Identification of Japanese Species of Evergreen Quercus and Lithocarpus (Fagaceae)

IAWA Journal, Vol. 32 (3), 2011: 383–393

Identification of Japanese species of evergreen Quercus and Lithocarpus (Fagaceae)

Shuichi Noshiro1 and Yuka Sasaki2

SUMMARY To identify archaeological oak woods with very large vessels (> 200 µm), the wood structure of eleven species of evergreen Quercus and Litho- carpus from Japan were studied. Species groups could be identified by the size and frequency of vessels and the ray structure. Quercus phillyraeoides of subg. Sclerophyllodrys had semi-ring-porous wood with small (< 100 µm on average), numerous vessels, and aggregate rays. Two species of Lithocarpus had aggregate to semi-compound rays that came to be divided by the development of vertical masses of fusiform cells. Among species of Quercus subg. Cyclobalanopsis, Q. gilva, Q. hondae, and Q. miyagii had very large vessels with a maximum vessel diameter over 200 µm. Within the species groups, individual species could not be identified just from wood structure, butQ. gilva could be distinguished when the distribution ranges of species were considered. The vertical splitting of semi-compound rays in Lithocarpus with the formation of a vertical wedge of fusiform cells differed from the ray development so far reported in Fagaceae or other taxa that have broad rays, and occurred only in the subgenus Pasania. Key words: Fagaceae, identification, Japan,Lithocarpus , Quercus subg. Cyclobalanopsis, wood structure.

Introduction

When rice cultivation was introduced into Japan around the beginning of the Yayoi period (c. 500 years BC), wood of evergreen oaks was selected to make various tools for agriculture and processing such as hoes, spades, mallets, and axe handles (Ito & Yamada 2011). While identifying materials of wooden tools of the subsequent, early Kofun period (late third to mid seventh centuries AD), the authors noticed that oak wood with very large vessels (> 200 µm) was exclusively selected for hoes and spades. To identify this distinct wood, the wood structure of Japanese species of evergreen Quer- cus and Lithocarpus was studied. These include two species of Lithocarpus, L. edulis (Makino) Nakai and L. glaber (Thunb. ex Murray) Nakai, eight species of Quercus subg. Cyclobalanopsis, Q. acuta Thunb., Q. gilva Blume, Q. glauca Thunb., Q. hondae Makino, Q. miyagii Koidz., Q. myrsinifolia Blume, Q. salicina Blume, and Q. sessilifolia Blume, and one species of Quercus subg. Sclerophyllodrys, Q. phillyraeoides A. Gray

1) Forestry and Forest Products Research Institute, Tsukuba, Ibaraki 305-8687, Japan [E-mail: [email protected]]. 2) Paleo Labo Co., Ltd., Shimomae 1-13-22, Toda, Saitama 355-0016, Japan.

Downloaded from Brill.com09/29/2021 03:03:38PM via free access 384 IAWA Journal, Vol. 32 (3), 2011

(Table 1; Ohba 2006). Most of the studied species grow southwestward from middle Honshu to Kyushu, some growing also in the Nansei islands (Fig. 1). Two species have narrower distribution ranges than other species, Q. hondae in southwestern Shikoku and Kyushu and Q. miyagii in Nansei islands. The wood structure of Japanese species of oaks has been studied by Onaka (1939) and Shimaji (1954, 1956, 1959). To identify archaeological woods, Onaka (1939) compared features of vessels and rays of six species of subg. Cyclobalanopsis and one species of Lithocarpus that grew in the Yamato district around an archaeological site in middle Honshu. He found that the size and arrangement of vessels, the size of axial parenchyma cells, and the occurrence and size of prismatic crystals in axial parenchyma can be used to distinguish species. However, his observation of wood anatomical features is presented only in a table without any information about the specimens studied, and it is not known how much of the species-level variation is reflected in his observations. To clarify systematic patterns in the wood structure of Fagaceae, Shimaji (1954, 1956, 1959) studied the wood structure of Japanese species of Quercus and Lithocarpus and

Table 1. Japanese species of evergreen Quercus and Lithocarpus, details of studied specimens, and the ranges and means (in parentheses) of their vessel features. Sp = specimen number, TVD = tangential vessel diameter.

Species Sp Latitude Longitude Stem TVD Maximum Vessel Vessel (°N) (°E) diam. (µm) TVD frequency area ratio (cm) (µm) (no./mm2) (%) Lithocarpus L. edulis (Makino) Nakai 7 26.8–33.2 128.3–131.5 10–29 77–128 134–184 1.4–5.6 1.4–6.3 (101) (153) (3.9) (3.7) L. glaber (Thunb. ex Murray) Nakai 3 31.9–33.1 131.2–132.7 16–28 101–115 168–200 4.1–5.8 5.7–6.6 (110) (183) (5.2) (6.3) Quercus subg. Cyclobalanopsis Q. acuta Thunb. 24 30.3–35.2 129.2–140.2 6–96 76–112 120–193 2.8–8.1 2.4–9.6 (95) (155) (4.9) (4.5) Q. gilva Blume 13 31.5–35.2 130.5–140.1 8–57 87–156 141–254 2.7–5.5 3.9–12.1 (127) (210) (3.9) (6.5) Q. glauca Thunb. 18 31.7–35.2 130.3–138.2 10–65 53–117 90–188 2.5–5.9 0.8–6 (95) (157) (4.5) (4.1) Q. hondae Makino 3 31.9–31.9 131.1–131.2 9–41 111–145 203–252 2.7–5.6 3.8–6.3 (123) (229) (3.7) (5.4) Q. miyagii Koidz. 4 24.4–27.8 123.8–128.9 12.5–30 112–149 179–233 3.4–6.8 4.9–8.7 (128) (197) (4.7) (7.3) Q. myrsinifolia Blume 7 31.9–35.2 130.5–140.1 8–75 91–113 147–196 2.3–8.5 2.3–7.4 (101) (171) (5.1) (5.3) Q. salicina Blume 24 28.3–35.3 129.2–136.4 8–76 61–121 114–200 2.2–9.6 2.1–6.9 (92) (150) (5.1) (4.2) Q. sessilifolia Blume 14 31.8–35.0 130.5–136.2 10–39 70–140 119–223 3.7–7.6 2.8–9.7 (99) (167) (5.2) (5.1) Quercus subg. Sclerophyllodrys Q. phillyraeoides A.Gray 7 30.3–34.4 130.4–136.9 5.5–33 46–82 76–163 6.4–22.1 2.6–4.3 (60) (121) (10.8) (3.5)

Downloaded from Brill.com09/29/2021 03:03:38PM via free access Noshiro & Sasaki — Japanese Quercus and Lithocarpus 385

Figure 1. Distribution of studied specimens and the distribution range of Quercus gilva. The northern limit of Q. acuta shows the northern limit of evergreen Quercus and Lithocarpus species in Japan (distribution ranges modified from Kurata 1964). presented a key to identify these species from wood structure. In the key for the species with radial-porous wood, the colour of wood was used as the first feature to distinguish groups, and the arrangement of vessels and the occurrence of prismatic crystals in chambered or non-chambered cells were used in the following keys, and the vessel size was used only in a minor key. The colour of wood is not applicable to archaeological woods, and preliminary observation showed that some of the features used in the key, such as the number of vessel rows and the chambering of crystalliferous cells, are too variable to be used as good criteria. Thus, we studied the wood specimens deposited at TWTw, Tsukuba, Japan to clarify the species-level variation in wood structure among Japanese species of Fagaceae with radial-porous wood. In this paper, to conform to the terms used by Shimaji (1962) to describe the onto genetic trends in ray development of Fagaceae where 2–3-seriate narrow rays aggregate and become fused to form broad rays, we refer to broad rays as compound rays and to rays that are partly aggregate and partly compound as semi-compound rays.

Downloaded from Brill.com09/29/2021 03:03:38PM via free access 386 IAWA Journal, Vol. 32 (3), 2011

Materials and methods

Among specimens of Lithocarpus and evergreen Quercus deposited at TWTw, 124 specimens with records of the stem diameter and localities were used (Table 1). All the specimens were obtained from the trunk at around breast height avoiding junctions of branches. The diameter was measured at the height of collection with bark from two directions and was averaged. For four specimens without records of the stem diameter, the diameters were estimated from the radius of the wood specimens (indicated with asteriks). To supplement the lack of available specimens, a branch wood specimen was also studied for Q. hondae. The specimen numbers for each species are as follows:

Lithocarpus edulis: 16054, 18896, 19332, 19367, 20154, 20973, 21240. L. glaber: 18908, 21255, 23494. Quercus acuta: 9315*, 15659, 15852, 16025, 16951, 18813, 19013, 19058, 19548, 19600, 20156, 20264, 21011, 21121, 21210, 22611, 22654, 23471, 23641, 23719, 24010, 24045, 25318, 25412. Q. gilva: 414, 811*, 9317*, 13280, 17750*, 18794, 18899, 21248, 22621, 22624, 25138, 25233, TI-4813 (a microscopic slide with a collection record). Q. glauca: 17570, 18395, 18790, 18863, 18883, 19046, 20193, 21254, 21693, 22546, 23415, 23686, 24035, 25151, 25260, 25261, 25330, 25478. Q. hondae: 17749 (both stem and brach wood specimens), 25272. Q. miyagii: 12867, 17391, 23286, 23351. Q. myrsinifolia: 3344*, 18857, 18906, 21645, 22625, 23416, 25293 Q. salicina: 16089, 16143, 16977, 18475, 18898, 18907, 19089, 19449, 19543, 19634, 20194, 20250, 21203, 21270, 21559, 21633, 22541, 23515, 23597, 23752, 24013, 25316, 25366, 25415. Q. sessilifolia: 18840, 18895, 18904, 19095, 19097, 20204, 21238, 21600, 21709, 22632, 23487, 25156, 25282, 25354. Q. phillyraeoides: 15495, 15516, 16125, 19131, 23697, 23750, 25432.

Quantitative features of vessels were obtained by image analysis of transverse sections. All the vessels in transectional areas of 2.7–15.4 mm in radial width by 2.6–3.4 mm in tangential width were analyzed using ImageJ 1.43o (W.S. Rasband, U.S. National Institutes of Health, Maryland, USA). Vessel frequency is the number of vessels per square millimetre, and vessel area ratio is the proportion of the total transectional area of vessels to the measured area. Curve fitting was done using the power curve option of DeltaGraph v. 6.0 (Red Rock Software, Salt Lake City, UT, USA).

Results

All the species had vessels arranged radially, usually in 1–3 rows (Fig. 2–4). The diameter of vessels usually decreased gradually toward growth ring boundaries, but was occasionally largest in the middle of growth rings. Quercus phillyraeoides tended to have slightly larger vessels that occurred disjunctively at the beginning of growth rings and were followed by a radial population of smaller vessels (Fig. 4). Mean vessel diameter ranged from 92 to 128 µm among the studied species, but was significantly smaller, 60 µm, in Q. phillyraeoides (Table 1). Quercus gilva, Q. hondae, and Q. miyagii had

Downloaded from Brill.com09/29/2021 03:03:38PM via free access Noshiro & Sasaki — Japanese Quercus and Lithocarpus 387

Figure 2. Wood structure of Lithocarpus edulis (TWTw-21240, DBH = 16 cm, H = 9 m). – a: TS, vessels in radial files and indented growth ring boundaries. – b: TLS, a semi-compound ray with a vertical wedge of fibres and uniseriate rays. — Scale bars = 200 µm. significantly larger vessels averaging over 130 µm. Trends were similar for maximum tangential diameter. Quercus phillyraeoides had significantly smaller values, and Q. gilva, Q. hondae, and Q. miyagii had significantly larger values. OnlyQ. gilva, Q. hondae, Q. miyagii, and Q. sessilifolia had specimens with a maximum tangential diameter over 200 µm. Against stem diameter, maximum tangential vessel diameter continued to increase without any plateaus even in trunk diameter ranges from 20 to 60 cm, but in Q. acuta and Q. salicina, the increase became gradual between 20 and 40 cm (Fig. 5). Vessel frequency ranged from 3.7 to 5.2 vessels/mm2 among most species, but was significantly larger, 10.8 vessels/mm2, in Q. phillyraeoides (Fig. 6;

Downloaded from Brill.com09/29/2021 03:03:38PM via free access 388 IAWA Journal, Vol. 32 (3), 2011

Figure 3. Wood structure of Quercus gilva (TWTw-18899, DBH = 43 cm, H = 15 m). – a: TS, semi-ring-porous wood with large earlywood vessels. – b: TLS, a semi-compound ray divided by oblique files of fibres. — Scale bar = 200 µm.

Table 1). Vessel area ratio was significantly high in Q. gilva, Q. hondae, Q. miyagii, and L. glaber, but did not differ among the rest of the species (Table 1). In all the species, axial parenchyma was arranged in irregular narrow bands and composed of cells of similar diameter (Fig. 2–4). Axial parenchyma strands often included prismatic crystals in chambered cells or non-chambered enlarged cells, but this varied within species. Rays consist of uniseriate ones and large aggregate to compound ones, and composi- tion of large rays differed between Lithocarpus, Quercus subg. Cyclobalanopsis, and Q. subg. Sclerophyllodrys (Fig. 2–4). In Lithocarpus, semi-compound rays formed in

Downloaded from Brill.com09/29/2021 03:03:38PM via free access Noshiro & Sasaki — Japanese Quercus and Lithocarpus 389

Figure 4. Wood structure of Quercus phillyraeoides (TWTw-25432, DBH = 19 cm, H = 9 m). – a: TS, one to three slightly larger vessels occurring disjunctively at the beginning of growth rings, followed by a radial population of smaller vessels. – b: TLS, an aggregate ray with vertical to oblique files of fibres. — Scale bar = 200 µm.

inner growth rings developed a vertically elongated mass of fusiform cells inside the rays, outside the c. 5th growth rings, which later enlarged and divided the semi- compound rays vertically (Fig. 2). This division formed indented growth ring boundaries in transverse sections. In Q. subg. Cyclobalanopsis, large aggregate rays mostly became compound as the wood matured and split into smaller segments with intrusion of oblique files of fusiform cells (Fig. 3). In Q. subg. Sclerophyllodrys, aggregate to semi-compound rays continued to be formed into mature wood (Fig. 4). Ray cells often included prismatic crystals in chambered cells or non-chambered enlarged cells; crystals in chambered and non-chambered cells occasionally occur in single specimens.

Downloaded from Brill.com09/29/2021 03:03:38PM via free access 390 IAWA Journal, Vol. 32 (3), 2011

Discussion Identification of Japanese species of evergreen Quercus and Lithocarpus with features of vessels and broad rays Among Japanese species of evergreen Quercus and Lithocarpus, vessel size and frequency and ray structure had enough variation to distinguish species groups. Quercus phillyraeoides of subg. Sclerophyllodrys was most distinct, having semi-ring-porous wood with small (< 100 µm on average), numerous vessels, and aggregate rays not get- ting fused. Two species of Lithocarpus had similar vessel features to species of Quer- cus subg. Cyclobalanopsis, but had aggregate to semi-compound rays that came to be divided by the development of a vertical mass of fusiform cells outside the c. 5th growth rings from the pith. Among species of Quercus subg. Cyclobalanopsis, two groups were recognized, one consisting of Q. gilva, Q. hondae, and Q. miyagii and the other consisting of all others. The former group had very large vessels with a maximum vessel diameter over 200 µm, but the latter had smaller vessels with a maximum diameter mostly less than 200 µm. Both groups similarly had aggregate to semi-compound rays with occa- sional prismatic crystals. However, two species of Lithocarpus and species in the two groups of Quercus subg. Cyclobalanopsis were indistinguishable from wood structure. Trends in the maximum tangential diameter against the stem diameter implied that the maximum diameter of mature wood has not been attained in the studied specimens of Q. gilva and Q. sessilifolia (Fig. 5). The stem diameter of mature trees attains 150–200 cm and 60 cm in Q. gilva and Q. sessilifolia, respectively (Kurata 1964; Ohba 2006).

300

200

L. edulis L. glaber Q. acuta 100 Q. gilva Q. glauca Q. hondae Q. miyagii Maximum tangential diameter (µm) Q. myrsinifolia Q. salicina Q. sessilifolia Q. phillyraeoides 0 0 20 40 60 80 100 Stem diameter (cm) Figure 5. Stem diameter and maximum tangential diameter of vessels among Japanese species of evergreen Quercus and Lithocarpus.

Downloaded from Brill.com09/29/2021 03:03:38PM via free access Noshiro & Sasaki — Japanese Quercus and Lithocarpus 391

If the obtained curve fitting is extrapolated to 60 cm in stem diameter, the maximum tangential diameter of vessels becomes 240 µm and 215 µm in Q. gilva and Q. sessilifolia, respectively. Thus, at this stem diameter, Q. sessilifolia should be included in the species group with the maximum vessel diameter over 200 µm. Because this stem diameter is usually the maximum for Q. sessilifolia, wood with the maximum vessel diameter over 220–230 µm derives from Q. gilva, Q. hondae, or Q. miyagii. Among these three species, Q. gilva can be distinguished when the distribution ranges of species are considered, because Q. miyagii grows only in Nansei islands and Q. hondae in southwestern Shikoku and Kyushu. When the threshold of the maximum vessel diameter is increased to 220–230 µm, very mature wood of Q. gilva and Q. sessilifolia growing in Honshu can be distinguished with a high probability. Studied specimens of Q. gilva collected in Honshu had a maximum vessel diameter of less than 200 µm, but this is probably because these specimens were obtained from trees with stem diameters less than 30 cm that are too young to attain the vessel size of mature wood (Fig. 5). Although there are some overlaps with other species, Q. gilva tended to have fewer, but larger vessels and have larger vessel area ratio compared with other species (Fig. 6). The archaeological woods that initiated this study were obtained from the Sorimachi and adjacent sites, Saitama Prefecture in middle Honshu and were made of trees less than 40 cm in diameter. They had a maximum vessel diameter over 200 µm and thus could be identified asQ. gilva (Noshiro et al. 2009).

300

Q. gilva

Q. hondae, Q. miyagii 200 Q. sessilifolia L. edulis, L. glaber

Q. phillyraeoides

Maximum tangential diameter (µm) Q. acuta, Q. glauca, Q. myrsinifolia, Q. salicina

100

0 0 2 4 6 8 10 Vessel frequency (no./mm2) Figure 6. Ranges of maximum tangential diameter of vessels and vessel frequency among the studied specimens with stem diameters of over 20 cm.

Downloaded from Brill.com09/29/2021 03:03:38PM via free access 392 IAWA Journal, Vol. 32 (3), 2011

Vertical splitting of semi-compound rays of Lithocarpus and its phylogenetic im- plication Broad semi-aggregate rays of Lithocarpus formed in inner growth rings have a vertical wedge of fusiform cells with vessels, banded axial parenchyma, and uniseriate rays in outer growth rings (Fig. 2). In transverse sections, this wedge of fusiform cells usually makes indented growth ring boundaries not found in the studied species of evergreen Quercus. Although development of single broad rays was not followed through serial sections, our observation of preparations and wood specimens shows that, once nearly compound rays are formed, a vertical file of fusiform cells soon forms in their central part. As this file widens and lengthens into a vertically elongated wedge of fusiform cells and uniseriate rays, the nearly compound rays split into a pair of vertically tall semi-compound rays separated by this wedge (Fig. 2b). The wedge first has fibres, axial parenchyma, and uniseriate rays, and then has vessels as this wedge widens tangentially. Thus, this splitting of semi-compound rays differs from the splitting of broad rays reported by Barghoorn (1940) and Moseley (1948) where broad rays are split into shorter, narrower segments with the insertion of usually obliquely elongated files of fusiform cells. This vertical splitting of semi-compound rays differs also from the scheme presented by Shimaji (1962) for the ontogenetic development of aggregate to compound rays in Fagaceae. In his scheme, species of evergreen Quercus and Lithocarpus show ag- gregation of 1–3-seriate rays in the first growth ring and tall semi-compound rays in the tenth growth ring. These semi-compound rays then split into smaller compound rays of the adult type with the formation of obliquely elongated files of fusiform cells in outer growth rings. According to our observation, this type of ray development oc- curs in the species of evergreen Quercus subg. Cyclobanalopsis (Fig. 3b). In the two species of Lithocarpus, on the contrary, tall semi-compound rays split vertically by the formation of a wedge of fusiform cells into nearly equal halves (Fig. 2b). Besides, the formation of compound rays and their splitting occur repeatedly, not only in inner growth rings, but also in outer growth rings. In Lithocarpus, this type of splitting of compound rays is observed in several species in temperate to tropical Asia. Judging from photos in wood atlases and the wood collection at TWTw, the splitting of compound rays seems to occur in Pasania terna- ticupula (Hayata) Schot. (= L. hancei (Benth.) Rehd.) (Yang & Huang Yang 1987) and L. brevicaudatus (Skan) Hayata (TWTw-1026) of Taiwan, L. fordianus (Hemsl.) Chun (Cheng et al. 1992) or L. glaber (Thunb.) Nakai (Cheng 1980) of China (for Chinese species, different species names are applied to the same photos), and L. pachyphyllus (Kurz) Rehder of India (Rao et al. 1991), and L. grandifolius (DC.) S.N. Biswas of Nepal (Joshi 2004). For species of Lithocarpus from Southeast Asia only truly compound rays are reported (Lemmens et al. 1995; Ogata et al. 2008), but similar splitting seems to occur also in L. leptogyne (Korth.) Soepadmo from Sabah (TWTw-9767). Although more species should be studied, so far the vertical splitting of compound rays is found in species belonging to the major clade of subgenus Pasania consisting of Asian species (Manos et al. 2000) and probably has some phylogenetic background.

Downloaded from Brill.com09/29/2021 03:03:38PM via free access Noshiro & Sasaki — Japanese Quercus and Lithocarpus 393

Acknowledgements

This study was partly supported by the Grant-in-Aid from the Ministry of Education, Culture, Sports, Science and Technology, Government of Japan (No. 21300332).

References

Barghoorn, E.S. 1940. The ontogenetic development and phylogenetic specializaion of rays in the xylem of dicotyledons. I. The primitive ray structure. Amer. J. Bot. 27: 918–928. Cheng, J.-Q. 1980. Chinese tropical and subtropical timbers, their distinction, properties and application. Science Press, Beijing (in Chinese). Cheng, J.-Q., J.-J. Yang & P. Liu. 1992. Anatomy and properties of Chinese woods. China For- estry Publ. House, Beijing (in Chinese). Itoh, T. & M. Yamada (eds.). 2011. Archaeology of wooden artefacts from Japan. Kaiseisha Co., Ohtsu (in Japanese). Joshi, L. 2004. Wood anatomy of the family Fagaceae. Bull. Dept. of Plant Resources No. 23. 32 pp. National Herbarium & Plant Laboratories, Godawari. Kurata, S. 1964. Illustrated important forest trees of Japan. Chikyu Shuppan, Tokyo. Lemmens, R.H.M.J., I. Soerianegara & W.C. Wong (eds.). 1995. Plant Resources of South-east Asia, No. 5(2). Timber trees: Minor commercial timbers. Backhuys, Leiden. Manos, P.S., Z.-K. Zhou & C.H. Cannon. 2000. Systematics of Fagaceae: phylogenetic tests of reproductive trait evolution. Int. J. Plant Sci. 162: 1361–1379. Moseley, F.M. 1948. Comparative anatomy and phylogeny of the Casuarinaceae. Bot. Gaz. 110: 231–280. Noshiro, S., Y. Sasaki & Y. Murakami. 2009. Wood selection at the Sorimachi site. In: Ar- chaeological Report of the Sorimachi Site, I: 315–345. Saitama Cultural Deposits Research Corporation, Kumagaya (in Japanese). Ogata, K., T. Fujii, H. Abe & P. Baas. 2008. Identification of the timbers of Southeast Asia and the Western Pacific. Kaiseisha, Ohtsu. Ohba, H. 2006. Fagaceae. In: K. Iwatsuki, D.E. Boufford & H. Ohba (eds.), Flora of Japan, vol. IIa: 42–60. Kodansha Co., Tokyo. Onaka, F. 1939. On the ancient wooden implements of “Yayoi” type (over 2000 years ago) dis- covered at Karako, Yamato Province, Japan. Jap. J. Forestry 439–545 [in Japanese]. Rao, V.R., B. Sharma, R. Dayal & R.D. Raturi. 1991. Occurrence of broad rays in Indian Cas- tanopsis (D.Don) Spach; Lithocarpus Blume and Quercus semiserrata Roxb. (Fagaceae). J. Ind. Acad. Wood Sci. 22: 25–40. Shimaji, K. 1954. Anatomical studies on the wood of Japanese Quercus. I. On subgenus Lepi- dobalanus (Nara group). Bull. Tokyo Univ. For. No. 46: 193–210. Shimaji, K. 1956. Anatomical studies on the wood of Japanese Quercus. II. On Subgenus Cy- clobalanopsis (Kashi group). Bull. Tokyo Univ. For. No. 47: 125–143. Shimaji, K. 1959. Anatomical studies on the wood of the Japanese Pasania, Castanea and Castanopsis: With a key to the 22 Japanese representative species of the Fagaceae. Bull. Tokyo Univ. For. No. 55: 81–99. Shimaji, K. 1962. Anatomical studies on the phylogenetic interrelationship of the genera in the Fagaceae. Bull. Tokyo Univ. For. No. 57: 1–64. Yang, K.-C. & Y.S. Huang Yang. 1987. Minute structure of Taiwanese woods. Hua Shiang Yuan Publishing, Taipei.

Downloaded from Brill.com09/29/2021 03:03:38PM via free access