950). I found no s. 1. are not distinguishable from one another. evidence to contradict this (Fig. 1-13). Papillae Winton and Winton (1939) came to the same (Fig. 14, 15), however, may be multicellular, as conclusion after examining various cultivars of on the petals of Acmena smithii (Fig. 14). Hairs the Myrtoideae. of various myrtaceous genera have been described 428 AMERICAN JOURNAL OF BOTANY [Vol. 59

as bicellular, but actually there is a single proto­ plast which withdraws into the lower part of the • cell, the upper part being shed after a septum • forms between the two parts (Uphof, 1962). .:.~.,~.... As in most species of American Myrtaceae o •• (McVaugh, 1968), in Eugenia the initial pubes­ •••. .. cence usually disappears soon after flowering • . (McVaugh, 1963a, b). Presumably the distal . . .., parts of the hairs break off by the mechanism just .. : . noted. In addition, at least in herbarium material, • • those hairs attached merely by small pegs (Fig. 4, 6-8, 12, 13) seem especially susceptible to breakage. • •• The presence or absence of pubescence, its • type, and its distribution afford useful taxonomic distinctions both in the American Myrtaceae and in Eugenia s. s. (McVaugh, 1963a, b, 1968). Svzygium s. l. is mostly glabrous whereas Eugenia s. s. is mostly pubescent. Tannins-The Myrtaceae are characteristically tanniferous (Metcalfe and Chalk, 1950). The flowers of Syzygium s. l. are moderately to very Fig. 16-19. Variation in bundle arrangements.-Fig. tanniferous, those of Eugenia s. s. usually only 16. Polycyclic, with two or more concentric rings of slightly so. bundles.-Fig. 17, 18. Monocyclic, with a single ring Distribution of tannin in my material does not of bundles; Fig. 17 with a definite number; Fig. 18 with an indefinite number.-Fig. 19. Zonocyclic, with many seem to be taxonomically significant. In a number bundles dispersed in a banded ring, definite concentric of species (e.g., Acmena smithii, Syzygium panic­ rings of strands as in Fig. 16 not apparent. ulatum) much tannin occurs in the epidermal and subepidermal layers and particularly in the sta­ mens and style. It is tempting to relate this dis­ flowers of 25 species of Myrtaceae, in which he tribution of tannin to a protective function against found three patterns (excluding the seed): (1 ) snails, insects, and pathogens, as presumed by sclereids present in flowers, and sometimes also earlier botanists (see Eyde, 1966, for references in the buds; (2) sclereids present only after an­ and a detailed discussion of the occurrence of thesis, during fruit maturation; (3) sclereids ab­ tannin in flowers). It would seem particularly sent from all reproductive stages. important for a plant to evolve protective devices I found sclerenchyma (sclereids and/or espe­ for its fertile parts. Tannin also frequently occurs cially phloem fibers) only in flowers of Syzygium in phloem parenchyma cells (significance: pro­ s. l. In the larger flowers sclerenchyma, especially tection against aphids?). fibers, occurs throughout the floral tube and ovary; in several cases it occurs even in the sepals but Crystals-Numerous druses, but relatively few never in the petals, stamens, or style. This dis­ prismatic crystals, usually occur in some or all of tribution suggests that evolution of sclerenchyma the floral parts, depending upon the species and in certain species might have paralleled the evolu­ perhaps even the individual. Raphides are absent. tion of zonocycly, which seems to have resulted Anthers, in particular, often contain dense clusters . as numbers of stamens increased as an adaptation of druses (e.g., Acmena smithii, Syzygium aro­ to bird pollination (Schmid, 1971). maticum) . Matthews and Knox (1926) and Namikawa (1919) suspected that the calcium VASCULATURE-There is so much variability oxalate of crystals might play a role in causing in floral vasculature, both between different spe­ dehiscence of the anther by breaking down cells cies and within a single species (even between in the connective. However, the fact that anthers different flowers of the same plant), that general­ of some species (e.g., S. jambos) apparently lack ized descriptions of vasculature were not prepared. crystals militates against this assumption, at least Nevertheless, the following features of vasculature for the group I studied. seem salient. Sclerenchyma-Pass (1940) discussed in con­ Floral tube-The major vascular bundles of the siderable detail the distribution of sclereids in floral tube exhibit the following patterns in the Myrtaceae (Schmid, 1971; in press b, c): 8 In keeping with botanical precedent, I consider the darkly staining cells or parts of cells to be tanniferous, (1) Monocyclic (not "unicyclic," which is a although no attempt was made at chemical identification. Latin-Greek hybrid): with a single ring of bundles April, 1972] SCHMID-EUGENIA-SYZYGIUM CONTROVERSY 429

cies of Eugenia s. s. exhibit irregular patterns. A regular relationship clearly seems related to mono­ cycly with a definite number of bundles (Fig. 17) (Schmid, 1971). Stamens-A single strand, sometimes only one or two vessel elements thick (e.g., Eugenia vene­ zuelensis),. supplies each stamen. Distally the strand typI~ally expands, often considerably, in the connective, where it frequently abuts against a large terminally situated secretory cavity. The shape of this distal expansion may be a poten­ tially useful taxonomic character. . Ovular supply-The species of Eugenia s. l. display two pathways of vasculature supplying the ovules, transeptal and axile (Fig. 20-23). In the transeptal. ovular system, the vascular supply to the ovules IS from the sides, via the peripheries of the septa (Fig. 22), and there is no vascular tissue in the center of the ovary below the pla­ centae (Fig. 22, 23). By contrast, in the axile Fig. 20-23. Generalized diagrams of axile (Fig. 20, ~)Vular system, the vascular supply to the ovules 21) and transeptal (Fig. 22, 23) ovular systems in Myr­ IS through the center of the gynoecium, via the taceae. Arrows at Fig. 20 and 22 indicate approximate bases of the septa (Fig. 20), and all transections levels of transections shown in Fig. 21 and 23, respec­ of the lower parts of the ovary will reveal vascular tively. tissue in the center (Fig. 20, 21). The axile sup­ ply occurs in Syzygium s. 1. whereas the transeptal (Fig. 17, 18); e.g., Acmena smithii, Cleistocalyx one occurs in Eugenia s. s. operculatus, all species of Eugenia s. s. and Jos­ I purposely avoided designating bundles re­ sinia. lated to the ovular supply as "ventral carpellary (2) Polycyclic: with two or more concentric bundles" since I am uncertain of the homology rings of bundles (Fig. 16); e.g., Eucalyptus (Carr, between ventral bundles in other families and Carr, and Milkovits, 1970), also Alangium of the those bundles of Eugenia s. 1. comprising the axile Alangiaceae (Eyde, 1968). Bundles mayor may and transeptal systems. not occur along definite radii. I use the adjective "axile" rather than "axia1." (3) Zonocyclic (Greek zone, meaning belt !h~ f~,rm~r is w~th~ut J?orphological implications, girdle, zone): with many bundles dispersed in ~ axile Simply indicating a central position for banded ring, definite concentric rings of strands the vascular supply to the ovules (as opposed to not apparent (Fig. 19); e.g., Cleistocalyx myr­ a lateral or transeptal one); in contrast, "axial" toides, all species of Syzygium s. s. could be construed as meaning that the placentae may be cauline in nature (see Jackson 1928). In Only types 1 and 3 occur in Eugenia s. 1., but decr~e the~e are intermediate types. For example, spite of Eames' (1961, p. 235) that "no sv­ ar~ zygtum aromaticum is monocyclic proximally, placentae morphologically cauline," whether zonocyclic distally (Schmid, 1971, in press b). carpels, particularly their placentae, are of an Two variations occur in the monocyclic pattern: appendicular (foliar) or axial (cauline ) nature (a) .usually eight major bundles present, as in all ~as by no means been conclusively decided. Moe­ species of Eugenia s. s. and Iossinia (Fig. 17); hono (1970), for example, presents detailed argu­ and (b) an indefinite number of bundles present, ments for an axial interpretation of placentae in as in Acmena smithii and Cleistocalyx operculatus the Caryophyllales. (Fig. 18). It is best to apply these terms to regions of the Style-In many species of Eugenia s. 1. the flower in the vicinity of the loculi since all zono­ number of stylar bundles as seen in transection cyclic flowers are monocyclic in their very base ~ncreases toward the stylar apex. This, of course, before the strands have divided. ' IS partly due to branching of bundles. However, o~ten there are distally occurring stylar bundles Floral tube versus perianth parts-Many species Without any basal connections. Possibly such of Eugenia s. s. (e.g., E. biflora) display a very "free-floating" strands result from the rapid ex­ regular relationship between the major bundles pansion of the style. Fusions between stylar of the floral tube and the vasculature to the bundles frequently occur. perianth members. Syzygium s. 1. and other spe- The bundles often expand near the apex of the 430 AMERICAN JOURNAL OF BOTANY [Vol. 59 style, generally forming two arcs or even a con­ duced species have skewed the extent of their tinuous ring of vascular tissue. Arber (1931, p. differences-e.g., the New World E. uniilora and 36), incredibly, believed a similar situation in the Old World S. cumini, S. jambos, S. mataccense, the Cruciferae to be "connected with the hin­ and S. panicutatum obviously do not adequately drance to further flow of sap when the distal reflect the weltering procession of the many other region of the organ is approached." Melville taxa in their hemispheres. (1962) discussed such stylar vascular expansions Perhaps largely for these reasons, the distin­ ("brushes") in terms of a "dichotomous branch guishing characters used by Merrill and Perry system" and invoked an intricate hypothesis in (1937, 1938a-c, 1939, 1942a, b), Merrill (1937, support of his gonophyll theory. 1939, 1950a, b), Perry (1950), and other authors, Rastogi (1951) found five stylar bundles in have been rejected as inconsistent by more con­ Eugenia jambotana (= ) and in­ servative botanists (Airy Shaw, 1949; Bailey, terpreted these as indicating a pentacarpellary 1930, but not 1949; Henderson, 1949; Merrill, nature for the stylar region, with three of the initially in 1917; Wilson, 1957, but not 1960), carpels in the presently bicarpellary ovary having some of whom have worked in different geograph­ been lost. By Rastogi's (1951) reasoning the 5­ ical areas and have reverted to the Bentham and 13 bundles in the base of the style of Acmena Hooker (1862-67; Bentham, 1869) concept of a smithii would necessitate a very elaborate phy­ "super" genus Eugenia to include everything. logeny. In recent times Merrill and Perry, by virtue of In contradistinction to these philosophers, I the extended series of papers just cited, have had believe the branching and distal expansion of the greatest influence on the taxonomy of Eugenia stylar bundles to be functionally related to the s. t. and its Old World segregates. Departing from secretion of stigmatic fluid and to the concomitant most previous workers by placing greater emphasis receptivity of the stigma for pollen. on fruit rather than on floral characters, Merrill and Perry (1938b, c, 1939) suggested that the Other features-Other features of the vascula­ Old and New World species could be effectively ture of Eugenia s. l. are even more variable. distinguished as follows (descriptive terms from Generalization thus is difficult. The nature of their 1939 paper): some of this variability is evident from the de­ Syzygium: (1) dried fruit not too easily broken; tailed descriptions in Schmid (1971; in press b, c). (2) testa roughish and more or less loosely ad­ The considerable variability in level and manner hering to the pericarp; (3) cotyledons of embryo of origin of the dorsal carpellary bundles of distinct, usually attached near the middle of the Syzygium s. t. is detailed in Schmid (in press a). opposing faces, concealing the hypocotyl within; (4) floral tube ("calyx limb") prolonged beyond DISCUSSION-It is evident from the nomen­ the ovary; (5) stamens much more incurved in clatural and taxonomic history outlined above bud than in Eugenia; (6) inflorescence largely that there is a plethora of discordant opinion re­ centrifugal, with the panicles branching by threes garding the validity and disposition of the various or with secondary cymes. Old and New World segregates of Eugenia s. t. Eugenia: (1) pericarp easily crushed, thinner The Old World segregate genera have perhaps than in most Old World taxa; (2) testa smooth, been more controversial than the New World ones chartaceous to cartilaginous, and mostly lustrous, or the former, at least, have received the greater free from the pericarp (and adherent to the cotyle­ attention. dons); (3) cotyledons united, the embryo un­ It is perhaps tempting on the basis of phyto­ divided and thus pseudomonocotyledonous; (4) geographical considerations alone to accept the floral tube less prolonged than in Syzygium; (5) division of Eugenia s. t. into at least two genera: stamens much less incurved in bud; (6) inflores­ the mainly American Eugenia and the exclusively cence centripetal, with one-flowered pedicels (or Old World Syzygium. However, obtaining valid peduncles) solitary, clustered, or in a short organographic justification for their taxonomic raceme. recognition has been extremely difficult because Merrill and Perry regarded characters (2) and of their immensity and diversity. Authors wishing (3) as "the basic distinctions between the two to split up Eugenia s. t. have, of course, proposed genera" (l938c, p. 206), but they also stressed various criteria that they believed effectively (1939) that both floral and fruit characters are separated these and additional groups, but un­ essential for the elucidation of Eugenia s. l. fortunately these distinctions have never been Subsequent workers (Amshoff, 1958, 1963; made on a worldwide basis. In fact, there exist Bailey, 1949; Kausel, 1956, 1957b, 1966; Mel­ no monographs on Eugenia s. t. or even Syzygium chior, 1964) have generally accepted these dis­ s. t. other than regional ones. Examination of the tinguishing characters, particularly that of the keys authors have used to separate Eugenia, divided versus undivided cotyledons. In fact, Syzygium, and their respective segregates can be Kausel (1956, 1957b, 1966), a splitter par ex­ very misleading because a few commonly intro- cellence, used only this character of the embryo April, 1972] SCHMID-EUGENIA-SYZYGIUM CONTROVERSY 431

NEW WORLD OLD WORLD

EUGEN IA S. S.

.' ...... \ .. Jossinla ·~r~m:~~~}:

••...... \...

SYZYGIUM S. L. Sect: Acicalyptus Cleistocalyx Syzygium (lnc, Caryophyllus and Jambosa) -----

CHARACTER EUGENIA S. S. SYZYGIUM S. L. "PUBESCENCE VEG. AND/OR REPRO. PARTS MOST SPECIES (ca 90%) VERY FEW SPECIES (ca 5%) "POSITION OF THE INFLORESCENCE PREDOMINANTLY LATERAL USUALLY TERMINAL MATURATION FLOWERS ON INFLORESCENCE CENTRIPETAL CENTRIFUGAL "STRUCTURE OF THE INFLORESCENCE SOLITARY FLOWERS OR SOLITARY FLOWERS OR trsu­ USUALLY RACEMES ALLY PANICLES OR CYMES "BRACTEOLES SUBTENDING FLOWERS MOSTLY PERSISTENT AND MOSTLY FUGACIOUS AND CONSPICUOUS (ca 80%) INCONSPICUOUS (ca 95 %) SIZE OF BUD NEAR STAGE OF ANTHESIS VERY SMALL TO MODERATE VERY SMALL TO VERY LARGE "PSEUDOPEDICEL (ATTENUATE FL. BASE) RARELY PRESENT (ca 1%) MOSTLY PRESENT (ca 90%) PROLONGATION FLORAL TUBE ABOVE OVARY NONE TO SLIGHT NONE TO MUCH "VARIATION IN NUMBER PERIANTH MEMBERS CONSTANT (4+4) RATHER VARIABLE (4-5 +4-5) MAGNITUDE OF THE PERIANTH PARTS USUALLY LARGE AND MINUTE TO LARGE AND CONSPICUOUS CONSPICUOUS DURATION OF THE LOBES OF CALYX USUALLY PERSISTENT DECIDUOUS OR PERSISTENT DISTINCTNESS OF THE LOBES OF CALYX DISTINCT OBSCURE TO DISTINCT "COROLLINE OR CALYCINE CALYPTRAE NEVER OFTEN (ca. 50%), MOSTLY COROLLINE INCURVATION OF STAMENS IN THE BUD LESSER GREATER POSITION OF LOCULI IN THE OVARY MIDDLE NEAR TOP "DISCRETENESS OF THE COTYLEDONS USUALLY FUSED USUALLY DISTINCT SURFACE OF THE SEED-COAT SMOOTH ROUGH

AMOUNT SCLERENCHYMA IN THE FLOWER NONE NONE TO MUCH TEXTURE OF THE OVARIAN TISSUE COMPACT USUALLY SPONGY AMOUNT OF TANNIN IN THE FLOWER USUALLY SLIGHT MODERATE TO VERY MUCH ARCHITECTURE OF VASCULAR SYSTEM CLOSED OPEN TO CLOSED "LARGE BUNDLES OF THE FLORAL TUBE MONOCYCLIC MONO· TO MOSTLY ZONO· CYCLIC NUMBER BUNDLES IF PATTERN MONOCYCLIC DEFINITE, USUALLY 8 INDEFINITE FLORAL TUBE BUNDLES AND PERIANTH PARTS OFT REGULAR RELATIONSHIP IRREGULAR RELATIONSHIP "PATHWAY OF VASCULAR SUPPLY TO OVULES TRANSEPTAL AXILE CONNECTIONS OF THE DORSAL BUNDLES FREQUENT TO OTHER NOT UNITED WITH OTHER CARPELLARY BUNDLES CARPELLARY BUNDLES Fig. 24. Diagram and table of organographical and anatomical differences between Eugenia s. s. and Syzygium s. I. Tabulated characters elaborated in text; those deemed most significant marked with asterisk. Other important differences occur in bark, wood, and pollen (see text). Diagrams of flowers depict patterns of bundles in floral tube (transections) and ovular vascular supply (longisections, the plane of septum facing viewer). Taxa arranged to show suggested taxonomy. 432 AMERICAN JOURNAL OF BOTANY [Vol. 59

TABLE 1. Quantification of several generic diiierences' an adventive, perhaps the surest single distinguish­ ing feature indicated in Fig. 24 is where the plant Character Eugenia s. s. Svzygium s. 1. lives-by itself, however, hardly a valid basis for Pubescence present on generic distinction. vegetative and (or) Many of the organographic characters listed in reproductive parts ca90% ca 5%} Fig. 24 have been emphasized previously, but a Bracteoles persistent ca80% ca 5 % of species few, particularly those concerning pubescence, Pseudopedicel present ca 1 % ca90% bracteoles, and pseudopedicels, deserve broader appreciation since their specific aspect is quite a Compiled from data in Henderson (1949), McVaugh diagnostic (Table 1). For example, Henderson (1958, 1963b, 1969), Merrill and Perry (1939, 1942a), and Perry (1950). Actual values rounded off to nearest (1949, p. 14) stated that "the great majority of 5 % and biased toward similarity between the taxa. species [of Syzygium s. l.] are glabrous in all their parts," but, as shown above, most species of Eugenia s. s. have pubescent vegetative and/or to place Eugenia and Svzygium in separate sub­ reproductive parts. However, Merrill (1950b, p. families. 358), in comparing the sparingly pubescent Jos­ However, Henderson (1949), Wilson (1957), sinia with the "characteristically entirely glabrous" and McVaugh (1958, 1963b) in essential agree­ Syzygium, is the only worker I know who has ment with these two workers, rejected the segre­ utilized this rather obvious distinction. Secondly, gation of Syzygium from Eugenia since they found the nature of the bracteoles as a diagnostic feature, the degree of fusion of the cotyledons and the mostly persistent in Eugenia s. s. (McVaugh, extent of adherence of the testa to the pericarp 1963b) but mostly fugacious in Syzygium s. l. to be quite variable. The former character varied (Henderson, 1949; Perry, personal communica­ even within a single species; the latter varied tion, 1971), also seems to have escaped the at­ depending upon whether living, dried, or dried­ tention of most taxonomists. Only McVaugh and-boiled material was examined. In addition, (1963b) used this feature in a key to these taxa. Amshoff (1958) noted that the African members Finally, the usual presence of a pseudopedicel in of Eugenia s. s., which Merrill (1950a, b) would Syzygium s. l. (Henderson, 1949) but its general have included in Jossinia, resemble the American absence in Eugenia s. s. (McVaugh, personal com­ species in having a pseudomonocotyledonous munication, 1971) has been used as a distinguish­ embryo but differ in having the testa "probably ing character by only a few workers (Guillaumin, adhering to the pericarp," the presumed character 1938; Hosokawa, 1940; McVaugh, 1963b; Nie­ of Syzygium (Merrill and Perry, 1938b, c, 1939). denzu, 1893). All the evidence, however, indicates that the de­ In my opinion, the totality of organographic gree of union of the cotyledons, while somewhat characters tabulated in Fig. 24 (see also Table 1) variable within a taxon, may still be useful in is adequate to differentiate nearly all species of distinguishing between many Old and New World Eugenia s. s. and Syzygium s. l., particularly if the forms, although certainly by itself it is not a hitherto neglected features concerning pubescence, good generic character. In light of the studies of bracteoles, and pseudopedicels are emphasized. In Henderson (1949) and Wilson (1957), however, addition, both Henderson (1949) and Merrill and the extent of adherence of the testa to the peri­ Perry (1939) suggested that good distinguishing carp (or, conversely, to the cotyledons) seems to characters might be found in inflorescence struc­ be a worthless character. ture. Consequently, the over-all ensemble of Taxonomists have generally used various char­ organographic differences (Fig. 24) would seem acters of the inflorescence and flower, particularly sufficient by itself to warrant the taxonomic ac­ of the floral tube and perianth, in attempts to ceptance of these two groups. However, there is delimit Eugenia s. s. and Syzygium s. l. However, fairly extensive additional evidence from anatomy there are so many species with intermediate char­ and palynology now available, including new data acter states that one or a few character differences from floral anatomy presented here, that further should not be expected to separate these taxa. demonstrates the basic disparity of Eugenia and For example, the prolongation of the floral tube Syzygium. beyond the ovary, one of the most favored char­ Anatomically, the two genera appear quite dis­ acters of taxonomists, is absent in S. aromaticum tinct. Moll and Janssonius (1918) indicated that but ranges up to 1.1 mm in E. salamensis var. their arrangement of species on the basis of wood hiraeijolia, which is the converse of the presumed structure does not support the subdivision of generic distinction. Nevertheless, it seems that a Eugenia into Jambosa, Syzvgium, etc. However, multitude of characters might adequately sort out the more comprehensive studies of wood by Ingle either genus. Figure 24 presents a diagrammatic and Dadswell (1953; also the preliminary report representation and a tabulation of the differences by Dadswell and Ingle, 1947) reenforce the con­ between Eugenia s. s. and Syzygium s. l. If one cept that at least two genera are represented within can eliminate the possibility of a given plant being Eugenia s. l.: (1) "Eugenia A" (a designation April, 1972] SCHMID-EUGENIA-SYZYGIUM CONTROVERSY 433

of Ingle and Dadswell, 1953), including the New and organographical differences are largely cor­ World species plus several Old World species related with geographical separation in the Old possibly referable to Iossinia; and (2) the quite and New World strongly indicates that we are different "Eugenia B," including Acmena, Cleisto­ dealing with two long-separated lines of evolution. calyx, Syzygium, and a number of Australasian The transeptal type of ovular system occurs species described under Eugenia but probably re­ rather infrequently in angiosperms (Eyde, 1967; ferable to Syzygium because of similar wood Schmid, unpublished data). In the a (Ingle and Dadswell, 1953). Chattaway (1959) transeptal ovular supply is known only in the also studied wood of Eugenia s. 1., upheld the dis­ Oliniaceae (Rao and Dahlgren, 1969), Onagraceae tinction between "Eugenia A" and "B," and in­ (Baehni and Bonner, 1948, 1949; Bonner, 1948; cluded seven additional species in "Eugenia A" Carlquist and Raven, 1966; van Tieghem, 1875), and 12 in "B." Evidence from bark anatomy Punicaceae (Berg, 1857-59, pl. 8, 9; Sinha and (Chattaway, 1959) and palynology (Pike, 1956) Joshi, 1959; Tung, 1935), Thymelaeaceae (Heinig, also supports the separation of "Eugenia A" and 1951), and Myrtaceae. In the Myrtaceae most "Eugenia B." Unfortunately, the cytological evi­ genera of the subfamily Myrtoideae have a tran­ dence is so scanty and inconclusive that it is thus septal ovular supply (Schmid, 1970; unpublished). far of no help (see Fedorov, 1969; Moussel, 1965, However, only an axile system apparently occurs for literature). in the Leptospermoideae. I discovered an axile My studies (Schmid, 1971) show that there system in Eucalyptus micrantha, are also some profound anatomical differences in scoparium, and Tepualia stipularis (Schmid, 1970). the flower between the mainly American species In addition, published figures indicate only the of Eugenia s. 1. (Ingle and Dadswell's "Eugenia axile supply to be present in the following Lepto­ A") and most of the extra-American species spermoideae: Angophora (Petit, 1908), ("Eugenia B"). These differences are listed in (Rastogi, 1951), (Polunina, 1958), Fig. 24. (Petit, 1908), Eucalyptus (Bac­ Syzygium s. 1. has an axile ovular system whereas carini, 1884; Carr et al., 1970; Niedenzu, 1893; Eugenia s. s. has a transeptal one (Fig. 20-24). Petit, 1908; Zucconi, 1958), Leptospermum I consider this character to be a basic one indi­ (Rastogi, 1951), (Baccarini, 1884; cating the generic discreteness of these taxa. In Leins, 1965), and (Baccarini, 1884). fact, Eyde (1967, p. 177), who discussed the The ovular system in series Corymbosae of Euca­ significance of these types of ovular supply and lyptus (Carr et al., 1970) may be somewhat in­ their occurrence in various angiospermous families; termediate between the axile and transeptal types, regarded the transeptal ovular system to be "a but the very clear diagrams of Eucalyptus in the family character of [the] Cornaceae" and used it little-known paper by Baccarini (1884) show a to exclude such taxonomically puzzling genera as very definite axile system. Curtisia, Corokia, Melanophylla, and Kaliphora. In view of its sporadic occurrence in the angio­ Figure 24 gives a number of other anatomical sperms and in the Myrtaceae, the transeptal sup­ differences between the flowers of Syzygium s. 1. ply clearly seems to be derived. However, this and Eugenia s. s. The fact that the former tend does not necessarily mean that Syzygium s. 1., with to be moderately to strongly tanniferous, whereas an axile ovular system, is ancestral to Eugenia s. s., the latter tend to be only slightly so, suggests that with a transeptal system. If the other characters these taxa may also differ chemically. The occur­ listed in Fig. 24 are evaluated by the traditional rence of sclerenchyma in flowers of several species dogmas of anatomy and organography, it is evi­ of Syzygium s. 1. and its absence in Eugenia s. s. dent that presumably primitive and presumably may be related to floral size and perhaps the con­ derived characters are fairly well combined in comitant adaptation to ornithophily or chiroptero­ each of the two genera. Thus, the many organo­ phily (Schmid, 1971). These modes of pollina­ graphic similarities between Eugenia s. s. and tion have been reported for Eugenia s. 1. (Grant, Syzygium s. 1., which have made life so difficult 1950; Knuth, 1904; van der Pijl, 1936, 1956; for the working myrtalean taxonomist, presumably Porsch, 1941 ; Werth, 1900). resulted from parallel evolution in the two groups Thus on the basis of the facts from both vege­ after they had arisen from the same ancestral tative and especially reproductive anatomy, it is stock, an ancestor no doubt with an axile ovular difficult to escape the conclusion that there are system. The divergence of Eugenia and Syzygium at least two largely allopatric, co-ordinate groups apparently occurred in the distant past, and embraced by Eugenia s. 1.: the strictly Old World neither genus is directly ancestral to the other. genus Svzygium s. 1. and the mainly New World This echoes the situation of the two main lines in genus Eugenia s. s. Furthermore, this anatomical Eucalyptus (Carr and Carr, 1962). The Myr­ evidence warrants the taxonomic acceptance of taceae, indeed the Myrtales as a whole, are well these two groups despite the fact that the gross represented in both Paleogene and Neogene de­ organographic differences between them are not posits of India, and Syzygium is known from the entirely clear-cut. The fact that the anatomical of India (Lakhanpal, 1970). The con- 434 AMERICAN JOURNAL OF BOTANY [Vol. 59 siderable vanation in organography and vascu­ BAILEY, L. H. 1930. Several arrangements of names lature of Eugenia s. l. suggests that the complex necessitated by recent studies. Gentes Herb. 2: may still be undergoing rapid evolution. 164-174. 1949. Manual of cultivated . Rev. ed. In Fig. I purposely displayed the taxa I 24 Macmillan Co., New York. recognize as "clusters" rather than as "phylo­ BAILLON, H. 1880. The natural history of plants. Vol. genetic lines"; the rationale for the circumscription VI. L. Reeve & Co., London. of these taxa is given in Schmid (1971; in press b). BENTHAM, G. 1869. Notes on Myrtaceae. J. Linnean With a literal interpretation of vasculature (see Soc. Bot. 10: 101-166. Schmid, in press a, for the pitfalls of this), and ---, AND J. D. HOOKER. 1862-67.' Genera plan­ particularly with a penchant for reduction series, tarum. Vol. I. Reeve & Co., London. a botanist might attempt to contrive a phylogenetic BERG, O. 1855-56. Revisio Myrtacearum Americae scheme for the species of Eugenia s. l. reported hue usque cognitarum s. Klotzschii "Flora Americae aequinoctialis" exhibens Myrtaceae. Linnaea 27: here. However, in view of our present state of 1-472. ignorance about the Myrtaceae, I believe that un­ 1857-59. Myrtaceae, p. 1-655. In C. F. P. bridled phylogenetic speculation would accom­ de Martius [ed.], Flora Brasiliensis. Vol. XIV, Pars plish little and would probably lead to error. Fi­ 1. Frid. Fleischer in Comm., Lipsiae. nally, the recent papers of Carlquist (1969) and BLUME, C. L. 1849-51. Museum botanicum Lugduno­ Eyde (1971) remind us that consistent differences Batavum. Vol. I. E. 1. Brill, Lugduni-Batavorum, in histology or vascular patterns can adequately Netherlands. serve as useful taxonomic markers although their BONNER, C. E. B. 1948. The floral vascular supply in Epilobium and related genera. Candollea 11: 277­ phylogenetic or even functional significance re­ 303. mains obscure. BROWN, W. H. 1938. The bearing of nectaries on the phylogeny of flowering plants. Proc. Amer. Phil. NOTE ADDED IN PROOF-The analysis of an Soc. 79: 549-595. additional 19 species of Eugenia s. s. (including BULLOCK, A. A., AND S. G. HARRISON. 1958. Nomen­ the segregate lossinia) and an additional 21 spe­ clatural notes: IV. The correct name for the . cies of Syzygium s. l. (including samples of such Kew Bull. 1958: 52. segregates as Acmena, Caryophyllus, Cleistocalyx, CANDOLLE, A. P. DE. 1828. Prodromus systematis nat­ Aphanomyrtus, Piliocalyx, Pareugenia, Tetra­ uralis regni vegetabilis. Pars 3. Treuttel et Wurtz, eugenia, Pseudoeugenia, Acmenosperma, Xeno­ Paris. dendron, Acicalyptus, and lambosa) substantiates 1842. Memoire sur la famille des Myrtacees. Mem. Soc. Phys. Hist. Natur. (Geneve) 9: 1-61. the anatomical differences listed in Fig. 24. A few CARLQUIST, S. 1969. Toward acceptable evolutionary of the large-flowered American species of Eugenia interpretations of floral anatomy. Phytomorphology s. s., however, have sclerenchyma in the flowers 19: 332-362. (this perhaps to be expected from a functional ---, AND P. H. RAVEN. 1966. The systematics and viewpoint), and a number of species have very anatomy of Gongylocarpus (Onagraceae). Amer. tanniferous flowers. J. Bot. 53: 378-390. CARR, D. J., AND S. G. M. CARR. 1962. Natural groups LITERATURE CITED within the genus Eucalyptus, p. 426-445. In G. W. Leeper [ed.], The evolution of living organisms. AIRY SHAW, H. K. 1949. Additions to the flora of Melbourne University Press, Parkville, Victoria. Borneo and other Malay islands. XX. The Myr­ ---, AND ---. 1970. Idem. II. Development taceae of the Oxford University Expedition to Sara­ and structure of oil glands in the embryo. Aust. J. wak, 1932. Kew Bull. 1949: 117-125. Bot. 18: 191-212. 1966. J. C. Willis' A dictionary of the flower­ CARR, S. G. M., AND D. J. CARR. 1961. The func­ ing plants and ferns. 7th ed. University Press, tional significance of syncarpy. Phytomorphology Cambridge. 11: 249-256. AMSHOFF, G. J. H. 1958. Notes on Myrtaceae. VII. ---, AND ---. 1969. Oil glands and ducts in Myrtaceae of French Equatorial . Acta Bot. Eucalyptus L'Herit. I. The phloem and the pith. Neer, 7: 53-58. Aust. J. Bot. 17: 471-513. 1963. Myrtaceae, p. 333-351. In C. A. Backer and R. C. Bakhuizen van den Brink, Flora of ---, ---, AND L. MILKOVITS. 1970. Idem. III. Java (Spermatophytes). Vol. I. N. V. P. Noord­ The flowers of series Corymbosae (Benth.) Maiden. hoff, Groningen, Netherlands. Aust. J. Bot. 18: 313-333. ARBER, A. 1931. Studies in floral morphology. I. On CHATTAWAY, M. M. 1959. The anatomy of bark. VII. some structural features of the cruciferous flower. Species of Eugenia (sens. lat.). Trap. Woods 111: New Phytol. 30: 11-41. 1-14. BACCARINI, P. 1884. Osservazioni anatomiche sopra CORNER, E. J. H. 1952. Wayside of Malaya. 2nd alcuni ricettacoli fiorali. Ann. R. 1st. Bot. (Roma) ed. Government Printing Office, Singapore. 1: 66-88. DADSWELL, H. E., AND H. D. INGLE. 1947. The wood BAEHNI, C., AND C. E. B. BONNER. 1948. La vascu­ anatomy of the Myrtaceae, I. A note on the genera larisation des fleurs chez les Lopezieae (Onagracees). Eugenia, Syzygium, Acmena, and Cleistocalyx. Candollea 11: 305-322. Trap. Woods 90: 1-7. ---, AND ---. 1949. La vascularisation du tube DAVIS, G. L. 1966. Systematic embryology of the floral chez les Onagracees, Candollea 12: 345-359. angiosperms. John Wiley & Sons, New York. April, 1972] SCHMID-EUGENIA-SYZYGIUM CONTROVERSY 435

DIELS, L. 1922. Die Myrtaceen von Papuasien. Bot. KNUTH, P. 1904. Handbuch der Bllitenbiologie. Bd. Jahrb. 57: 356-426. III. Die bisher in aussereuropaischen Gebieten ge­ EAMES, A. J. 1961. Morphology of the angiosperms. machten bllitenbiologischen Beobachtungen. Ed. by McGraw-Hili, New York. E. Loew. Tl. 1. Cycadaceae bis Cornaceae. Wil­ ERDTMAN, G., AND C. R. METCALFE. 1963. Affinities helm Engelmann, Leipzig. (III. Bd, not translated.) of certain genera incertae sedis suggested by pollen LAKHANPAL, R. N. 1970. Tertiary floras of India and morphology and vegetative anatomy. Kew Bull. their bearing on the historical geology of the region. 17: 249-256. Taxon 19: 675-694. EYDE, R. H. 1966. Systematic anatomy of the flower LEINS, P. 1965. Die Inflorescenz und friihe Bllitenent­ and fruit of Corokia. Amer. 1. Bot. 53: 833-847. wicklung von F. Muell. (Myr­ 1967. The peculiar gynoecial vasculature of taceae). Planta 65: 195-204. Comaceae and its systematic significance. Phyto­ MCVAUGH, R. 1956a. Nomenclatural notes on Myr­ morphology 17: 172-182. taceae and related families. Taxon 5: 133-147; 1968. Flowers, fruits, and phylogeny of Alan­ 162-167. giaceae. J. Arnold Arboretum 49: 167-192. 1956b. Tropical American Myrtaceae. Notes 1971. Evolutionary morphology: distinguish­ on generic concepts and descriptions of previously ing ancestral structure from derived structure in unrecognized species. Fie1diana: Bot. 29: 143-228. flowering plants. Taxon 20: 63-73. 1958. Myrtaceae. In Flora of Peru. Field FAWCETT, W., AND A. B. RENDLE. 1926. Flora of Mus. Natur. Hist., Bot. Ser., 13(pt.4): 567-818. Jamaica. Vol. V, Pt. 3. British Museum (Natural 1963a. Tropical American Myrtaceae, II. History), London. Notes on generic concepts and descriptions of pre­ FEDOROV, A. A. [ed.], 1969. Chromosome numbers of viously unrecognized species. Fieldiana: Bot. 29: flowering plants. V. L. Komarov Botanical Insti­ 391-532. tute, Academy of Sciences of the U.S.S.R., Lenin­ 1963b. Myrtaceae. In Flora of Guatemala. grad. (In Russian.) Fieldiana: Bot. 24(pt. 7): i-viii; 283-405. GAGNEPAIN, M. F. 1917. Classification des Eugenia. 1968. The genera of American Myrtaceae­ Bull. Soc. Bot. France 64: 94-103. an interim report. Taxon 17: 354-418. GRANT, V. 1950. The protection of the ovules in 1969. Myrtaceae. In B. Maguire et al., The flowering plants. Evolution 4: 179-201. botany of the Guayana Highland-Part VIII. Mem. GRAY, A. 1854. United States Exploring Expedition. N. Y. Bot. Garden 18: 55-280. During the years 1838-1842. Under the command MATTHEWS, J. R, AND E. M. KNOX. 1926. The com­ of Charles Wilkes, U.S.N. Vol. XV. Botany. parative morphology of the stamen in the Ericaceae. Phanerogamia. Vol. I. Sherman, Philadelphia. Trans. Proc. Bot. Soc. Edinburgh 29: 243-281. GUILLAUMIN, A. 1938. Materiaux pour la flore de la MAURITZON, J. 1939. Contributions to the embryology Nouvelle-Caledonie, 1.111. Revision des Myrtacees of the orders Rosales and Myrtales. Lunds Univ. a fruit charnu precedee de quelques notes supple­ Arsskrift N. F., Avd, 2, 35(2): 1-121. mentaires sur les Myrtacees a fruit sec. Bull. Soc. MELCHIOR, H. 1964. A. Engler's Syllabus der Pflanzen­ Bot. France 85: 626-653. familien. 12. Aufl. Bd. II Angiospermen. Gebrlider HEINIG, K H. 1951. Studies in the floral morphology Borntraeger, Berlin-Nikolassee. of the Thymelaeaceae. Amer. J. Bot. 38: 113-132. MELVILLE, R. 1962. A new theory of the angiosperm HENDERSON, M. R 1949. The genus Eugenia (Myr­ flower: I. The gynoecium. Kew Bull. 16: 1-50. taceae) in Malaya. Gard. Bull. (Singapore) 12: MERRILL, E. D. 1917. Alabastra Borneensia. J. Straits 1-293. Branch Roy. Asiatic Soc. 77: 189-247. HOSOKAWA, T. 1940. Materials of the botanical re­ 1937. Aphanomyrtus Miquel and Pseudo­ search towards the flora of Micronesia (XIX). J. eugenia Scortechini. Blumea (Suppl.) 1: 107-111. Jap. Bot. 16: 535-545. 1939. Le genre "Cleistocalyx" Blume (Myr­ HUMMEL, K, AND K STAESCHE. 1962. Die Verbreitung tacees) serait-il a preferer au genre "Acicalyptus" der Haartypen in den natlirlichen Verwandtschafts­ A. Gray? Bull. Soc. Bot. France 86: 377-379. gruppen, p. 207-250. In 1. C. T. Uphof, cited below. 1950a. On the synonymy of Iossinia reinward­ INGLE, H. D., AND H. E. DADSWELL. 1953. The anat­ tiana (Blume) Blume. J. Arnold Arboretum 31: omy of the timbers of the southwest Pacific area. 329-333. III. Myrtaceae. Aust. J. Bot. 1: 353-401. 1950b. Readjustments in the nomenclature of JACKSON, B. D. 1928. A glossary of botanic terms. Philippine Eugenia species. Philippine J. Sci. 79: 4th ed. Gerald Duckworth & Co., London. 351-430. JAYAWEERA, D. M. A. 1957. Variation in the flower ---, AND L. M. PERRY. 1937. Reinstatement and of Eugenia malaccensis Linn. 1. Linnean Soc. Lon­ revision of Cleistocalyx Blume (including Acicalyp­ don Bot. 55: 721-728. tus A. Gray), a valid genus of the Myrtaceae. J. KAUSEL, E. 1956. Beitrag zur Systematik der Myrta­ Arnold Arboretum 18: 322-343. ceen. Ark. Bot. 3: 491-516. ---, AND ---. 1938a. A synopsis of Acmena 1957a. Idem. II. Ark. Bot. 3: 607-611. DC., a valid genus of the Myrtaceae. J. Arnold 1957b. Myrtaceae, #28. In J. Angely, Arboretum 19: 1-20. Catalogo e estatistica dos generos botanicos fanero­ ---, AND ---. 1938b. On the Indo-Chinese spe­ gamicos. Instituto Paranaense de Botanica, Curi­ cies of Syzygium Gaertner. J. Arnold Arboretum tiba, Parana, . 19: 99-116. 1960. Zur Systematik von Pilothecium Kiar­ ---, AND ---. 1938c. The Myrtaceae of China. skou. Ark. Bot. 4: 401-405. J. Arnold Arboretum 19: 191-247. 1966. Lista de las Mirtaceas y Leptosperrna­ ---, AND ---. 1939. The myrtaceous genus ceas . Lilloa 32: 323-368. Syzygium Gaertner in Borneo. Mem. Amer. Acad. 436 AMERICAN JOURNAL OF BOTANY [Vol. 59

Arts Sci. 18: 135-202. (Reprinted in Mem. Gray RIDLEY, H. N. 1922. The flora of the Malay penin­ Herb. Harvard Univ. 4: 135-202. 1939.) sula. Vol. 1. I.. Reeve and Co., London. ---, AND ---. 1942a. Plantae Papuanae Arch­ SCHMID, R. 1970. Comparative floral anatomy of Myr­ boldianae, IX. J. Arnold Arboretum 23: 233-297. taceae, with emphasis on Eugenia and its segregates. ---, AND ---. 1942b. Myrtaceae, p. 74-78. Tn Amer. J. Bot. 57: 744-745, viii. (Abstract.) A. C. Smith et al., Fijian plant studies, II. Botanical 1971. Floral anatomy of Eugenia sensu lato results of the 1940-41 cruise of the "Cheng Ho." (Myrtaceae). Ph.D. Thesis, The University of Sargentia 1: 1-148. Michigan. METCALFE, C. R., AND I.. CHALK. 1950. Anatomy of In press a. Floral bundle fusion and vascular the . Vol. I. Clarendon Press, Oxford. conservatism. Taxon. MOELIONO, B. M. 1970. Cauline or carpellary placen­ In press b. Floral anatomy of Myrtaceae. tation among dicotyledons (Axis-borne versus ­ I. Sy rygium S. I. Bot. Jahrb. borne ovules). The cauline ovules of centrosperms. ---. In press c. Idem. II. Eugenia S. s. J. Arnold 2 volumes. Van Gorcum & Compo N. V., Assen, Arboretum. Netherlands. SINHA, S. c., AND B. C. JOSHI. 1959. Vascular anatomy MOLL, J. W., AND H. H. JANSSONIUS. 1918. Mikro­ of the flower of Punica granatum I.. 1. Indian Bot. graphie des Holzes der auf Java vorkommenden Soc. 38: 35-45. Baumarten. Bd. III. E. J. Brill, Leiden. TIEGHEM, P. VAN. 1875. Recherches sur la structure MOUSSEL, B. 1965. Contribution a l'etude cyto-taxi- du pistil et sur l'anatomie comparee de la fleur. nomique des Myrtacees. Mem, Mus. Nat. Hist. Mem. Acad. Sci. Inst. Imperial France, Ser. 2, 21: Natur., Nouv. sa., Ser. B, Bot. 16: 91-125. 1-261. (Reprinted from Mern, des Savants Etrangers NAMIKAWA, I. 1919. Uber das Offnen der Antheren a l'Institut, Ser, 2, 21: 1-261. 1871.) bei einigen Solanaceen. Bot. Mag. (Tokyo) 33: TUNG, C.-I.. 1935. Development and vascular anatomy 62-69. of the flower of Punica granatum I.. Bull. Chinese NIEDENZU, F. 1893. Myrtaceae, p. 57-105. In A. Bot. Soc. 1: 108-128. Engler and K. Prantl, Die natiirlichen Pflanzen­ UPHOF, 1. C. T. 1962. Plant Hairs. Tn K. Linsbauer familien. Tl. 3, Abt. 7. Wilhelm Engelmann, Leip­ [ed.], Handbuch der Pflanzenanatomie. Bd. IV, Tl. zig. 5. Gebrlider Borntraeger, Berlin.-Nikolassee. PASS, A. 1940. Das Auftreten verholzter Zellen in WEBERLlNG, F. 1966. Additional notes on the myrta­ Bliiten und Bliitenknospen. Osterr. Bot. Z. 89: ceous affinity of eugenioides Schltr. Kew 119-164; 169-210. Bull. 20: 517-520. PERRY, I.. M. 1950. Notes on some Myrtaceae of Fiji. WELCH, M. B. 1920. glands. J. Proc. J. Arnold Arboretum 31: 350-371. Roy. Soc. New S. Wales 54: 208-217. PETIT, L.-A. 1908. Recherches sur la structure anat­ WERTH, E. 1900. Bllitenbiologische Fragmente aus omique du fruit et de la graine des Myrtacees. Ostafrika. Ostafrikanische Nectarinienblumen und Doctoral Thesis. Univ. Paris, Ecole Superieure Phar­ ihre Kreuzungsvermittler. Ein Beitrag zur Erkennt­ macie, Annee 1907-1908, #8. nis der Wechselbeziehungen zwischen Blumen- und PIJL, I.. VAN DER. 1936. Fledermause und Blumen. Vogelwelt. Verh. Bot. Ver. Provo Brandenburg 42: Flora 131: 1-40. 222-260. 1956. Remarks on pollination by bats in the WIGHT, R. 1841. Illustrations of Indian botany. Vol. genera Freycinetia, Duabanga and Haplophragma, II. J. B. Pharoah, Madras. and on chiropterophily in general. Acta Bot. Neer. WILSON, K. A. 1957. A taxonomic study of the genus 5: 135-144. Eugenia (Myrtaceae) in . Pacific Sci. 11: PIKE, K. 1956. Pollen morphology of Myrtaceae from 161-180. the southwest Pacific area. Aust. J. Bot. 4: 13-53. 1960. The genera of Myrtaceae in the south­ POLUNINA, N. N. 1958. Floral biology and embryology eastern United States. J. Arnold Arboretum 41: of Callistemon lanceolatus Sweet. Bot. Zh. 43: 270-278. 1169-1178. (In Russian.) WINTON, A. 1.., AND K. B. WINTON. 1939. The struc­ PORSCH, O. 1941. Ein neuer Typus Fledermausblumen. ture and composition of foods. Vol. IV. Sugar, BioI. Gen. 15: 283-294. sirup, honey, tea, coffee, cocoa, spices, extracts, RAO, V. S., AND R. DAHLGREN. 1969. The floral anat­ yeast, baking powder. John Wiley & Sons, New omy and relationships of Oliniaceae. Bot. Notis. York. 122: 160-171. ZUCCONI, I.. 1958. Organogenesi del fiore ed em­ RASTOGI, S. P. 1951. Vascular anatomy of the flower briologia in . Dehn. Ente of some species of the Myrtaceae with special ref­ Nazionale per la Cellulosa e per la Carta, Pub­ erence to the inferior ovary. M.Sc. Thesis, Agra blicazioni del Centro di Sperimentazione Agricola University. e Forestale, Roma 2: 59-86.