ANATOMY Aid SYSTEMATIC POSITION of CENTAURODENDRON and YUNQUEA (COMPOSITAE)

Reprinted from BRITTOXA, VoL 10, No. 2, April 1958 Printed in U. S. A.

ANATOMY AiD SYSTEMATIC POSITION OF CENTAURODENDRON AND YUNQUEA (COMPOSITAE)

SHERWIN CARLQUIsT

Claremont Graduate Sehoo, Ilaneho Santa Ana Botanic Gardeit Claremont, California

I I ANATOMY AiND SYSTEMATIC POSITION OF CENTAURODENDRON AND YUNQUEA (COMPOSITAE)

SHERWIN CARLQUIST Claremont Graduate School, Raneho Santa Ana Botanic Garden Claremont, California

The genera Centaurodtnciron and Yunquea are remarkable in that they are both “rosette trees” which have been interpreted as belonging to the tribe (yn areae (Skottsberg l93 and are endemic to the island Masatierra in the Juan Fer nandez Islands. (‘ nta,rodendro’o consists of two species. (‘. (11u(aen1Juh( S .JOIIOH and (. palin.iforme Skottsherg, whereas Ynnquea tenii Skottsberg is the single species of its genus. The main purpose of the present study is to show how anatomical evidence bears on the systematic relationships which have been pro posed on the basis of gross morphology, Because (entauroclendron and Ynnquea are the only arboreseent genera referred to Cynareae, and because they show little superficial similarity to other Cynareae in vegetative characters, a study of their anatomy is desirable in determining if this is, indeed, the correct dis position of these genera. A secondary problem is created by the genus Yunquee, which Skottsberg (1929) erected on the basis of scanty egetative material. Allhough subsequent collections have furnished no flowers of this plant, which evidently flowers with extreme infrequency, anatomical data from the vegetative portions of the plant and from recently collected mature achenes appear to be decisive in assessing Skottsberg ‘s (1 929) suggestion that it should be regarded as a genus closely related 10 (‘Entuu,’odI ndi’on. Individuals of iunquea. are rare, and the localities in which they grow are difficult of access (Kunkel 1957; Skottsberg 1953). It may become extinct, as has so much of the antoelithonous Juan Fernandez vege tation,withont hiaviiig been. collected in the flowering state. The recent discovery (Skoltsberg 1957) of (]entaurodendron palmfforme is significant, for this species shows greater similarity to Yunquea than does C. (lracaenonlnc. Moreover, anatomical characters can be used in contrasting the two species of (n to urodendron.

MATERIALS ANt) METHODS Centaurodendi’on druea.enoides is only slightly more abundant than C. patmi forome or Yunqnui. but has been collected, at rare intervals, in flower. Therefore, the writer was able to use herbarium material for studies of the anatomy of inflorescence parts. Except for this material, listed below, the present study was based entirely on material collected by Dr. Skottsberg or collectors who assisted him during ins visit to the islands in 1955. Skottshcrg’s own collections were preserved in dilute ethyl alcohol, and consisted of stems, leaves, and shoot tips. wood samples were also collected and dried. Although no herbarium specimens were prepared to document these collections, which in a sense are doenniented by Dr. Skottsherg’s intensive knowledge of the Juan Fernandez flora. lhc writer is retaining the liqoid-preserved samples on which the studies were based. The specimens are as follows:

BRITTONIA 10: 78-93, April 1958. —78- 1958] CARLQUIST: CENTAURODENDRON AND YUNQUEA 79

Centaurodendron dracaenoides. Vegetative. Coil. C. Skottsberg, Sept. 3, 1955. Portezuelo do Villagra, Masatierra. Liquid. Same. Flowering. Coil. H. Weber, without data, 1937 (University of California Herbarium, Berkeley. Centaurodendron palmiforme. Vegetative. Coil. C. Kunkel, Sept. 3, 1955. Masatierra. Dried. Same. Weathered infiorescence. Coil. B. Sparre, Jan. 10, 1955. Masatierra. Dried. Same. Old stem. Coil. B. Sparre, March 5, 1955. Masatierra. Dried. Yunquea tenzii. Vegetative. Coil. C. Skottsberg, June 3, 1955. El Yunque, Masatierra. Liquid. Same. Aehenes. Coil. G. Kunkel, June 3, 1955. El Yunque, Masatierra. Dried. The writer wishes to express sincerest appreciation to Dr Skottsberg for the material mentioned, and to Dr T. H. Goodspeed, through whose good offices arrangements for these collections were made. The cooperation of the University of California Herbarium in allowing use of the specimen listed above is grate fully acknowledged. Herbarium material was prepared by treatment with 2.5 per cent NaOH in a 600 C oven. When material was sufficiently cleared and expanded by this method, it was washed and stored in 50 per cent ethyl alcohol. Both the alcohol- fixed material and the “revived” herbarium specimen portions were dehydrated according to Johansen’s (1940) tertiary butyl alcohol series, and embedded and sectioned according to the usual techniques. Treatment with hydrofluoric acid for several weeks was found necessary for softening the rather sclerotic stems of Centaurodendron dracaenoides before dehydration. Sections were stained by means of tannie acid and ferric chloride, followed by a safranin-fast green combination. In addition to sectioned material, whole mounts of leaves, involue ral bracts, flowers, and pollen were prepared by the NaOH technique mentioned above, followed by dehydration and staining with safranin. The wood samples were sectioned on a sliding mierotome after boiling in water, followed by immersion for about six weeks in 36 per cent hydrofluoric acid. Stems of C. palmiforme were sectioned without such treatment. Safranin in absolute ethyl alcohol was employed as a stain.

VEGETATIVE ANATOMY Leaves of Centaurodendron differ markedly from those of Yunquea in gross morphology. Yunquea has petiolate, toothed leaves with prominent ribs (corre sponding to the major veins) on the lower surface; the ribs and petiole have a light covering of uniseriate, non-glandular hairs. Centaurodendron leaves are sessile (rarely somewhat petiolate), nearly emarginate (the teeth obscure, although more obvious in C. palmiforrne), and smooth and glabrous on both surfaces. Study of foliar ontogeny demonstrates that leaves of the two genera are basically more alike than these differences would suggest, although study of their mature anatomy reveals additional characters which distinguish the genera. The shoot apex of Yunquea shown in Figure 1 serves to demonstrate the leaf origin in both of these genera. Leaf primordia originate in a spiral around the dome-shaped apical meristem. The meristem itself is rather clearly zonate. Three layers of tuniea are discernible, and the corpus contains a central mother- cell zone which has abundant pectic accumulations at cell interstices. Prominent rib-meristem action is initiated a short distance below the mother-cell zone. These clearly defined files of cells are not responsible for the formation of the entire pith, however, but only the central portion of it. The shoot apex of Centaurodendron dracaenoides shows the same features except that the groups

side.

adaxial

the on

occur

laeumiae large although portion, abaxial the in

present are

spaces

intercellular Smaller tissue. spongy and palisade into tion

differentia

with

present, are

layers more numerous that in lamina the from

anatomically

differs C. dracaenoides

base of The leaf 12. Figure in mnaticall

diagram

illustrated that to corresponds section This 3. Figure in dracae’no’ides

C. for

shown is

lamina the base subtending leaf the of portion narrowed

The

petiole. defined

clearly a lacks C’ntaurodendron Node. and Petiole

respects. in all dracaenomd.es

of that to

be identical to observed was paliniforme C. of structure Lamina

alike. quite is genera two in the lamina the of structure above, mentioned

points the

from

Aside structure. a such have also leaves Centaurode’ndro’n of teeth

obscure

relatively

The construction. in hydathodic are leaves Yuiiquea of margins

the teeth

prominent The Centanrodendron. do in often they as a vein jacket

completely not and do

abundant are less they although veins, larger the of sheaths

bundle the

within are present Fibers left). 5, (Fig. leaf the of surface lower the

on rib

a with prominent associated those are extensions sheath with veins present;

also are sheaths

bundle In Yu’nquea, left). 4, (Fig. veins smaller the of some even

and 4, right) (Fig. veins larger the around abundant are fibers addition, In

(Fig. 4, right). extensions bundle-sheath conspicuous have veins larger and

in Centaurodcndron. developed well are sheaths Bundle numerous. more are

layers course, occur, of veins Where has six. which qnea, Yen in than layers,

nine

cell- have which of Centeurodenthon. species both in greater is layers piiyll

number of ineso total the Moreover, quea. fun in occurs layer single a whereas

droccceuoides, in present characteristically are palisade of layers two (1938),

by Skottsherg mmotecl As respects. of number a in 5) (Fig. Yunqnc of that

(Fig. from 4) differs cliacacnoith’s (cniaurodeudron of lamina niature The

suggest. Centauiea in clots’’

‘‘glandular of similar 193s) (1929. reports Skottsherg’s as Cynareae, in also

occur probably 1957a), they (Carlqu:ist Hesperornaunia Mntisieae, of member a

in only i 1 ompos.itae for been demonstrated have tricliomes glandular biseriate

sunken Although collapsed. and be non-functional to appear however, leaves,

on mature hairs Glandular fun queG. and (‘c-n(aurodeadi’oa in both leaves,

the of both 5) surfaces on 4, (Figs. pockets into sunken become they Rather,

soon degenerate not do glandular hairs biseriate the part, most the For

both genera. in leaves young on present

are which hairs, uniseriate of preservation relative of is matter a qnea, then,

lun and Centavrodendron between of leaves vesture in difference The hairs.

glandular surrounding pockets of to the formation is which prior 2. Figure in

shown stage is the in evident triehomes such uniseriate of Degeneration cI.roa.

Ceniacrodc’n of mature leaves on be lacking wholly to appear They on petioles.

axils, and the in leaf leaves, of surfaces lower on the the ribs along only served

are pre hairs uniseriate leaves, Yuiiquea mature On Yunqnea. in they are as

genus, just that in herbage of portions on immature present hairs are glandular

non uniseriate the laurodendron, Ceo in leaves of nature glabrous the Despite

of photograph). bottom top and (near glandular biseriate and photograph) of

(top non-glandular uniseriate as be classified may trichomes These Ynqttea. of

petiole immature an on hairs types of two of presence the shows 2 Figure

below. mentioned

stage, early at an deuiareated are sciereids pith become to destined cells of

10 VOL. BRTT’TONIA 1958] CARLQUIST: (ENTAURODENDRON AND YUNQUEA 81

aC C WIi

Fin. 1. Yunquea enzii, longifildinal section of shoot apex; X 135. Pie. 2. Y. tnzii, portion of longitudinal section of young petiole; X 170. Pie. 3. Centaurodendron dracac noides, transection of sheathing leaf base, adaxial face at right: X 60. Fin. 4. C. draeac noideg. portion of lamina trausection, aclaxial face above; x 78. Pie. 5. P. (enzii. portion of lamina transection, adaxial face above; X 115. PIGs. 6, 7. Portions of transeetions from stems with some secondary growth, 108. PIG. 6, C. draeaenoides. Fj. 7, P. tenzii. 82 BRTONIA [VoL. 10 scierified hypodermis is prominently differentiated in the adaxial face (Fig. 3, right). Collenchyma is present in the abaxial layers. Around the bundles, fibers differentiate tardily in comparison with the rapid scierification of the adaxial hypodermis. Several secretory canals surround each bundle in the leaf base (Fig. 3); five or more are present around the larger veins. These canals are continuous with those associated with leaf traces in the cortex (Fig. 6). The leaf base of C. palmiforme differs from that of C. dracaenoides only in lacking a sclerified hypodermis (collenchyma is present on both faces) and in possessing fewer (usually two) secretory canals around the larger veins. Comparison of the leaf base of C. dracaenoides (Fig. 12) with the leaf traces in the cortex of a given node (Fig. 11, below) reveals that more numerous bundles are present in the leaf base. This condition results from repeated branching of the most lateral pair of leaf traces in the leaf base. Each of these laterals may branch several times. Other leaf traces are unbranehed. Figure 12 shows larger and smaller veins alternating in the leaf base. The larger veins depart from the vascular cylinder at a lower level than the smaller veins. Approximately sixteen traces, associated with sixteen gaps, were characteristi-. cally present in each node of the C. dracaenoides material examined. The petiolate nature of the leaf in Yunquea results in certain anatomical differences in comparison with Centaurodendron. Although the sheathing basal portion of the leaf (Fig. 9) is not unlike that of Centaurodendron (Fig. 12), the distal portion of the petiole (Fig. 10) has a central group of bundles in addition to the outer are of large and small bundles. This central group of bundles is further distinguished by the orientation of xylem and phloem, an orientation at variance with that of the outer arc. These bundles do not origi nate from an adaxial meristern. Study of successive sections indicates that they branch from the larger bundles of the outer arc. As in Centaurodendron, numerous small secretory canals surround each of the larger veins, both at the level shown in Figure 9 and that in Figure 10. As in Centaurodendron, they are continuous with secretory canals of the leaf traces in the cortex below a node, although they fade out near the lamina base and are not present in the lamina. At the level shown in Figure 10, secretory canals are present only around the bundles of the outer arc. The course of secretory canals in the cortex adjacent to leaf traces, and their extension into leaf bases, where they terminate below the lamina, has been reported by Van Tieghem (1872) in Xeranthemurn cytindraceum, Cirsium arvense, and Lctppa gra’ndiflora. The petiole of Yunquea shows resemblance to the leaf base of Centaurodendron in that collenehyma is present exterior to the outer arc of bundles in the petiole. The nodal venation of Yunquea differs from that of Centaurodendron in its greater abundance of veins. The number of traces which supply a leaf of Yunquea is approximately twenty; each of these is associated with a gap in the vascular cylinder. The larger number of traces may be correlated with the fact that the leaf base extends more nearly around the circumference of the stern than it does in Centaurodendron. As in Centaurodendron the most lateral pair of leaf traces in each leaf base of Yu’nquea branch toward the margin, increasing the number of veins. Other veins throughout the leaf base branch; some are formed de novo in the leaf base (i.e. without downward connection to leaf traces). Thus a much greater number of veins is present in the leaf base of Yunquea than in that of Centaurodendron. 1958] CARLQUIST: CENTAIJRODENDRON AND YUNQTJEA 82

Precise data on nodal venation in other Cynareae is not available. The type illustrated here, however, is that which one would expect to be associated with the broad, sheathing leaf bases typical of many Cynareae. Stem. Figures 6 and 7 illustrate comparable portions of the cortex of Cen taurodendron dracaenoides and Funqvea tenzii respectively. In both of these stems an accumulation of secondary growth was present. Cork formation, initiated in the hypodermal layer, is seen in both. In C. dracaenoides and C. paimiforme, a few ]ayers of collenehyma are formed in the outer cortex, whereas none is present in Funquea Funquea, however, characteristically develops numerous sclereids in the outer cortex, whereas no selereids were observed in cortex of stems of either Centaurodendron species. In both Figures 6 and 7, a leaf trace is shown. The prominent cap of fibers is similar in both genera. Surrounding the bundle cap, various numbers of secretory canals are present. The number of secretory canals around leaf traces of Centaurodendron varies from one to three, whereas from three to five are characteristically present in such leaf traces of Funquea. These canals are never present near the xylem face of the bundle. Some secretory canals are adjacent to vascular bundles of the vascular cylinder proper in both genera. These secretory canals are usually one per bundle; this number increases at a higher level in the stem for any given bundle, as the bundle becomes demarcated as a leaf trace. In addition, a few secretory canals arranged without reference to leaf traces were observed in the cortex of Funquea.

FIGS. 8—10. Yunquea fenzii. FIG. 8. Reconstruction (based on serial sections) of a node, showing veins to a single leaf oniy; x 3.3 Flo. 9. Diagrammatic transection of leaf base at the level at which the leaf is cut in Figure 8; X 3.3. FIG. 10. Diagrammatic transec tion of petiole; X 4.8. Fios. 11, 12. Centaurodendron draeaenoides. FIG. 11. Reconstruc tion of a node, showing veins to a single leaf only; X 6. FIG. 12. Diagrammatic transection of leaf base at the level at which the leaf is cut in Figure 11; X 4. Black portions of bundles in Figures 9, 10, and 12 are xylem. 84 BRITTON IA [VoL. 10

The presence of secretory canals in Compositae has been the subject of several surveys, such as those of Van Tieghem (1872) and CoT (1903). A particular feature of interest, however, is the presence of niore than one or two canals per leaf trace iii Cynareae, whereas in other tribes of the family they are, if present, only one or two per trace. Although some Cynareae have only a pair, such as Cynara scolyrnus (Jeffrey 1917) or have a large number abaxial to a trace, as in Aretium (Jeffery, op. cit.), certain examples which resemble the condition iii Centaxrode’ndron and Yunquea may be cited. Such an example is Seriatula centauroules (Van Tieghem 1872) ; several others, also from the snbtribe Centanrinae, are offered by Col (1903). Distinctive conditions of selerification are found in the pith of Centauroden thou and Yunqueo. In Ynnquea (Fig. 14) and C. palmiforme the inner margins of bundles have prominent caps of fibers which extend for considerable distances mto the pith; in the pith of these, however, no isolated scicreid groups occur, and the parenehyma is thin-walled. Iii Centcrurodendron draeaenoides (Fig. 13), on the contrary, numerous large nests of selereids are scattered throughout the pith. Sclereuehymatous inner margins of bundles also occur in C. draeae noides, und sonic of the selereid nests, which consist of similar cells, are con nected with these, The scicreid nests of C. clraeaenoic?es are clearly fore shadowed in the apical regions of the shoot before selerifieatioi. takes place. Cells destined to become selereids are much smaller in diameter than cells which will develop into thin-walled parenehyma. The presence of selerenehyma as inward extensions of bundles in Ce’nlauro. dendrou. and Y’u’nquea, and as selereid nests in C. dracaenoides pith, may have some relation to the arboreseent growth-form of these genera. If the accumula tion of secondary xylem is slow (as compared with that, in truly arboreal plants), additional selerenehyma may be of value in providing mechanical support for an arhorescent form. it is interesting to note that such arboreseent Compositae

‘‘-

I3: ‘

Fm. 13. Gee faurodeedro’n draeaenoides, portion of pith transection; X 95. Pm. 14. Yea qnea tenzii, portion of pith transection, showing inner margins of vascular bundles; X 95. 1958] CARLQUIST: CENTAURODENDRON AND Y1NQUEA 85 as Hes2eroma 0010 (Carlquist l.957a) and certain Guayana Mutisieae (Cariquist 19;38a) have conditions quite similar to C. dracoenoith’s. Likewise, the less woody species of Fitchia an arboreseent. genus of Reliantheae, demonstrate a wide variety of pith selerification (Cariquist 1957e). Ilerbaceous or shrubby members of these tribes at large seem mostly to lack such selerenchyma. If such scierification were a secondary acquisition, as it appears to be in Centauroden circa, rather than pruiitive1y present, it might be interpreted as a partial indi cation that a truly arboreal form is derived in Compositae. Such a suggestion would certainly not. rule out the possibility of woody fornis among primitive Compositae. however. The lack of pith sciereicts in (Cilt(1l(iOd(flthOfl 7xilniifoimo and Yunqvca may be related to the life-form of these two species. They are mon oearpi c, whereas C. (1151(0 coo ldes is hapaxanthic (Skottsberg, persoi ial corn nuuiieation)

Wood. Beeause Ce ntait i’od n dron and Vu nq 1( a are the only arborescent genera assigned to Cynareae, study of anatomy of their secondary xylem seems desirable. The oldest stein samples of Yvnqn.ea and C. pal.rniforrne had a radial width of approximately 3 mm of secondary xylem. The stem of the longer-lived C. d,aeaeuoidcs collected by Skottsberg, however, came from ‘‘the last mature generation of branches’’ awl had a thick cylinder or xylem. Despite this ddfercnee in wood samplcs--——prohahlv related to the life form of the plants— differences in the structure of the secondary xylem in the two genera can he demonstrated. These differences were verified by cOlilparison of younger stein samples of C. dlaCaCfloldl s with 11w stems of C. paltnifoeiuc and Iunquea, which had a similar aecuniulation of xylem. The measurements in Table I are presented as a means of comparing the species and genera with each other, and of comparing the two genera with other woody Compositae.

TABLE 1. Wood measurements of Centaurodendron and Von çpua.

a a a C C) x C2 — ,..n 8 u C) C) 5 C)Q.) ‘ f tiC)

‘ , u 4

SC) SC) S5 •C) n -i’ nS C. dracaenoides 26 90 53 244 1.80 396 198 21 .9 C. palmiforme 23—82 48 270 1.60 202 23 1.1

i:. teiiz; 3 60 40 335 1.36 402 — 15 1.5

Coniparisoii of C. dracaenoidcs with In nquea reveals an overall tendency toward longer, narrower elements in in nqu(a. Ceo to urodendron palmiforine is intermediate between the two taxa in this regard. Detailed comparisons ill volving qualitative characters also may be arranged as follows VESSELS. Vessel clement dimensions illustrate the tendency toward longer, narrower elements in innquca.. A further difference occurs in respect to vessel grouping. In inn qnea. vessels are inUre often solitary or hi more limited groups

120. All tenzii. quea 20, Fun 19, FIGs.

C. patmiforme. dracaenoides. Centaurodendron 17, 18, FIos. 16, 15, sections. tial FIGs.

tangen transections. 20, 18, 19, of FIGs. 17, 16, FIGS. 15, Sections xylem. secondary

Figures of 15-20 Explanation

ally clearly fibers and short wide, the C. poimiforme, of rays the in cells bent

of procum of The presence Yunquea. and that C. dra’ncaenoides of that between

be C. be of to said may wood pairniforme the intermediate In summary,

C. of palmiforme. wood observed all iii were

of ray Yunquea cells at 19). No (Fig. in were resins noted droplets resin sional

C. occa draeaetwides; cells some in ray deposits fill Resin wide. cells four least

at rays all appear multiseriate of the quea, Van wood the In species. cienclron

of in auro Cent are both wood present. rays width) cells maximum three or (two

C. or Some rather iiarrow palmiforrne Yunquea. wood of in were observed none

dracaeno’ides; wood of Centaurodendron in are infrequent rays Uniseriate 20).

(Fig. Yunquea rays of in lacking almost completely are cells procumbent Such

16, of 18). (Figs. Centawrode’ndron rays of portion central the in chiefly cells

procumbent of the narrow in presence occurs genera two the between difference

A 20). conspicuous 16, 18, (Figs. both in genera rays the of margins the along

erect fiber-like are cells some There be seen. may cells, ray erect prominently

by rays connected superposed two be to appear and extent, vertical their

of some along are parts rays narrowed which however, genera, both In height.

in over centimeter a be vertical may rays some which in Yunquea, than height

ray limited much more show of Ceataurodendron species two The RAYS.

17). 15,

(Figs. of Centaurodendron that in than 19) (Fig. Yvnquea of wood in present

cells are such Fewer groups. vessel or vessels surrounding wide, cell-layer

a than single more never sheath, incomplete scanty a of form the in Yunquea

and aurodendron in Cent present is parenchyrna vasieentric Compositae, in

is As typical 1957b). (Carlquist Mutisieae of certain wood the in been reported

bands have parenchyma Apotracheal them. to exhibit enough old been have not

may examined stems the although C. palmiforme, or Yunquea in observed were

bands parenehyma such No fibers. wood libriform of form derived genetically

a fact phylo in are cells parenchyma apotracheal such that 1958b), (Carlquist

Dubautia in case the be to appeared as he, it may and evident, is fibers to

similarity Their pareiichyma. apotracheal called be must cells these occurrence,

of mode and characteristics of their account On 15). of Figure the center across

visible is clearly band such (one stem the around extend which bands tangential

occur in cells latter These center). 16, (Fig. pattern storied a somewhat in

occur elements these ends; blunt or with rounded elements thinner-walled shorter,

are there right), extreme 16, (Fig. ends tapered fibers with thick-walled to long,

addition In dimorphic. arc fibers that say may one however, C. dracaenoides,

In 18). (Fig. short and 17) (Fig. wide uniformly are fibers C. palmiforme,

In Centaurodendron. in those with compared 20) (Fig. long 19) and (Fig.

narrow

are

in Yunquea fibers Wood PARENCHYMA. WOOD FIRERS; LIBRIFORM

or C. Yunquea. paimiforme

of vessels in observed were none although 16) 15, (Figs. C. dracaenoides

vessels of in occur deposits resin Prominent genera. both in pits bordered lar

circu alternate numerous of consists on vessels pitting wall Lateral respect. this

in is intermediate palmiforme Centaurodendron C. dracaenoides. those of than

10 [VoL. BRITTONIA 86 1958] CARLQUIST: CENTAURODENDRON AND YUNQUEA 87

L :11 88 BRITT0N1A lyon. 10

this species to C. (lrace( noules. The lack of apotracheal pareiiehyiiia in C. palrn.’i_ forme, as in Yunquea, may be a reflection of the rnonoearpie life form of these species, as opposed to the hapaxanthic habit of C. dracaenoides.

INFLORESCENCE ANATOMY The ascular cylinder of the infloreseenee axis and its branches in Centauro (4evdron dracoenoides is much like that of the vegetative stem, although these branches are narrower in diameter. Some of the bundles occupy a position slightly exterior to the vascular cylinder proper; these supply the outer involu eral bracts, wherea.s the remaining bundles supply the inner involueral bracts and flowers. Several secretory canals are present aroimd the phloem pole of bundles in the infloreseciice axis. Pith and pith rays of the infloresenee branches consist of thin-walled parenehyma. In the material of C. palmifor;ne examined, pith and pith rays were composed of thin-walled selereids winch were connected with the fibers of prominent bundle caps. This difference in selerification, however, may be due to the more advanced age of the C. paimiforme infloreseence as compared to that of C. dracaenoc1es which was studied. Involucres of C. pulndforme were too weathered to permit anatomical study. Consequently, ony the involucre of C. clracc&enoicies could be studied. The struc ture of an outer involueral bract is illustrated in Figure 21. The outer face of the bract at the level indicated consists of several layers of selerenehyma, whereas on the inner face, only the epidermis and a few cells of the layer beneath it are sclcrified. The remainder of the bract consists of thin-walled parenehynia. Vascular bundles occur in a single are in the parenehyma, and are not jacketed by fibers. A secretory canal is usually present at the phlocm face of each bundle. A transection near the apex of an outer involucral bract shows a marked altera tion of this pattern (Fig. 22). Several layers of sclerenchynia arc present on the inner, rather thait the outer, face of the bract; the outer face shows scierifi cation only on the epidcrmal cells. The mesophyll of the bract at this level, rather than. eonsistuig exclusively of thin-walled parenchyma, contains bands of selereids between the bundles. At the philoem pole of cacli bundle, a secretory canal is present. Inner involueral braets show a structure similar to that of the outer braets, differing only in that fewer layers of selerenchyma and mesophyll parenehyma are present. Receptacular bristles (Fig. 23) are quite simple in structure. They consist of a sclerificd epidermis, on which hiseriate glandular I1L hairs occur, overlying a ground tissue of thin-walled parenchyma. No vascular bundles were observed in rcceptaeuhar bristles. Among studies on anatomy of involucral braets in Cynareae, closest similarity to the condition observed in Centanrodendron clraeaenoides is ifound in Daniel’s (1890) descriptions and illustrations of C(ntaurea cineraria and Serralula tinctoria. Distribution of selerenehynma and occurrence of secretory canals in bracts of these species are reminiscent of, a]though not identical with, the situa tion in Cenfavroden.ciron draea.eaodes. Involueral bracts of other taxa differ more markedly. Although selerenehyma and secretory canals have been reported ._ for braets of Astereac (Napp-Zimui 1956) and Heliantheac (Napp-Zinu 1956: Cariquist l957c), they have a different pattern in these tribes. Time Mutisieae which. have been studied lack secretory canals in involueral bracts (Cariquist 1958a), and the Cynareac for which bract structure has been described, other than those mentioned above (Daniel 1890; Napp-Zinn 1956), do not by any means agree in. this respect with (‘entourodeadron dracaeno’ides. 1958] CARLQUIST: CENTAURODENDRON AND YFJNQUEA 59

FLORAL ANATOMY The careful descriptions of floral morphology of Centaurodendron d’racae noides given by Skottsberg (1938) are quite detailed, and few data need to be added. The floral venation of C. dracaenoides is of a simplified type found widely in Compositae, and corresponds exactly to the illustration given by the writer (1956) for Ifrio’phyllurn lanaturn. Short segments of median veins appear to be present at the corolla-lobe tips. Examination of these, however, proves them to be composed of fibers, not traeheary elements. No secretory canals were observed in the corolla of C. dracaenoides. The trichomes illustrated by Skotts berg on inner iii volticral bracts. receptacular bristles (shown here, Fig. 23), and

29m

Fins. 21—26, Centaaroclendron dracaenoides. Fin. 21. Portion of transection takeii mid way along the length of an outer involueral bract; adaxial face below. Fin. 22. Portion of transection taken near apex of an outer involueral bract, adaxial face above. Pie. 23 Transection of reeeptacnlar bristle; biseriate hair in section, below. FIG. 24. Portion of transection of wall of fertile mature acheiie. Fin. 25. Portion of transection of onter sur face of ovum from the same achene. Fin. 26. Transection of stigniatic. branch. FIGS. 27— 30, Yunquea tenzii. Fx. 27. Portion of transection of mature achene. Fin. 28, Portion of transection of dorsal face of ovule. PIG. 29. Portion of transection of lateral face of Same ovule. Fio. 30. Transection of two pappiis setae. Scale for all figures indicated,

details exine of figures Scale structure. for indicated. all

below. most 33. Optical FIG. of transection equatorial one third a of pollen showing grain,

of the furrows. Successive Fm. 32. optical transections exine; outermost of above, inner

31—33. Centaerodendron Fios. draeaenoicl

31. FIG. Entire grain, pollen showing one .

31 33

)

j -

‘ / r

•_

Heliantheae (Cariquist 1957c), which they are in internal to the bundles. Of

Cynareae (Lavialle 1912), in canals which are external to the bundles, and

few which secretory contain Those canals. which possess canals are members of

achenes Yunquea. of studied. Studies the achene of wall Compositae of reveal

Four bundles, with each an associated secretory canal, were observed in the

between the outer and inner selerenchyma at all the points on achene wall.

of layer of the thin-walled parenchyma. Some pareiichyma, is however, present

These anatomically differ from portions other by a wall of the greater thickness

of wall. the achene Prominent ridges are present. the achene on wall of Yunquea.

Some face. these sclereids of are scattered the through soft-walled parcnchyma

scierified, although less they markedly; are also smaller those than outer on the

of consist thick-walled selerenchyma.. Cells the achene lining cavity are also

plicated in structure. The outer epidermis, and two or three layers beneath it,

achene. wall The aehene Yunquea of (Fig. 27), the on contrary, is more com

a canal secretory opposite the phloern pole of of each the four veins the of

thin-walled parenchyma at maturity, is of and in interest the only presence of

The achene wall Centauiodenciron of dracaenoides (Fig. 24) consists of

been examined.

although they be appear to lacking in the majority Compositae of have which

1957c) been and have a reported in single genus of Mutisieae (Gueguen 1902),

literature. Secretory caiials are common in styles of Heliantheae (Cariquist

of Cynareae styles may be frequent, although mention is of this lacking in the

to show the presence of a secretory Occurrence canal. of secretory canals in

A a transection of stigmatic 0. branch dracanoides of (Fig. 26) is presented

they are more elongate not and sunkell pockets. into

setae pappus are biseriate glandular like those hairs of the herbage, although

[VoL. 10 nRn’roNIA _ 1958] CARLQUIST: CENTAURODENDRON AND YUNQUEA 91

the examples given by Lavialle, an achene wall identical to that of Centauroden dron draca.enoides is shown for Serratula- tinctoria. Presence of selerenehyma in. the achene wall, such as that of Yunquea, is found in certain species of Centaurea, such as C. c’i’rrhata (Lavialle, op. cit.). Lavialle also lists a number of species of Centaurea in which secretory canals are chareteristically present. Peculiar thickenings of the ovule epidermis are found in Centaurodendron dra,caenoides (Fig. 25). In C. dracaenoides, the epidermis is of approximately the same thickness at all points around the circumference of the ovule. In Yunquea, however, epidermal cells on the clorsiventral faces of the ovule (Fig. 28) are much longer than those on the lateral faces of the ovule (Fig. 29). The characteristic thickenings of the epidermis walls in both of these genera are identical with those reported by Lavialle (1912) in a iiumber of Cynareae, including $erratula, Centaurea, and Cyn era. This type of ovule epidermis is not found in certain other groups of Compositae. The pappus setae of Yunquea (Fig. 30) are identical to those of Centauro dendron dracaenoides as seen in transection. They consist of a small number of cells, similar in wall thickness, with no differentiation between the epidermis and center of the seta. Pollen. The pollen of Centaurodendron dracaenoides (Figs. 31—33) is of unusual interest in its highly complicated exine. The exine has three sculptured layers. The inner of these consists of numerous large bacula (Fig. 31; see also sectional views in Figure 32). The outer layers consist of fine rods. Although he does not illustrate it, Wagenitz (1955) gives some data concerning the pollen grain of C. dracaenoides. The writer’s observations agree with those of Wage nitz. except that fully turgid grains appear to be more nearly oblate, as shown in Figure 31. The polar to equatorial axis ratio given by Wagenitz is .98, indicating a virtually spherical form. The systematic importance of Wagenitz’ study on pollen of the subtribe Centauriuae seems to be considerable, because of the distinctive types of exine morphology which lie describes and finds refer able to the taxonomic system. The writer agrees with Wagenitz’ placement of C. dracaenoides pollen in t.he “Serra.tula-type” which includes (ientaurea (sec tions Plectocephalus and Cheirolophus only), Acroptilon, Ainberboa, Call/i cephalis, Crup’ina, Leuzea, Mantisalca, Plaqiobasis. Rhaponticum, and Serratula, hut excludes all other members of the subtribe Centaurinac.

SYSTEMATIC CONCLUSIONS The relatively precise and distinctive data concerning pollen offered by Wageriitz (1955) are perhaps most decisive in suggesting what the closest rela tives of Centaurodendron and Yunqvea may be. Data from secretory canal distribution, involucral bract anatomy, and achene anatomy indicate clearly that these genera belong within the tribe Cynareae, and that the closest com parisons are with the genera Centaurea and Serrat ala, in agreement with the data from pollen. The occurrence of pollen identical with that of Centauro dendron in Centaurea sect. Plectocephalus is of considerable interest, because Skottsberg (1938) considers this group in his discussion of the affinities of Centaurodendron, but rejects it on account of certain details of pappus and stigmatic branch morphology. These characters are often quite variable in Compositae, and the fact that Centaurea sect. Plectocephalus is represented by several species in Chile, the nearest land mass to the Juan Fernandez Islands, as well as the resemblance in pollen morphology, suggests that the possibility of 92 BRTTTONIA [VOL. 10 relationship between (‘e a to urodendro n and P1cc toceph al u.s should remain open. The specialized sexual eonditions in the head of (‘entaurodendron as compared to that of Oentovrea do not seem so impressive to the writer as they did to Skottsberg (1938). In another group of Juan Fernandez Compositae, the closely related gellera Rohinsonia, Rhetinothudron, and $yunphochaeta, the latter two genera appear to have evolved a dioecious condition from the monoecious condition fOund jU Robi’nsonia. Probably no species of Centaurinac clearly illus trates a desired ancestral type from which the highly distinctive genera eentourodendron and Yen queu can be derived, but the extreme closeness to this subtribc in respect to the anatomical characters reviewed above suggest that the two genera have considerable affinity to some members of Ceutauririae: a more discrete affinity, in fact than one can usually postulate between insular endemic genera and their putative mainland relatives. Data from vegetative anatomy also permit a relatively decisive outlook on relationships among the three species considered in this paper. The writer believes that closeness hi such features a.s glandular triehomes, leaf anatomy, node anatomy. distribution and number of secretory canals, and wood anatomy confirm Skottsberg’s (1929) contention that Yunqu(o must be related to Cen tamocie’ndron. The coincidence of these anatomical features could hardly be explained in any other way. Similarly, Skottsberg’s decision to erect a new genus from the rather fragmentary material of Yunqueo he received seems well justified in terms of anatomical data. The differences in leaf, cortex, and node anatomy are minor. The divergences in wood anatomy are somewhat greater. These differences, taken as a whole, are approximately what one would expect of plants which are generically different. On account of virtual identity in leaf, leaf base, node, and cortex anatomy, the new species Ce’ntourodendron palmiforme is correctly placed in that genus rather than in Yunquea. In some anatomical characters, however, it resembles Vu uquca more closely than does C. dracaeno ides. The lack of pith selereids, absence oE apotracheal parenehyma. and intermediacy in certain dimensions of tracheary cells between Van qvea and C. dracaenoides indicate this tendency. Other wood characters, such as the ray anatomy and fiber width, clearly mark it as referable to Centaurodenciron.

LITERATURE CITED Carlajiist, S. 1936. On the generic limits of Friophyltum (Compositne) and related genera. Madroflo 13:227—239. 1957a. Systematic anatomy of I[esperornaaniu. Pacif. Sd. 11:207—215. 1957b. Wood anatomy of Mutisiene (Compositae). Trop. Woods 106:29—45. 1957c. The genus Fitchia (Compositme). Univ. Calif. Pnhl. hot. 29:1—144. • 1958a. Aiiatomy of Iuayana Mutisieae. Part TI. Meni. N. Y. hot. Gard. 10:00 00. • 1 958b. Wood anatomy of Hehiautheae (Compositae). Trop. Woods 108:00 00.

Col, M. A. 1903. Recherehes sur 1 ‘appareil sieréteur nterne des eomposdes. Jour. (Ic hot. 17:252 318. Daniel, L. 1890. Reehcrehes aiiatomiques et physiologiqu(s sur les braeGes de l’involuere des eomposées. Thesis, (1. Massoii, Paris. Gueguen, F. 1 902. Anatomie do style et du stigmate des phanerogames. Jour. de hot. 16: 300 313. Jeffrey, E. C. 1917. Tli mnatomv of woody p1U1tS. university of Chicago Press, Chicago. Johansen, D. A. 1940. Plant. niicroteelinique. McGraw Hill, New York. 1958] CARLQUIST: CENTAURODENDRON AND YUNQUEA 93

Kunkel, 0. 1957. Beobaehtungen über die Vegetation auf dem Yunque-Massiv. Bot. Jahrb. 77:149—157. Lavialle, P. 1912. Recherches sur le développement de l’ovaire en fruit chez les eomposêes. Ann. Sd. pat. Bot. sér. IX. 15:39—149. Napp-Zirni, K. 1956. Beitrkge zur Anatomie und Morphologic der Isivolueral- und Spreu blatter der Compositen. Bot. Studien 6:1—116. Skottsberg, C. 1929. Notes on some recent collections made in the islands of Juan Fernandez. Medd. Ghteborgs Bot. Trkdg. 4: 155-171. 1938. On Mr. C. Bock’s collection of plants from Masatierra (Juan Fernandez), with remarks on the flowers of Centanrodendron. Medd. Göteborgs Bot. Trhdg. 12: 361-373. 1953. The vegetation of the Juan Fernandez Islands. Nat. Hist. Juan Fernandez and Easter Isl. 2:793—960. 1957. Une seconde cspêce de Centaurodendron Johow. Bull. Jard. Rot. Brunches 27: 585—589. Van Tieghem, P. 1872. Mémoire sur ls canaux sécréteurs des plantes. Ann. Sei. Nat. Bot. sdr. V. 16:96—201. Wagenitz, G. 1955, Pollenmorphologie und Systematik in der Gattung Centaurra L. s. I. Flora 142:213—279.