OBSERVATIONS ON THE VEGETATIVE ANATOMY OF CREPIDIASTRUM AND DENDROCACALIA (ASTERACEAE)

Although Crepidiastrum Nakai belongs to the tribe Lactuceae whereas Dendrocacalia Nakai is a member of Senecioneae. the two genera have been grouped here because both form woody species on offshore islands of Japan. As such, they provide material for further studies in wood anatomy and vegetative anatomy of Asteraceae; neither genus has been studied anatomically hitherto. Crepidiastrum consists of about eight species in eastern Asia, from Korea through Japan (including the Bonin Islands) to Taiwan (Walker 1976). Most of these occur on rocky places near the sea (Ohwi 1965). Crepidiastrum lanceolatum (Houtt.) Nakai in the present study is representative of these. I collected it on limestone boulders (worn portions of uplifted coral reefs) near the shore on Iriomote. Ryukyu Islands. This species and other beach species are not truly woody; they are rosette perennials with succulent stems. Shoots with elongate internodes are sometimes produced, but the plant body as a whole may be deemed succulent rather than woody and the habit designated as a subshrub. Understanding of how this habit and texture relate to anatomy is one of the goals of this study. Crepidiastrum ameristophyllum (Koidz.) Nakai is a rosette shrub, sparsely branched and sometimes with a single stem (Fig. 5). The stem has a woody texture. Crepidiastrum ameristophyllum occurs on the summit ridge of Hahajima. Bonin Islands. This habitat is markedly unlike the seashore habitats of the other species. The summit of Hahahjima is a humid cloud forest where periods of sunshine are limited. The marked difference in habit between C lanceolatum and C ameristophyllum invites comparison as a way of seeing how wood anatomy may shift with relation to habit. The fact that C ameristophyllum is so restricted geographically, whereas the other species are more widely distributed and also so relatively uniform in their rosette-subshrub habit suggests that the habit of C ameristophyllum might be derived from the habit of the other species. There are no cytological differences or other differences of an irreversible nature to give independent confirmation of the direction which has been followed in evolution of habit in this genus.


INTRODUCTION
Although Crepidiastrum Nakai belongs to the tribe Lactuceae whereas Dendrocacalia Nakai is a member of Senecioneae. the two genera have been grouped here because both form woody species on offshore islands of Japan. As such, they provide material for further studies in wood anatomy and vegetative anatomy of Asteraceae; neither genus has been studied anatomically hitherto.
Crepidiastrum consists of about eight species in eastern Asia, from Korea through Japan (including the Bonin Islands) to Taiwan (Walker 1976). Most of these occur on rocky places near the sea (Ohwi 1965). Crepidiastrum lanceolatum (Houtt.) Nakai in the present study is representative of these. I collected it on limestone boulders (worn portions of uplifted coral reefs) near the shore on Iriomote. Ryukyu Islands. This species and other beach species are not truly woody; they are rosette perennials with succulent stems. Shoots with elongate internodes are sometimes produced, but the plant body as a whole may be deemed succulent rather than woody and the habit designated as a subshrub. Understanding of how this habit and texture relate to anatomy is one of the goals of this study. Crepidiastrum ameristophyllum (Koidz.) Nakai is a rosette shrub, sparsely branched and sometimes with a single stem (Fig. 5). The stem has a woody texture. Crepidiastrum ameristophyllum occurs on the summit ridge of Hahajima. Bonin Islands. This habitat is markedly unlike the seashore habitats of the other species. The summit of Hahahjima is a humid cloud forest where periods of sunshine are limited. The marked difference in habit between C lanceolatum and C ameristophyllum invites comparison as a way of seeing how wood anatomy may shift with relation to habit. The fact that C ameristophyllum is so restricted geographically, whereas the other species are more widely distributed and also so relatively uniform in their rosette-subshrub habit suggests that the habit of C ameristophyllum might be derived from the habit of the other species. There are no cytological differences or other differences of an irreversible nature to give independent confirmation of the direction which has been followed in evolution of habit in this genus.
The distinctive habitats of C. lanceolatum and C. ameristophyllum, re- While the larger sheathing leaf bases of the C. ameristophylhim material might well be expected have more numerous veins that the C. lanceolatum material, in which leaf bases were narrower, variation in vein number and therefore nodal type is to be expected based on wideness of leaf base.
The stem of Dendwcacalia crepidifo/ia has a cylinder of bundles, with no pith bundles. Fibers are present as bundle caps on some bundles. Secretory canals are present adjacent to phloem of some bundles. The cortex is largely composed of cells grading from lacunar collenchyma near the outside to ordinary parenchyma near the cylinder of bundles. Nodes are multilacunar: five traces related to five gaps occurred in the material studied.
Secondary xylem. -Crepidiastmm lanceolatum (Carlquist 15679), stem (Fig. 2-4, 16). Growth rings evident (Fig. 2), most often as latewood poorer in fibers than earlywood. Vessels angular to round in transection. Mean vessel diameter 36.5 nm. Mean number of vessels per mm 2 . 362. Mean number of vessels per group = 3.2. Vessels mostly arranged in radial chains. Mean vessel wall thickness. 4 jum. Mean vessel-element length. 416 /xm. Lateral wall pitting of vessels essentially scalariform (Fig. 16). although more properly termed pseudoscalariform because of probable derivation from alternate pits much widened laterally. Perforation plates simple. Axial xylem composed largely of axial parenchyma, with libriform fibers present in annual bands. Mean libriform fiber diameter at widest point, 21 nm. Mean libriform fiber wall thickness. 2 fxm. Libriform fiber length not determined because of paucity. Axial parenchyma type not separable from the abundant zones of parenchyma. The abundant axial parenchyma also merges imperceptibly into upright ray cells, so that limits of rays are not visible and therefore dimensions cannot be determined. For the same reason, uniseriate rays are rarely discernible. Ray cell walls are thin walled and nonlignified (Fig. 3). Strands of interxylary phloem present in a few places in older secondary xylem ( Fig. 2. 3, 4). These strands of phloem do not occur in the most recent year's secondary xylem, and evidently are developed by means of subdivision of files of axial parenchyma cells. Wood not storied.

CONCLUSIONS
Leaf anatomy.-The leaf anatomy of Crepidiaslrum lanceolatum contrasts with that of C. ameristophyilum in ways correlated with the habitats of the two species. The leaf of C. lanceolamm is amphistomatal. Because this species grows on coral rocks, its leaves receive bright illumination from below even though they tend to be borne horizontally. The fact that mesophyll in this species is not clearly differentiated into palisade and other (spongy) chlorenchyma types prevents one from designating the leaf as clearly isolateral despite its amphistomatal condition. Xeromorphic features of the leaf of C. lanceolatum include sunken stomata. greater thickness of epidermal walls (as compared to C. ameristophyllum), more compact chlorenchyma with smaller air spaces, and a slightly greater leaf thickness. The leaf anatomy obviously matches the contrasting habitats of the two species, respectively. This reinforces the obvious difference in leaf size: the leaves of C. ameristophyllum are mesomorphic in having appreciably larger dimensions compared with the leaves of the seashore species.
The leaves of Dendrocacalia are likewise mesomorphic in their dimensions. The occurrence in Dendrocacalia leaves of secretory canals is not unexpected in view of the distribution of secretory canals in the family (e.g.. Col 1899). The occurrence of hypodermis in leaves of Dendrocacalia may be exceptional for Asteraceae as a whole. However, hypodermis in leaves of Asteraceae has been reported in genera which occur in humid (but probably bright) tropical areas, such as Petrobium and Oparanthus (Carlquist 1957) or various of the Mutisieae from the Guayana highlands of Venezuela (Carlquist 1958b).
Anatomy of stems and nodes.-The primary stems of Crepidiastrum are noteworthy in presence of pith bundles. Pith bundles, while unusual in Asteraceae as a whole, have been reported from a scattering of genera, including some in the tribe Lactuceae (Metcalfe and Chalk 1950). The occurrence of multilacunar nodes in Crepidiastrum ameristophyllum and Dendrocacalia crepidifolia relates to leaf size in these two taxa. In turn, leaf size in these two shrubs is probably related, as mentioned above, to a habitat where humidity is high and cloud cover frequent, so that overheating of leaves is unlikely to occur.
Wood anatomy.-The difference between Crepidiastrum ameristophyllum and C. lanceolatum with respect to parenchymatization is quite evident, as comparison of Fig. 2 and 7 reveals. This is a demonstration of succulence related to the rosette habit versus woodiness as related to the shrub habit. Moreover, the ease with which a transition from one to the other has occurred is evident, judging from the coexistence of the two in the same genus.
The presence of phloem strands in older secondary xylem of C. lanceolatum is quite distinctive: such interxylary phloem has not been reported in Lactuceae before. In three genera of the tribe Inuleae. Elytropappus. Metalaisia, and Phaenocoma, Adamson (1934) has reported interxylary phloem. However, in these three genera the pattern is not like that of Crepidiastrum; rather, those genera have cambia which produce cork to the outside and xylem containing interxyiary phloem strands to the inside. In this respect, the three genera of Inuleae show more resemblance to Stylidium (Carlquist 1981) than to Crepidiastrum.
The presence of scalariform perforation plates (with various degrees of modification) in a sizeable proportion of perforation plates of Crepidiastrum ameristophyllum vessel elements requires interpretation. One possible explanation applicable to C. ameristophyllum is given in the case of Patrinia vi/losa Juss. in an accompanying paper (Carlquist 1983). According to the data of Bierhorst and Zamora (1965), certain Lactuceae have scalariform perforation plates in primary xylem. Through paedomorphosis, this primary xylem pattern might be protracted into the secondary xylem. Such juvenilism can be deemed likely because the stems of C ameristophyllum never exceed two to three centimeters in diameter. Scalariform perforation plates are not of negative selective value in this species because it grows in humid mesic sites where flow of large volumes of water per unit time may never occur; if rapid transpiration and therefore flow of large volumes of water occurred, selection for simple perforation plates would be expected to occur. Consequently, a juvenile perforation plate pattern is spread into secondary' xylem. where simple perforation plates (in accordance with the almost universal occurrence of simple plates in wood of the family) would ordinarily be expected. Simple perforation plates do occur without exception in wood of C. lanceolatum, which occurs in more xeric sites. Modified scalariform perforation plates in secondary xylem of other Lactuceae have been reported in the case of Dendroseris sect. Phoenicoseris (Carlquist 1960). These are insular species of moist forest, a habitat much like that where C. ameristophyllum grows. One may note that the scalariform perforation plates of C. ameristophyllum are often much modified, with bars running in tangential directions (Fig. 10) rather than in radial orientation frequently. This argues in favor of a juvenilistic overlay concept in the secondary xylem rather than the alternative idea, a phylogenetic retention of scalariform perforation plates as a feature inherited from ancestors in which scalariform perforation plates were present in all secondary xylem vessels. The reader may have noted that the lateral wall pitting of vessels in C. ameristophyllum is alternate, not pseudoscalariform. as one might expect of vessels juvenilistic in all respects. However, the shrubby habit of C. ameristophyllum probably does place selective pressure on mechanical strength of vessel elements as well as libriform fibers, and alternate circular bordered pits represent a configuration of maximal strength. On the other hand, the lateral walls of vessels in C. lanceolatum bear pseudoscalariform pits (Fig. 16). This pitting, considered a juvenilistic feature (Carlquist 1962a), has been protracted into secondary xylem in C. lanceolatum because, in my interpretation, selection for mechanical strength is lowered in succulent forms. This interpretation has been offered earlier for plants with similar growth forms (Carlquist 1975). Thus, the role paedomorphosis plays in the secondary xylem is not an inflexible one. but species which show paedomorphosis will show various degrees and kinds of retention ofjuvenile features depending on the selective value of any given feature in any particular species.
The woods of Crepidiastrum and Dendrocacalia invite ecological interpretation on account of the distinctive habitats they occupy. The index I have termed Mesomorphy earlier (Carlquist 1977) can be used to compare these woods. The Mesomorphy value for Crepidiastrum lanceolatum is 42; that of C. ameristophyllum is 109 (stem) or 95 (root). Thus, the wood of C. ameristophyllutn does seem to reflect the more mesic habitat of that species as compared to C. lanceolatum. The difference in figures between root and stem of C. ameristophyllum are negligible and can be considered statistically the same. The wood of Dendrocacalia crepidifolia has a Mesomorphy value of 493 (small stem) or 630 (large stem), figures remarkably high, even for Asteraceae. Although C. ameristophyllum is geographically in the same area as D. crepidifolia, C. ameristophyllum is in more open, unstable areas whereas D. crepidifolia is in intact cloud forest. The large size of Dendrocacalia plants (compared to plants of Crepidiastrum) means that Dendrocacalia roots probably tap deeper soil levels which are perpetually moist, whereas the shallower soil levels in which C. ameristophyllum roots grow very likely fluctuate more more in moisture content. Thus, the high Mesomorphy value of D. crepidifolia compared to C. ameristophyllum is understandable.
The Mesomorphy value of D. crepidifolia shows increase with age of a stem; outer wood is more mesomorphic than inner wood. This is apparent not merely in vessel diameter, but in vessel-element length also. This trend with age of stems can be found not merely in this species, but in dicotyledons at large. The reason is not obscure. The Mesomorphy index, reflecting increase in vessel diameter as it does, reflects increased transpiration in that feature. Thus, the Mesomorphic index would be expected to advance as plants increase in height (foliage exposed to sunlight more) and roots lengthen (sources of water less subject to fluctuation available). Changes in wood of Dendrocacalia with age also include widening of rays; this is a common tendency in dicotyledon woods, as noted by Barghoorn (1941) and subsequent authors.
Increase in stoning with age, observed in Dendrocacalia crepidifolia, is to be expected in view of Bailey's (1923) account of storied cambium. In its storied cambium and wide multiseriate rays, the wood of Dendrocacalia crepidifolia closely matches that of other Scnecioneae, such as Senecio ecuadoriensis Hieron. and S. huntii F. Meull. (Carlquist 1962b). This pattern is probably common in Senecioneae and resemblance of any pair of taxa should not necessarily be taken as indicative of close relationship.