Pacific Science (1998), vol. 52, no. 4: 356—368 © 1998 by University of Hawai’i Press. All rights reserved

Wood Anatomy of (: Madiinae) in Relation to Adaptive Radiation’

SiRwIN CLQuIsT2

ABSTRACT: Qualitative and quantitative features are reported for stem wood of 13 collections of 12 species of the Hawaiian genus Dubautia. Although the species share a basic wood plan, quantitative expressions range widely, espe cially with respect to vessel element dimensions, vessel density, vessel grouping, length of libriform fibers, and dimensions of multiseriate rays. Ecology and habit explain most of the diversity. Variations in the ratio between vessel ele ment length and libriform fiber length are correlated with habit both within Dubautia and when Dubautia is compared with and Wilkesia. Other variation in wood is related mostly to ecology. The Dubautia species of wet forest have high mesomorphy ratio values. Low mesomorphy ratio values occur in species of recent or dry lava (e.g., D. scabra) or dry alpine areas (D. menziesii); mesomorphy ratio values in the xeric species are comparable with those in Argyroxiphium. Highly xeromorphic wood in the bog species D. waia lealae may reflect recent immigration from a dry habitat or peculiar features of the bog habitat. The lianoid D. 1atfolia has notably xeromorphic wood, which may reflect recent entry into wet forest or else the tendency for lianas in general to have xeromorphic features that confer conductive safety. All species of Dubautia show fiber dimorphism. Dubautia is a superb example of adaptive radiation, in contrast to the Hawaiian Schiedea (Caryophyllaceae), which has shifted into various habitats with little change in wood anatomy, or the Galapagos genus Scalesia, all species of which must survive periods of drought and have xeromorphic wood.

HAWAIIAN MA1IIN (Argyroxiphium, Du Wood anatomy has proven to be a sensi bautia, Wilkesia) constitutes a monophyletic tive indicator of adaptation to diverse eco group (Baldwin et al. 1990, Baldwin and logical regimes and may be seen with special Robichaux 1995) often cited as a premier ex clarity in families and genera that have di ample of adaptive radiation in (e.g., versified in recent geological time, with little Cariquist 1965, 1970, Carr 1985, Robichaux attendant extinction (Carlquist 1975). Thus, et al. 1990). Argyroxihium, with five species, we would expect the wood anatomy of Du and Wilkesia, with two (Carr 1985), show bautia to show adaptive radiation in wood distinctive aspects of this adaptive radiation, exceptionally well. Dubautia ranges from but the diversity of the 21 species now rec subshrubs on hot, dry lava (D. linearis, D. ognized within Dubautia (Carr 1985, 1990) scabra) to shrubs of mesic forest (D. micro illustrates entry into a wide range of habi cephala, D. plantaginea), rain forest trees (D. tats with attendant change in morphological, knudsenii), bog shrubs (D. waialealae), alpine anatomical, and physiological expressions. shrubs (D. menziesii), and a lianoid shrub (D. latfolia). The habitats and localities for Dubautia species are well described by Carr 1 Manuscript accepted 2 April 1998. (1985, 1990). 2 Barbara Botanic Garden, 1212 Mission Can yon Road, Santa Barbara, California 93105. Address Earlier studies that included data on wood for correspondence: 4539 Via Huerto, Santa Barbara, of Dubautia (Carlquist 1958, 1974, 1994) California 93110-2323. dealt with a small number of the species and 356 New Species of the Gecko Genus Bavayia—BAuER FE AL. 355 covered in New Caledonia are cryptic spe checklist and key to the herpetofauna of cies, diagnosable on the basis of minor sea New Caledonia, with remarks on bioge lation differences and molecular characters, ography. Proc. Calif. Acad. Sci. 47:17— others, like B. exsuccida and B. puichella and 45. the skinks Simiscincus aurantiacus (Sadlier BouciiEr, P., T. JAFFRE, and J.-M. VEILL0N. and Bauer 1 997b) and Lacertoides pardahs 1995. extinction in New Caledonia: (Sadlier et al. 1997), are relatively distinctive. Protection of scierophyll forests urgently The absence of such novelties from existing needed. Biodiv. Conserv. 4:415—428. museum collections is indicative that the dis JAFFRI, T., P. M0RAT, and J.-M. VEILL0N. covery and description of endemic lizards 1993. Etude floristique et phytogéographi from New Caledonia is as much, if not more, que de la forét sclérophylle de Nouvelle a function of prior undersampling as it is of Calédonie. Bull. Mus. Natl. Hist. Nat., current scrutiny of accumulated collections. 4th sér., sect. B (Adansonia) 15: 107—147. MoiAT, P., T. JAFn, J.-M. VEILL0N, and H. S. M.&cKi. 1986. Affinités floristiques et ACKNOWLEDGMENTS considerations sur l’origine des maquis miniers de la Nouvelle-Calédonie. Bull. We thank the New Caledonian author Mus. Nati. Hist. Nat., 4th sér., sect. B ities, especially Marcel Boulet (Province Sud) (Adansonia) 8: 133—182. and Christian Papineau (Province Nord) for O.R.S.T.O.M. (Office de la Recherche Scien permission to collect and conduct research in tifique et Technique Outre-Mer). 1981. New Caledonia. Jean Chazeau and the staff Atlas de la ,Nouvelle-Calédonie et dé of the ORSTOM Centre de Nouméa are pendences. Editions de l’Office de la thanked for their ongoing assistance to Recherche Scientifique et Technique herpetological research in New Caledonia. Outre-Mer, Paris, France. Specimens of Bavayia puichella were col SADLIER, R. A. 1986. A review of the scincid lected during the Province Nord Biodiversity lizards of New Caledonia. Rec. Aust. Project. Sarah Smith and Hervé Jourdan Mus. 39:1—66. assisted in the field, and line drawings were 1989. Bavayia validiclavis and Ba prepared by Melissa Tarby. The manuscript vayia septuiclavis, two new species of gek was reviewed by Glenn Shea. konid lizard from New Caledonia. Rec. Aust. Mus. 40 [1988]: 365—370. SAt’LIER, R. A., and A. M. BAuL3R. 1997a. LITERATURE CITED The terrestrial herpetofauna of the Loy alty Islands. Pac. Sci. 51:76—90. BAuER, A. M. 1990. Phylogenetic systematics 1997b. A new genus and species of and biogeography of the Carphodactylini lizard (Reptilia: Scincidae) from New (Reptilia: Gekkonidae). Bonn. Zool. Caledonia, Southwest Pacific. Pac. Sci. Monogr. 30:1—219. 51:91—96. 1995. Geckos of the genus Rhaco SLrER, R. A., G. M. SIA, and A. M. dactylus. Reptiles 3(7): 32—34, 36, 38—42, BAuER. 1997. A new genus and species of 44, 46—49. lizard (Squamata, Scincidae) from New BAuER, A. M., and R. A. SA.ELIER. 1993. Caledonia, Southwest Pacific. Zool. Neo Systematics, biogeography and conser caledonica, vol. 4. Mem. Mus. Natl. list. vation of the lizards of New Caledonia. Nat. Ser. A 171:379—385. Biodiv. Lett. 1: 107—122. SADUER, R. A., A. H. WHITAKER, and A. M. 1994. The terrestrial herpetofauna of BAuER. 1998. Lioscincus maruia, a new the Ile des Pins, New Caledonia. Pac. Sci. species of lizard (Reptilia: Scincidae) from 48:353—366. New Caledonia, Southwest Pacific. Pac. BAuER, A. M., and J. V. VIr’nuM. 1990. A Sci. 52:334—341. Wood Anatomy of Dubautia—CLQuIsT 357 with limited data sets. Therefore, a more de basal stems or major branches, portions ob tailed series of studies is attempted here. tainable when these plants were relatively Companion studies, on wood of Argyro common 40 or more years ago. xiphium (Carlquist 1997) and Wilkesia (Carl Wood samples were prepared by drying. quist in press), lend perspective to the ob Voucher specimens are listed in Table I. servations on Dubautia wood. Summaries of habitat information for the Adaptation by wood of dicotyledons to various species given by Carr (1985) are the varied ecological regimes is primarily related best source for comparisons with wood anat to vessel features: vessel diameter, vessel omy, and thus localities of the individual element length, and vessel density (Carlquist specimens are not given here. The wood of 1975). To this list, degree of vessel grouping Raillardiopsis muirii (A. Gray) Rydb. has not should be added (Cariquist 1984). Presence been included in Table 1; it was collected in and nature of growth rings (Carlquist 1980) Tehipite Valley, Middle Fork of the Kings and occurrence of vasicentric tracheids River, 1400 m, Fresno County, California, (Cariquist 1985a) in wood are also indicative J. T. Howell 33960 (RSA). The specimen of of ecology. D. plantaginea (Carlquist 1928), collected in Degree of woodiness (subshrubs, shrubs, Ko’olau Gap, is referable to D. plantaginea trees) is indicative of adaptation to various var. platyphylla Hillebrand, a variety now habitats in a phylad in which secondary placed into synonymy under D. plantaginea woodiness has occurred, as appears to be the subsp. plantaginea by Carr (1985). case in Dubautia and other island genera Portions of dried wood samples were (Carlquist 1974). Secondary woodiness is boiled in water and stored in 50% aqueous characteristic of the Hawaiian Madiinae, ethanol. Wood was sectioned on a sliding judging by their terminal position in clado microtome. Sections were stained either with grams (Baldwin et al. 1990, Baldwin and safranin or with a safranin-fast green combi Robichaux 1995). Basic to the Hawaiian nation. Macerations for measurement of ves genera are subshrubs of very limited wood sel element length and libriform fiber length iness, such as the Californian montane genus were prepared with Jeffrey’s Fluid and Raillardiopsis, which has been included in stained with safranin. this study for purposes of comparison. Means shown in Table 1 are based on 25 Adaptive radiation in wood of Dubautia measurements per feature except for rather should be understood in the context of the variable ones: vessel wall thickness, pit di Madiinae as a whole. ameter, and libriform fiber wall thickness figures are based on conditions judged to be typical. Number of vessels per group is com puted on the basis a solitary = MATERIALS AND METHODS of vessel 1, a pair of vessels in contact = 2, etc. Vessel The wood samples studied represent lumen diameter is used rather than outside mostly wood samples I collected over a vessel diameter because lumen diameter is period of years from naturally occurring considered a better indicator of conductive specimens. To these have been added stems characteristics. The lumen diameter of vessels from a few herbarium specimens. The sample oval in transection was measured as an aver of D. linearis wood was small (3 mm diame age between widest and narrowest diameter. ter), but the sample of D. waialealae, al Wood anatomy terms follow the IAWA though ca 6 mm in diameter, was from the Committee on Nomenclature (1964), except basal stem of a mature plant. The stem of for the terms vasicentric tracheid and vascu D. latfolia was taken from a dead 3-yr-old lar tracheid, which follow earlier usages branch and does not represent a basal stem, (Carlquist 1984, 1985a). Radial sections of but the scarcity of this species does not per all species were prepared, as is typical in mit harvesting of larger wood samples. For studies on wood anatomy, but radial sections all other species, the wood samples represent have not been illustrated here; they yield 358 PACIFIC SCIENCE, Volume 52, October 1998 some kinds of information well (e.g., shape of in large radial multiples together with wide ray cells), but more numerous features rele vessels; the narrow vessels often are difficult vant to the conclusions of this paper are evi to discern because they are similar in diame dent on transections and tangential sections. ter to libriform fibers. The tendency for vessel Statistical treatments other than means grouping to take the form of radial multiples have not been undertaken here because of the is illustrated by D. tatfolia (Figure 5), nature of sampling. Wood collections of Du whereas in D. sherffiana (Figure 9), vessel bautia are few because of rarity of the species groups are about as wide tangentially as and the infrequency with which wood collec radially. tions have been made. For statistical meas Vessel diameter (Table 1, column 2) also urements to be meaningful, one would have ranges widely in Dubautia. Mean vessel di to sample a series of mature specimens of ameter is widest in D. plantaginea (Figure 1), each species. Moreover, wood of branches narrowest in D. waialealae (Figure 13). Other and roots as well as of basal stem portions transections illustrated here show decreasing should be included for a study in which sta vessel diameter within these extremes: D. pa tistical methods are employed. The means leata (Figure 3: 50 pm), D. latfo1ia (Figure 5: given in Table 1 are suggestive of differences 42 pm), D. ciliolata (Figure 7: 39 pm), D. among the species, but do not have the sherffiana (Figure 9: 38 pm), and D. menziesii statistical validity obtainable from extensive (Figure 11: 15 pm). sampling of comparable samples. Statistical Vessel density (Table 1, column 3) is ordi tests can only meaningfully analyze material narily considered inversely proportional to studied; they cannot project what one would vessel diameter. This is approximately true in find in materials not studied. Dubautia (compare columns 2 and 3); in ex amining transections of species in which nar row vessels are common but easily confused with libriform fibers, this correlation not RESULTS may be evident at first. In D. latfolia (Figure 5), Growth Rings the number of vessels per square millimeter is 109, well above the average for Dubautia as a Growth rings were observed clearly only whole (last data line of Table 1), 68, whereas in two species, D. latfolia (Figure 5) and D. the vessel diameter in this species, 42 pm, is scabra. These growth rings can be recognized close to the average for the genus (46 pm). by rings of wide earlywood vessels external This is the strongest deviation in the genus to narrow latewood vessels; latewood vessels away from an inverse relationship between are also more numerous per square milli vessel diameter and vessel density, and is in meter. In species other than D. latfolia and the only species with a lianoid habit, as dis D. scabra, fluctuation with respect to vessel cussed below. diameter occurs, but not in the form of Vessel element length (Table 1, column 4) clearly demarcated growth rings. ranges from 420 pm in D. knudsenii to 174 pm in D. menziesii (Figure 11), and for both of Vessel Elements those species (as for most wood samples studied) mature branches rather than twigs The mean number of vessels per group were studied. Thus, differences among the is shown in Table 1, column 1, and ranges species are mostly not attributable to juve widely within the genus: from 1.3 in D. nilism of wood. Vessel element length will knudsenii and D. plantaginea to more than 10 be discussed with relation to ecology (see in D. waialealae (Figure 13); D. menziesii below). (Figure 11) also has some large vessel group Vessel wall thickness is relatively uniform ings, although the transection shown in Fig for the genus (Table 1, column 5). The thick ure 11 does not suggest this. In D. menziesii walled vessels of D. paleata and D. latfolia and D. waialealae, narrow vessels often occur (Figure 5) differ but little from the conditions TABLE 1

Wooo CHARAcmRIsrlcsOFDubautia

1 2 3 4 5 6 7 8 9 10 11 12 13 SPECIES COLLE(moN VG VD VM VL VW VP LF SF FW MH MW FV MR

D. ciliolata (DC.) Keck Carlquist 2051 (RSA) 1.9 39 36 184 3.0 5 448 229 4.0 1561 8.3 2.43 199 D. knudsenii Hillebr. Cariquist H15 (uc) 1.3 68 13 420 2.3 4 698 472 2.6 4623 3.4 1.66 2197 D. 1atfo1ia (A. Gray) Keck cult. J. Plews 4.2 42 109 296 2.7 5 650 424 5.2 2949 6.6 2.20 114 D. laxa Hook. &Am. Carlquist 1971 (RsA) 1.4 59 14 262 2.5 4 546 305 4.6 576 5.9 2.08 1104 D. linearis (Gaud.) Keck Rock 8123 (aisi-i) 1.8 41 39 179 2.2 4 374 314 4.0 1936 6.8 2.10 188 D. menziesii (A. Gray) Keck Cariquist H17 (uc) 13.4 15 157 173 2.2 5 442 286 3.0 1696 5.4 2.54 17 D. microcephala Skottsb. Cariquist H14 (uc) 2.0 52 20 320 2.2 4 564 390 3.0 601 3.4 1.76 832 D. paleata A. Gray Carlquist 1772 (RsA) 1.9 50 39 332 3.0 6 553 208 2.4 5972 6.3 1.67 426 D. plantaginea Gaud. Carlquist 1928 (RsA) 1.3 67 13 278 2.2 5 603 327 2.5 3892 6.2 2.17 1433 D. plantaginea Gaud. Carlquist 1362 (RsA) 1.5 86 13 324 2.2 5 695 536 5.0 1999 7.1 2.14 2143 D. raillardioides Hillebr. Cariquist H16 (uc) 1.7 42 22 331 2.6 5 654 286 2.4 2050 4.1 1.98 632 D. scabra (DC.) Keck Carlquist 2035 (R.sA) 3.8 32 126 244 1.2 4 469 254 2.5 2252 7.0 1.92 62 D. sherffiana Fosberg Carlquist 2356 (asA) 1.6 38 23 224 2.2 4 539 313 3.0 922 4.4 2.50 370 D. waialealae Rock Rock 5030 (uc) >10 11 333 208 2.5 4 462 193 2.4 1101 6.1 2.22 7 Dubautia collections averaged 3.8 46 68 270 2.4 4.6 558 324 3.1 2295 5.8 2.25 697

KEY TO COLUMNS: 1(VG), mean numberof vesselsper group; 2 (VD), mean vessellumen diameter, pm; 3 (VM), mean number of vesselsper square millimeter; 4 (VL),mean vesselelement length, pm; 5 (VW), mean thicknessof vesselwalls, pm; 6 (VP), mean axialdiameter of lateral wall vesselpita, pm; 7 (LF), mean length of longer libriform fibers,pm; 8 (SF), mean length of shorter libriform fibers,pm; 9 (FW), mean wall thicknessof long libriform fibersat widest point, pm; 10(MH), mean height of multiseriate rays, pm; 11(MW), mean width of multiseriate rays at widest point, pm; 12 (FV), F/V ratio (mean length of longer libriform fibers divided by mean vesselelement length); 13 (MR), mesomorphy ratio (mean vessellumen diameter times mean vesselelement length divided by mean number of vesselsper square millimeter.

pm). 10 division (each 1 = Figure

above

figures

all for

Scale

species.

this in

rays

tall of wide, is typical

(center)

ray multiseriate section; 4. Tangential

groupings.

in larger

rower,

nar

are

vessels

3. Transection; paleata.

D.

3—4.

storied. are fibers libriform section; Tangential 2.

paired. or

solitary

wide, are

vessels

Transection; 1. 1362).

(Carlquist plantaginea D. 1—2. Dubautia. of sections Wood 1—4. Fious FIGuaEs 5—8. Wood sections of Dubautia. 5—6. D. lat(folia. 5. Transection; wide vessels slightly above center of photograph are earlywood vessels indicative of a growth ring. 6. Tangential section; many libriform fibers are septate, only vaguely storied. 7—8. D. ciliolata. 7. Transection; a tangential band of thin-walled libriform fibers just below cen ter. 8. Tangential section; both thick-walled fibers (left) and thinner-walled fibers (right) are storied. Scale for all figures above Figure 1.

1. Figure figures above all for Scale section. the

of portion

a large

occupy

and

tall wide,

are rays section; Tangential 12. upper left). (e.g., strands as but also as bands,

mostly present

fibers

libriform thinner-walled

11. Transection; menziesii. D. 11—12. center). near fibers of strip

(dark

fibers

thicker-walled

and in

of

photograph) (most fibers thinner-walled in both storying section; Tangential 10. tiples).

mul (pore

grouped mostly vessels

Transection; 9. sherffiana. D. 9— 10. Dubautia. of sections Wood 9—12. FIGuREs Wood Anatomy of Dubautia—CA1uQuIsT 363 seen in transections of wood of the other narrow, and thicker-walled libriform fibers species (Figures 1, 7, 9, 11, 13). Likewise, di that exhibit storying less conspicuously or ameter of lateral wall pits on vessels (Table 1, not at all. Fibers intermediate with respect to column 6) is remarkably uniform. The uni these features are relatively uncommon, and formity of vessel wall thickness and lateral a bimodal distribution with respect to the wall pit diameter is one reason for saying that quantitative features is present. The contrast wood of Dubautia as a whole has a common between these two fiber types may be seen plan that has been altered by adaptation here most clearly in the tangential sections of various species to particular ecological of Figure 10 and Figure 12. The bands and regimes. patches of thin-walled fibers for these partic With respect to morphology, all vessels ular species are illustrated in the correspond of Dubautia have simple perforation plates ing transections, Figure 9 and Figure 11. and most frequently are cylindrical with few The phenomenon of fiber dimorphism was “tails.” Lateral wall pits are alternate and first observed in the wood of D. laxa and D. mostly oval, but a few laterally elongate pits menziesii, as well as in the wood of Wilkesia were recorded for D. ciliolata and D. planta gymnox4phium A. Gray (Carlquist 1958, ginea. No forms of helical sculpturing (e.g., 1994), and was assigned the term fiber di “helical thickenings”) were observed on walls morphism by Carlquist (1961). In the mono of secondary xylem vessels. graph on woods of Heliantheae (Carlquist 1958), fiber dimorphism was not observed Imperforate Tracheary Elements in all species of Dubautia. However, better materials and more extensive preparations In wood of D. menziesii and D. waialealae, have permitted me to recognize the occur imperforate fusiform tracheary elements with rence of fiber dimorphism in all of the species bordered pits are present in small numbers. of Dubautia studied here, although it is rela They are associated with narrow vessels tively inconspicuous in some, such as D. and may be considered as vessel elements plantaginea (Figures 1, 2), D. latjfolia (Figure too narrow to bear perforation plates. When 5), and D. paleata (Figures 3, 4). As seen found at termini of growth rings, such cells in transections, the shorter, thinner-walled are termed vascular tracheids. If distributed fibers may occur as long tangential bands throughout growth rings or in species with (Figure 7) or in shorter tangential bands out growth rings, these cells are termed vasi (Figures 9, 11), some of which are very short centric tracheids (Carlquist 1984, 1985a). tangentially and can be termed strands (e.g., Vasicentric tracheids are indicative of xero Figure 11, upper left). Earlier, I concluded morphy in woods (Carlquist 1985a). Vasi that the shorter, thinner-walled fibers should centric tracheids were not observed in woods be called axial (“vertical”) parenchyma of species other than D. menziesii and D. (Carlquist 1958). In this study, I have seen no waialealae. Macerations must be studied to instances in which the shorter fibers are sub ascertain if vasicentric tracheids are present, divided into strands as is the vasicentric axial because small perforation plates can easily be parenchyma in Dubautia (see below), so I missing or overlooked in sections of narrow now consider the shorter fibers still to be lib tracheary elements with bordered pits. riform fibers, although possibly functionally Other than the very small numbers of similar to axial parenchyma. vasicentric tracheids in the two species just In Table 1, mean lengths are given for cited, all imperforate tracheary elements in longer fibers (column 7) and shorter fibers Dubautia are libriform fibers. However, (column 8) for each collection. A few fibers within the wood of any given Dubautia, lib of length intermediate between the two riform fibers are not uniform. There are classes do occur in macerations, and these short, comparatively wide, and thin-walled were omitted from the measurements because libriform fibers that are usually clearly assignment to one class or the other would storied; these contrast with longer, relatively have been arbitrary. These omissions may J

Fioui.as 13—16. Wood sections of Dubautia (13—14) and Raillardiopsis (15—16). 13—14. D. waialealae. 13. Tran section; vessels are narrow (narrowest vessels are narrower than libriform fibers). 14. Tangential section; all fibriform cells in left quarter of photograph are narrow vessels. 15—16. R muirii. 15. Transection; wide rays are present. 16. Tangential section; vague storying present in fibriform cells, left quarter of photograph. Scale for all figures above Figure 1. Wood Anatomy of Dubautia—CLQuIsT 365 have the effect of widening the differences as multiseriate rays and average 188 pm in between the means for the two fiber types in height. any given species, but because intermediate The mean height of multiseriate rays is fibers are relatively scarce, relatively little given in Table 1, column 10. There is no bias is introduced. correlation between height of multiseriate The mean length of shorter fibers in Du rays and length of vessel elements within bautia as a whole is 324 pm; this is somewhat Dubautia. Notably short rays characterize D. greater than the mean vessel element length laxa, D. microcephala, and D. sherffiana. The for the genus (270 pm), indicating that appre tallest rays in the genus were observed in D. ciable elongation of the derivatives of fusi paleata (Figure 4), but tall rays are also form cambial initials does occur as they ma common in D. latfolia (Figure 6), D. planta ture into shorter fibers. Indeed, some shorter ginea (Figure 2), D. raillardioides, and D. fibers characteristically have a wide cylindri scabra. cal body at either end of which is an abruptly Rays occupy a relatively large proportion narrower tail; the tails are interpretable of the wood in D. menziesii (Figure 12) and as products of intrusive growth during cell D. scabra, both of which are small shrubs elongation after derivation from a fusiform or subshrubs. The Californian Raillardiopsis cambial initial. The ratio in length between muirii is relevant in this regard, because it, longer fibers and shorter fibers in the genus as too, is a subshrub and wide rays occupy a whole is 1.7, a ratio sufficiently high that a large proportion of the woody cylinder the dimorphism is evident in numerical (Figures 15, 16). terms. Ray width (Table 1, column 11) varies Wall thickness is given for the longer markedly within Dubautia. Relatively wide fibers only (Table 1, column 9). The range rays may be observed in D. ciliolata (Fig in this feature is from 5.2 pm in D. latfolia ure 8), D. latfolia (Figure 6), D. linearis, D. to 2.4 pm in D. paleata (Figure 3) and D. paleata (Figure 4), D. plantaginea, and scabra. Some septate fibers were observed in D. scabra. D. latjfolia. Ray cell walls are lignified but only mod erately thick in the genus. The range ob Axial Parenchyma served was from 1.5 pm to 2.4 pm. Borders were observed (especially on tangentially Axial parenchyma in Dubautia is vasicen oriented ray cell walls) on many of the pits tric: an incomplete sheath one cell in thick in ray cells of D. knudsenii, D. latfolia, and ness around vessels or vessel groups. The D. sherffiana, and on a smaller proportion of parenchyma occurs in strands of two to four the pits in ray cells of the other species. cells: mostly two in most species studied, but more commonly four cells in D. knudsenii Storying and D. microcephala. Strands of two cells may be seen in axial parenchyma cells in Storied patterns can be seen clearly in the Figure 2 (vertical strip of pale cells at far shorter libriform fibers (e.g., Figure 10), but right). Elongate diagonally oriented pits can also, to a lesser extent, in all fibers (Figures 2, be seen on the walls of axial parenchyma 4, 8, 12, 14). Vessels, vasicentric tracheids, cells, all of which are lignified in Dubautia. and axial parenchyma conform to the storied These diagonal pits face helices of alternate pattern. oval pits on adjacent vessel elements. Deposits and Other Features Rays Oleoresins, commonly seen as droplets or Rays are almost exclusively multiseriate in massive deposits in cells of some Asteraceae, Dubautia except for D. microcephala. In that are scarce in woods of Dubautia. No crystals species, uniseriate rays are nearly as common were observed in wood of Dubautia. A few

bog species the that hypothesis The & erence. (Rock A. kauense 8 in is from range

pref ecological in shift phyletic possibly also 1997): the (Carlquist Argyroxiphium of stems

and a in liana expected features some for hibits obtained those with figures low relatively

ex thus latfolia Dubautia diameter. of series vessel this compare can One (7). lealae

and density vessel between correlation verse waia D. and (62), scabra D. (17), menziesii

in an of basis the on expected than higher D. (114), lat(folia D. (188), linearis D. (199),

density vessel shows that genus the of species ciliolata D. are figures, ratio mesomorphy

sole is the latfolia D. and diameter, vessel small with relatively spectrum, the of end

mean their considering dicotyledons arboreal the other At regard. this in notable are (632)

than density vessel greater have also Lianas raillardioides and D. 2143), (1433, taginea

1985b). (Carlquist noted are lianas which for plan D. (832), microcephala D. (1104), laxa

wide vessels to the addition in D. tracheids (2197), knudsenii D. forest: rain of shrubs

or vasicentric vessels narrow of number large and trees the occur in figures ratio morphy

a have lianas a many liana; in pattern wood meso highest The length. element vessel

xeromorphic a of value selective exemplify and density, vessel diameter, lumen vessel

may it Alternatively, Kaua’i. of region incorporates which 13), 1, column (Table

Koke’e the of forests wetter the into itats ratio mesomorphy the by revealed well are

hab drier from shifted have may subgenus, species Dubautia the of regimes Moisture

its in species sole species, this of ancestors species. Dubautia

The Dubautia. species of other those of of with majority do the than sites windier likely

compared (114) value ratio thus mesomorphy and exposed more occur in which but

low relatively has a it in genus; the shrub of branches, number a limited with shrubs

sole scandent the D. to latfolia, apply small may are relatively that species sherffiana,

explanation similar A island. another from D. and menziesii, D. ciliolata, in D. occur

possibly even or on a dry Kaua’i zone from Dubautia within values F/V highest The

perhaps bog habitat, the to immigrant cent shrubs. small to large from mostly ranges

re relatively a be might waialealae Dubautia that genus a for appropriate is Dubautia

species. any of Dubautia value ratio morphy of value F/V intermediate The genus. that

meso lowest the has ironically, which, but in value F/V higher the with correlate may

1980), (Carlquist rainfall the greatest with sites) exposed in occurs often (which phium

locality the Hawaiian in grows lealae, which gymnoxi W. of stems polelike unbranched

to waia D. might apply considerations same the contrary, the On that. reflect may genus

These surface. foliar small and relatively that fibers in short and the value selective

size bog limited shrubs of characterize not low of likely is strength mechanical stems,

does diameter of wide vessels require would such in condensed; to acaulescent from range

that Transpiration habitat. bog the in geous Argyroxiphium stems of The press). in quist

disadvanta is not wood xeromorphic that (Carl of Wilkesia stems midlevel in 2.86

and recent has been this shift is that 1997) and 1997) (Carlquist in Argyroxihium 1.89

(Carlquist given interpretation The pattern. with compares This 2.25. is the ratio whole,

wood mesomorphic a more to release a by as Dubautia For 12. column 1, Table in

been reflected has not bog habit to the shift Dubautia for given is which length), element

A. to caliginis, gave rise that to plants vessel similar by divided length fiber (libriform

is subsp. macrocephalum A. sandwicense ratio If F/V the regimes: moisture with than

22. Meyrat, (A. Gray) macrocephalum subsp. rather habit with correlates figure One

DC. A. sandwicense silversword, Haleakalã

CONCLUSIONS

the of that to close very is of which 24, value

ratio a mesomorphy has a species, bog bes,

For caliginis Argyroxiphium species. margin

bog a Degener, (Hillebrand) grayanum A. observed. not were Tyloses

in 165 to Degener I. & Degener Neal) M. cells. ray in observed were grains starch small

1998 October 52, Volume SCIENCE, PACIFIC 366 Wood Anatomy of Dubautia—C..iu.QuIsT 367

A. caliginis and D. waialealae have acquired of large numbers of narrow, short vessels that xeromorphic wood characteristics in response offer redundancy and serve for conductive to the bog habitat is conceivable, but this safety where moisture availability fluctuates, does not seem applicable to A. grayanum, a but are of no selective value where soil mois bog margin species, and would require a re ture is abundant. version to xeromorphic wood features from a One notes that vessel dimensions correlate more mesomorphic plan that was ultimately well with leaf area in Dubautia, as illustrated likely derived from ancestors with xeromor for a selection of species earlier (Carlquist phic wood like that of the Californian tar- 1974: 153). Leaf area is an indicator of tran weeds (see below). spirational characteristics in Dubautia, Vessel grouping proves closely correlated several features of which are discussed by with ecology in Dubautia. Only in such a Robichaux et al. (1990). wide-ranging genus as Olearia (Asteraceae) The ecological significance of fiber can one find a greater span among species dimorphism in Dubautia is uncertain. This with respect to number of vessels per group phenomenon is present in Argyroxihium and (Carlquist 1984). When one compares the Wilkesia also (Carlquist 1997; in press), but vessels per group figures for the various spe has not been observed in the Californian cies of Dubautia (Table 1, column 1) with the Madiinae (Carlquist, unpubi. data). Further mesomorphy ratio values (Table 1, column studies of fiber dimorphism both with respect 13), one finds the relationship to be very close to physiology and to details of histology are to inverse. This not only reinforces the con desirable. cept of vessel grouping as a feature related to In summary, adaptive radiation in woods xeromorphy (Cariquist 1984), it also serves of Dubautia is truly exceptional when one as another indicator of adaptive radiation in compares it with wood anatomy of other en wood features of Dubautia. Large groupings demic island genera. For example, Scalesia of vessels potentially safeguard conductive (Asteraceae, tribe Heliantheae) on the Gala pathways during times when high tensions pagos Islands is relatively uniform with re in the secondary xylem might disable many spect to vessel features (Carlquist 1982). The vessels. upland Galapagos forest, habitat for some If as cladograms show (Baldwin et al. Scalesia species, is really not mesic because 1990, Baldwin and Robichaux 1995), mon rainfall is seasonal, so that wood of the tane California subshrubs such as Raillar upland Scalesia species must cope with the diopsis are basal to the Hawaiian Madiinae, demands of a dry season. Schiedea (Cary the features of these subshrubs are essential ophyllaceae) is a Hawaiian genus of shrubs to our understanding of adaptive radiation or subshrubs that is present in both dry and in Dubautia. If the wood of Raillardiopsis genuinely mesic habitats, but its wood anat (Figures 15, 16) is indicative, the Hawaiian omy is xeromorphic with only minimal shift Madiinae began with xeromorphic wood: toward mesomorphy in the species of more vessel elements of narrow diameter and short mesic habitats (Carlquist 1995). Schiedea length, numerous per square millimeter of may be unable to form large shrubs and trees transection, in relatively large groups. Rail because of a combination of features. One lardiopsis and Madia bolanderi have tall, indicator is that Schiedea has rayless woods, wide multiseriate rays and few or no uni and rayless woods are not found in large senate rays, like wood of xeromorphic spe shrubs or trees. Certainly in island genera cies such as D. menziesii or D. scabra. If the that have radiated autochthonously into a wood of Raillardiopsis or Madia bolanderi range of habits and habitats, wood anatomy represents ancestral features, the diversifica may parallel that radiation to various de tion in wood features of Dubautia represents grees. Dubautia is exceptional in the degree widening of vessels to accommodate greater to which adaptive radiation is illustrated by peak transpiration of broader leaves and loss quantitative features of wood anatomy. 368 PACIFIC SCIENCE, Volume 52, October 1998

LITERATURE CITED regions; review of vasicentric tracheids. Aliso 11:37—68. BALDWIN, B., and R. H. RonicHAux. 1995. 1985b. Observations on functional Historical biogeography and ecology of wood histology of vines and lianas: Vessel the Hawaiian (Aster dimorphism, vasicentric tracheids, narrow aceae). New molecular perspectives. Pages vessels, and parenchyma. Aliso 11: 139— 259—287 in W. L. Wagner and V. Funk, 157. eds. Hawaiian biogeography. Evolution 1994. Anatomy of tropical alpine on a hot spot archipelago. Smithsonian plants. Pages 111—128 in P. W. Rundel, A. Institution Press, Washington, D.C. P. Smith, and F. C. Meinzer, eds. Tropical BALDwIN, B., D. W. KYH0s, and J. DvoRAK. alpine environments. Cambridge Univer 1990. Chloroplast DNA evolution and sity Press, Cambridge. adaptive radiation in the Hawaiian silver- 1995. Wood anatomy of Cary sword alliance (Asteraceae—Madiinae). ophyllaceae: Ecological, habital, system Ann. Mo. Bot. Gard. 77:96—109. atic, and phylogenetic implications. Aliso CARLQuIsT, S. 1958. Wood anatomy of 14:1—17. Heliantheae (Compositae). Trop. Woods 1997. Wood anatomy of Argyro 108: 1—30. xiphium (Asteraceae): Adaptive radiation 1961. Comparative plant anatomy. and ecological correlations. J. Torrey Bot. Holt, Rinehart, & Winston, New York. Soc. 124:1—10. 1965. Island life. Natural History In press. Wood anatomy of Wilkesia Press, New York. (Asteraceae) with relation to systematics, • 1970. Hawaii: A natural history. organography, and habit. J. Torrey Bot. Natural History Press, New York. Soc. 1974. Island biology. Columbia CJR, 0. D. 1985. Monograph of the Ha University Press, New York. waiian Madiinae (Asteraceae): Argyroxi • 1975. Ecological strategies of xylem phium, Dubautia, and Wilkesia. Allertonia evolution. University of California Press, 4:1—123. Berkeley. 1990. Dubautia. Pages 292—308 in 1980. Further concepts in ecological W. L. Wagner, D. R. Herbst, and S. H. wood anatomy, with comments on recent Sohmer, Manual of the flowering plants of work in wood anatomy and evolution. Hawai’i. University of Hawai’i Press and Aliso 9:499—553. Bishop Museum Press, Honolulu. 1982. Wood and bark anatomy IAWA COMMITTEE ON NoMENcLATuIu. 1964. of Scalesia (Asteraceae). Aliso 10:301— Multilingual glossary of terms used in 312. wood anatomy. Verlagsbuchanstalt Kon 1984. Vessel grouping in dicotyledon kordia, Winterthur, Switzerland. woods: Significance and relationship to RonIcHAux, R., G. D. CAIu, M. LIEBMAN, imperforate tracheary elements. Aliso and R. W. PnRcY. 1990. Adaptive radia 10: 505—525. tion of the Hawaiian silversword alliance • l985a. Vasicentric tracheids as a (Compositae—Madiinae): Ecological, drought survival mechanism in the woody morphological, and physiological diver flora of southern California and similar sity. Ann. Mo. Bot. Gard. 77:64—72.