Aliso: A Journal of Systematic and Floristic Botany Aliso: A Journal of Systematic and Floristic Botany

Wood anatomy of the recently described Degeneria roseiflora differs from that of D. vitiensis by possessing narrower vessels, much thicker-walled vessels and fiber-tracheids, abundant uniseriate rays, and greater numbers of ethereal oil cells in rays. Because both large and smaller wood samples of D. vitiensis were studied, ontogenetic changes in the wood are presented and separated from those features that probably vary with the species. Tyloses and perforated ray cells are newly reported for Degeneria. Anatomy of mature bark of D. roseiflora is described. Wood anatomy of Degeneria is moderately primitive. Although Degeneria is often compared to Himantandraceae and Magnoliaceae, Eupoma tiaceae also seem very close, if not closer.

The discovery of a new species, Degeneria roseif/ora J. M. Miller (Miller 1988) and the availability of wood and bark material of this species have provided an opportunity for study of bark of Degeneria, hitherto little known, aiid for comparison of wood of the two species. John M. Miller kindly placed liquid-preserved material of wood and bark of D. roseif/ora (Miller 1200) at my disposal. 'The wood sample, collected in 1987, was 66 mm in diameter; the bark was 6 mm thick.
The wood of D. vitiensis studied here comes from a mature wood sample (R 1193-1) provided by the Forestry Commission of New South Wales. The rays of this wood block are parallel, indicating that the sample was obtained from a large trunk. In contrast, a wood sample 33 mm in diameter (Carlquist 695) representing a basal shoot from an uninjured tree, provided a relatively small accumulation of secondary xylem. The latter wood sample was collected in Fiji in 1962, thanks to the aid of John W. Parham. These two wood samples offer contrast in age and offer a way to determine which wood features alter with ontogeny in Degeneria wood. Miller (1988) suggests that D. roseif/ora may have neotenic features in its floral structures. One would not necessarily expectjuvenilistic features in the sense of paedomorphosis in wood of this species, however, because typically woody dicotyledons do not show paedomorphosis in wood features as do dicotyledons with special growth forms (e.g., rosette trees, stem succulents; for a review, see Carlquist 1988). In any case, the wood available of D. vitiensis permits study of change in wood features from pith to the outside of a large stem; the wood sample of D. roseiflora is intermediate in size between the two D. vitiensis samples studied, and thus differences among the samples that are related to ontogeny can be differentiated from those that may be related to species limits.

MATERIALS AND METHODS
The wood sample Miller 1200 of D. roseiflora was preserved in formalin-aceticalcohol in the field, whereas wood samples of D. vitiensis were dried. The latter wood samples were boiled in water and stored in 50% aqueous ethyl alcohol.
Woods were sectioned on a sliding microtome. Sectioning by means of a sliding microtome proved unsuccessful for sampleR 1193-1 of D. vitiensis because of the thin-walled nature of wood cells. For that sample, therefore, an alternative technique that involves softening in ethylene diamine followed by sectioning in paraffin (Carlquist 1982) was employed. This technique provided the sections illustrated in Figures 10-14. This method was also used for the bark sections of D. roseiflora illustrated in Figures 15-18. Wood sections were stained in safranin or in a safranin-fast green combination. Macerations were prepared with Jeffrey's solution and stained with safranin.
Some sections of D. roseiflora wood cut with a sliding microtome were placed between glass slides and allowed to dry. These sections were observed with an ISI WB-6 scanning electron microscope ( Fig. 5-7).
Wood terminology follows that of the IAWA Committee on Nomenclature (1964). Vessel diameter is measured as lumen diameter at widest point. All quantitative data are based upon 25 measurements per feature except for vessel wall thickness, fiber-tracheid wall thickness, and fiber-tracheid diameter at widest point; for these three features, figures for typical conditions rather than means were obtained (e.g., nonobliquely sectioned cells selected; cell wall thickness measured not at cell comers). Number of vessels per group is calculated as: a solitary vessel = 1, a pair of vessels in contact = 2, etc. In addition to observations made on outer stems, observations were made on wood of twigs of D. vitiensis in order to find the most juvenile expressions of ray types and ray histology, although no quantitative data were computed from these. Specimens documenting ·the Carlquist and the Miller collections are located in the herbarium of the Rancho Santa Ana Botanic Garden.

Bark
The bark of D. roseiflora is illustrated in Figures 15-18. Outer bark (Fig. 15) shows disruptions because of formation of successive periderms; bark tends to form breaks at the phellogen in sectioning. Periderm formation occurs only in outer bark; Figure 16 shows periderm only at top. Periderm typically consists of four or five layers of phellem, and, internal to the phellogen, a single layer of phelloderm.
Axial portions of secondary phloem consist of a succession of plates of fibers, alternating with thin-walled cells (Fig. 16). The fibers are thick walled (Fig. 18,  top). The thin-walled cells consist of sieve tube elements, companion cells, and phloem parenchyma. The phloem parenchyma may accumulate massive deposits of dark-staining compounds (Fig. 18), and some of the phloem parenchyma cells acquire lignified walls (Fig. 18: near bottom, lower right).
Phloem rays consist of thin-walled cells that tend to be stretched tangentially as stem diameter increases (Fig. 16). A few of these cells acquire lignified walls (Fig. 17, upper left). Druses are common in phloem ray cells (Fig. 17).

DISCUSSION
The three collections of Degeneria studied represent ontogenetic stages in the sequence D. vitiensis (Carlquist 695), D. roseiflora (Miller 1200), and D. vitiensis (SFCw-R 1193-1). Quantitative data when arranged in this order oan be used to demonstrate changes during ontogeny of particular wood features. In addition, a few sections from the center of stems of D. vitiensis and D. roseiflora were used to determine the earliest point in ontogenetic sequences of these characters. Features subject to ontogenetic change are discussed first, so that a residue of features by which the two species might differ then can be considered.
Quantitative features that show change include (least juvenile expression given): longer vessel elements, longer fiber-tracheids, taller multiseriate rays, and wider multiseriate rays. The changes in ray dimensions are apparently minimal, however. Ray histology shifts to a high proportion of procumbent cells (upright cells may be found in the multiseriate portions of multiseriate rays in younger stems of D. vitiensis). In rays of the largest stem studied, cells are exclusively procumbent except for a few sheathing cells and tip cells (cells at the upper and lower tips of rays), and the procumbent cells are markedly elongate radially. Uniseriate rays are uncommon in D. vitiensis, but grow even less common (essentially they are absent) as ontogeny proceeds. All of these tendencies are in accord with Barghoorn's (1941) findings on ray ontogeny. The quantitative data offered by Taka- Fig. 15-18. Transections of bark of Degeneria roseiflora (Miller 1200).-15 . Outer portion of bark (surface above); break is present in a periderm, center.-16. Inner portion of bark (innermost periderm above), showing plates of fibers in secondary phloem.-17. Ray cells from secondary phloem, showing lignified cell (above) and druses (center, below). -18. Portion of axial secondary phloem; fibers, above; dark-staining deposits in phloem parenchyma cells, center; phloem parenchyma with lignified cell walls, near bottom. (Fig. 15, 16, magnification scale above Fig. I; Fig. 17, 18, scale above Fig. 17 [divisions= 10 f.LID].) hashi (1985) suggest that he was studying relatively small stems; he reports mean vessel element length shorter than that of the sample Car/quist 695, and he found square as well as procument cells in the multiseriate portions ofmultiseriate rays.
Features by which the two species differ that seem unrelated to ontogenetic change include (expressions in D. roseif/ora given): narrow vessels, thicker-walled vessels, thicker-walled fiber-tracheids, greater abundance of pits on fiber-tracheids, and greater abundance of ethereal oil cells in rays. Uniseriate rays are appreciably more common in the D. roseif/ora sample, despite the fact that its size is greater than that of Car/quist 695 (which has fewer uniseriate rays; R 1193-1 has essentially no uniseriate rays); as noted above, uniseriate rays decrease in number with age. None of the features by which the two species differ offer a true presenceversus-absence kind of contrast, and all are differences of degree. The differences between the two species do not seem related to differences in ecology; from the observations of Smith (1981) and Miller (1988), the ecology of the two species is similar. Lemesle and Duchaigne (1955a, b) use terms that suggest new or uncommon phenomena, but I believe their observations can be explained more simply. The "parenchymatous cells" with bordered pits they cite could be either axial parenchyma cells or ray cells. Both kinds of parenchyma cells in wood have borders much more commonly than is indicated in wood literature at present, although the pits in axial parenchyma and ray cells of Degeneria, where bordered, are much less conspicuously bordered than those of many dicotyledons. Bordered pits in parenchyma cells of wood are best observed in sectional view; the borders are very poorly seen if the pits are studied in face view, and that accounts for lack of mention of bordered pits on those cells in the literature (see Carlquist 1988). The "pseudotracheids" these authors claim may be axial parenchyma cells with bordered pits. Takahashi's (1985) claim that pits of fiber-tracheids of D. vitiensis could be either bordered or simple could not be confirmed. The possibility remains that borders, always quite inconspicuous on fiber-tracheids of Degeneria, may be present but not easily seen and therefore overlooked.
The low degree of grouping of vessels in Degeneria accords with the mesomorphic ecology in which this genus occurs; grouping is only a little more than random adjacence (if one calculates for the given density and vessel diameter what random placement of vessels would be produced: David A. Hoekman, unpublished). Woods with tracheids tend to have minimal grouping of vessels, and in woods with fiber-tracheids rather than tracheids, vessel grouping is a little more than random distribution would dictate even in woods from very mesomorphic areas (Carlquist 1984).
The wood of Degeneria is less primitive than that of some other dicotyledons, such as Euptelea or Illicium. Degeneria wood is specialized in ray histology, corresponding to Heterogeneous Type liB, tending toward Homogeneous Type II. Other relatively specialized features include presence of fiber-tracheids rather than tracheids, presence of various types of parenchyma aggregation (e.g., banded) and the only moderately long perforation plates (fewer than 25 bars). The comparisons offered by Bailey, Nast and Smith (1943) and by Canright (1955) between Degeneriaceae and Magnoliaceae or Himantandraceae are based in part on similar