Betulaceae s.l.

The Betuloideae (fig. 2) and Coryloideae (fig. 3) have been treated variously as subfamilies within the Betulaceae or as separate families, Betulaceae and Corylaceae. The Li et al. (2004) analysis did not recover a clear grouping of Alnus Miller and Betula versus the Coryloideae (Carpinus L., Corylus L., Ostrya Scop., Ostryopsis Decne), but Forest et al.’s (2005), which used a different set of genes and examined a different number of species, did. Vessels in the Coryloideae typically have radial/diagonal/dendritic arrangements, while those in Alnus and Betula are randomly arranged (table 1).

Isolated pieces of Betula wood are easy to identify because of their unique combination of vessels that are solitary and in short radial multiples that are randomly arranged (fig. 2A), minute alternate intervessel pitting (fig. 2B), vessel-ray parenchyma pits similar to intervessel pits (fig. 2C), exclusively scalariform perforation plates (fig. 2C), and homocellular rays (fig. 2B, 2C). Alnus woods are also distinctive, with vessels that are solitary and in radial multiples that are randomly arranged (fig. 2E), scalariform perforation plates (fig. 2F, 2G), opposite to alternate intervessel pits (fig. 2G), and, in many species, aggregate homocellular rays (fig. 2E, 2F; e.g., Tippo 1938; Hall 1952; Brazier and Franklin 1961; Panshin and deZeeuw 1980; Schweingruber 1990). Coryloideae woods often have a radial/diagonal vessel arrangement (fig. 3A–3C); aggregate rays occur in Carpinus and Corylus (fig. 3A), and simple perforation plates occur in Ostrya and some Carpinus and Corylus. Crystals are unknown in the Betuloideae, but crystalliferous ray cells occur in some Coryloideae (table 2; fig. 3G).

Müller-Stoll and Mädel (1959) reviewed occurrences of fossil betulaceous woods. The earliest occurrence of wood with features of Betula is from the middle Eocene Clarno Nut Beds, Oregon (Scott and Wheeler 1982; Wheeler and Manchester 2001). Alnus-like woods with the abovementioned combination of features are more common and are reported from the Paleocene of England (Crawley 2001) and the Denver Basin of the United States (Wheeler and Michalski 2003), the Paleocene-Eocene of the Russian Far East (Blokhina and Snezhkova 1999), and the early Eocene of Yellowstone National Park (Wheeler et al. 1977). This is congruent with the fruit and seed record (Forest et al. 2005; Liu et al. 2014). Unequivocal Coryloideae woods have been recovered from the Upper Cretaceous (Campanian) of California (Carpinoxylon ostryopsoides Page 1970), the early Eocene of Yellowstone National Park (Carpinus absarokensis Wheeler, Scott & Barghoorn and Carpinus saximontana Wheeler, Scott & Barghoorn; Wheeler et al. 1977), and the Oligocene of Europe (Coryloxylon nemejcii Prakash, Březinová & Bůžek and Coryloxylon tertiarum Prakash, Březinová & Bůžek; Prakash et al. 1971).

Casuarinaceae

Moseley (1948) surveyed Casuarinaceae wood anatomy. Since that time, Casuarina has been split into four genera: Allocasuarina L.A.S. Johnson, Casuarina L. s.s., Ceuthostoma L.A.S. Johnson, and Gymnostoma L.A.S. Johnson (Johnson and Wilson 1989; Wilson and Johnson 1989). Steane et al.’s (2003) phylogenetic analysis supports the recognition of these four genera. Some Casuarina species appear similar to some Quercus L. species (the evergreen oaks) because they share solitary vessels (fig. 4A–4D), vasicentric tracheids (fig. 4E), two size classes of rays (fig. 4H–4J), and apotracheal axial parenchyma (fig. 4A–4D). However, vessel-ray pits in Casuarinaceae do not have markedly reduced borders and are not irregularly shaped (fig. 4G), whereas they are in the Castaneoideae (fig. 7A–7D).

In the 1990s, M. Molenaar (Rijksherbarium, Leiden University; M. Molenaar, unpublished data) began a wood anatomical study of the family, sectioning woods of Allocasuarina, Casuarina, and Gymnostoma from a variety of habitats; samples of Ceuthostoma were not available. He observed variation in vessel diameter and density correlating with variation in habit and habitat, with shrubby species that grow in arid regions having narrower, more numerous vessels than tree species that grow in moist habitats. Descriptions based on Molenaar’s work and photographs of his sections are available on the InsideWood website. These data show that there is potential for using wood anatomy to distinguish the genera (tables 1, 2), to investigate how wood anatomical features correlate with xeric conditions, and to use fossil woods to track the history of the family.

However, reliable reports of fossil woods of Casuarinaceae are rare, and the descriptions of putative Cretaceous and Paleogene woods of the family do not include the details necessary to confirm their affinities with the family. In 1937, Shimakura thought that large compound rays as seen in a poorly preserved small axis with pith from the Cretaceous of Japan occurred only in Casuarinaceae, Fagaceae, and Proteaceae. He considered its vessel arrangement most similar to that of Casuarinaceae and so named it Casuaroxylon japonicum Shimakura. However, large rays and predominantly solitary vessels also occur in other families (e.g., Aphloiaceae, Cardiopteridaceae, Cunoniaceae, Dilleniaceae, Ericaceae, Icacinaceae, Menispermaceae). Shimakura’s (1937) description mentions simple perforation plates with a question about the occurrence of scalariform perforation plates; a line drawing shows a scalariform perforation plate, and photographs show only steeply inclined vessel element end walls, so it is not clear what type of perforation plate predominates. Vessel-ray parenchyma pits are not described or illustrated. The pith of this Cretaceous wood sample has sclereids; however, the pith of Allocasuarina stricta (Macklin) L.A.S. Johnson does not have sclereids (Moseley 1948, fig. 27). The wood collection of the Naturalis Biodiversity Center has slides of two other Allocasuarina species and two Casuarina species with pith included in the sections; there were no sclereids.

Hofmann (1944) assigned an Oligocene wood from Europe (Austria) to Casuarinaceae. However, the description of this wood was brief, and some important diagnostic characters (e.g., vessel-ray parenchyma pits) were not mentioned, so the familial assignment is questionable. Vessel-ray parenchyma pit type is particularly useful for distinguishing Casuarinaceae from Fagaceae (table 1).

Moore and Wallace (2000) reported that multiple samples of Casuarinaceae wood occurred in the Miocene of New Zealand. They did not provide anatomical details, but the one cross section figured shows anatomy consistent with Casuarinaceae, particularly with those species from moist habitats. More recently, Vanner (2019) described Casuarinoxylon ildephonsi Vanner from the Miocene of New Zealand, also hypothesized to be from a mesic site.

Fagaceae

The Fagaceae are often treated as having three subfamilies, Fagoideae, Quercoideae, and Castaneoideae (Kubitzki 1993), but phylogenetic analyses reveal that the Castaneoideae are paraphyletic, with Quercus intervening (Oh and Manos 2008; Larson-Johnson 2016), supporting Nixon’s (1989) recognition of just two subfamilies, Fagoideae and Castaneoideae. Consistent with the phylogenetic topology of Li et al. (2004) and as known for decades (Metcalfe and Chalk 1950), the Fagoideae (fig. 5; vessels solitary and in multiples, random vessel arrangement, and heterocellular rays variable in size—ranging from narrow to wide) have anatomy distinct from that of the Castaneoideae (figs. 6, 7; vessels exclusively solitary, usually in a radial/diagonal to dendritic arrangement, rays exclusively uniseriate or of two distinct sizes—uniseriate plus very wide and tall or uniseriate plus aggregate rays—and predominantly homocellular; tables 1, 2). Vasicentric tracheids, the occurrence of which is often correlated with wide and exclusively solitary vessels (Carlquist 1984, 2001), are common in the Castaneoideae and rare in the Fagoideae. Grimmson et al. (2016) suggested that Fagus L. split from other Fagaceae by the Late Cretaceous.

There is considerable overlap in the wood anatomical characteristics of Lithocarpus Blume and Quercus. There is also overlap in the wood anatomy of Castanea Miller, Castanopsis (D. Don) Spach, and Chrysolepis Hjelmq. (tables 1, 2). The molecular analyses of Li et al. (2004) did not find clear distinctions among these genera. The overlap in the appearance of diffuse-porous Quercus and diffuse-porous Lithocarpus is such that, at present, it is not clear whether their woods are distinguishable, supporting the merger of Quercoideae and Castaneoideae. Manos et al. (2008) recognized the western North American endemic Notholithocarpus densiflorus (Hook. & Arn.) Manos, Cannon & S.H. Oh (synonym: Lithocarpus densiflorus (Hook. & Arn.) Rehder) and considered it more closely related to Castanopsis than to Lithocarpus. InsideWood’s images of tangential and transverse sections of these two genera give some support for that proposed relationship: aggregate rays in Lithocarpus are composed of wider rays than those that occur in Notholithocarpus Manos, Cannon & S.H. Oh and some species of Castanopsis.

Trigonobalanus Forman was recovered as sister to the castaneoid genera in the analyses of Li et al. (2004) and Larson-Johnson (2016). Nixon and Crepet (1989) proposed that the three species of Trigonobalanus be split into three monospecific genera. Available wood anatomical information for Trigonobalanus does not provide support for such a split. Trigonobalanus wood is diffuse-porous and has uniseriate and aggregate rays and so is similar to that of Lithocarpus and diffuse-porous Quercus spp. (Forman and Cutler 1967), supporting Forman’s (1964) placement of the genus in the Quercoideae sensu Kubitzki. The three trigonobalanoid taxa form a clade that is sister to a clade of both quercoid and castaneoid genera in the analysis of Larson-Johnson (2016).

Being able to recognize different species of Quercus has been a preoccupation of North American and European wood scientists since the 1800s (Abromeit 1884, cited in Tillson and Muller 1942). It is also an issue in protecting the endangered species Quercus mongolica (CITES app. III; https://trade.cites.org), a white oak from the Russian Far East that has been mixed with American and European oaks in flooring and furniture (https://eia-global.org/subinitiatives/russia). The accepted wisdom is that all ring-porous red oaks (subgenus Quercus, section Lobatae) have latewood pores that are thick walled and rounded in outline, while ring-porous white oaks (subgenus Quercus, section Quercus) have latewood pores that are thin walled and angular in outline (e.g., Panshin and DeZeeuw 1980); evergreen oaks often have semi-ring-porous to diffuse-porous wood with vessels in a diagonal to radial arrangement (table 1). Identification of individual species is considered impractical, however. Although it would be of value to museum conservators and antique dealers who are interested in the provenance of materials made of oak wood, there does not appear to be a reliable way to use wood anatomy to distinguish European white oaks from American white oaks (e.g., Brazier and Franklin 1961).

Fossil Fagoideae wood. In his monograph on Fagus-like fossil woods, Süss (1986) accepted Fagoxylon hokkaidense Stopes and Fujii (1910) from the Late Cretaceous of Japan as Fagus-like. However, Takahashi and Suzuki (2003) were not able to locate and reexamine material of F. hokkaidense and consider that the original diagnosis lacks details necessary for showing that the wood can be only Fagus. We concur with Takahashi and Suzuki.

Paleogene fossil woods that were thought to have features of Fagus include Fagus grandiporosum Beyer (1954) from the early middle Eocene of Yellowstone National Park, Wyoming, and Fagoxylon kraeuselii Hofmann (1952) from the Oligocene of Austria. The description of F. grandiporosum by Beyer (1954, p. 563) mentions “intervessel … pits not as abundant as in living species,” and the paper illustrates only widely spaced alternate intervessel pits; vessel-ray parenchyma pits were not described or figured. It is questionable that the affinities of this fossil wood are with Fagus. Unfortunately, the description of the Austrian wood is not detailed, so it is not possible to reevaluate the affinities of this wood. Denk (2003) and Denk and Grimm (2009) suggested that Fagus originated in the northern Pacific Basin during the early Tertiary and spread to Europe and Central Asia during the Oligocene. At least 10 Fagus-like woods are known from the Mio-Pliocene of Europe and the United States (http://insidewood.lib.ncsu.edu; Süss 1986; Gregory et al. 2009). There are late Eocene woods from Oregon with features of Fagus (E. A. Wheeler and S. R. Manchester, unpublished data).

Fossil Castaneoideae wood. Quercinium centenoae Estrada-Ruiz, Martínez-Cabrera & Cevallos-Ferriz (Estrada-Ruiz et al. 2007) from the late Campanian/early Maastrichtian Olmos Formation of Mexico and Paraquercinium cretaceum Wheeler, Lee & Matten from the Maastrichtian McNairy Formation of southern Illinois (Wheeler et al. 1987) are the oldest known occurrences of woods with features of the Castaneoideae. Both have diffuse-porous to slightly semi-ring-porous wood, few and relatively wide exclusively solitary vessels in a somewhat radial/diagonal arrangement, aggregate rays and exclusively uniseriate rays that are predominantly homocellular, and vasicentric tracheids.

Quercus cretaceoxylon Suzuki & Ohba (1991) from Hokkaido, Japan, has characteristics of a ring-porous oak, but the sample was a pebble in a modern river deposit, so its exact stratigraphic placement is not known, and in spite of its name it is likely Cenozoic (Takahashi and Suzuki 2003). A distinctly ring-porous wood would be unusual in the Cretaceous. Ring-porous wood, considered an adaptation to seasonal climates, did not become common until after the Eocene (Wheeler and Baas 1991, 2019). Ring-porous Castaneoideae woods were common in the Miocene-Pliocene of Asia, Europe, and North America (e.g., Watari 1952; Müller-Stoll and Mädel 1957; Prakash and Barghoorn 1961a, 1961b; Selmeier 1971; Privé 1975; Petrescu 1978; Suzuki and Ohba 1991; Gottwald 2004).

By the Paleocene, woods of both the Quercus/Lithocarpus type and Castanea/Castanopsis/Chrysolepis type had occurred. A wood from the Paleocene Williamsburg Formation, South Carolina, was assigned to Fagaceoxylon Scott & Wheeler (Melchior 1998), but it has exclusively solitary vessels, which is not a characteristic of Fagaceoxylon (a fossil combining features of the Betulaceae and Fagaceae; Scott and Wheeler 1982). Exclusively solitary vessels, along with uniseriate and aggregate rays and irregularly shaped vessel-ray parenchyma pits, suggest that Melchior’s Paleocene wood belongs in the semi-ring-porous to diffuse-porous Quercus and Lithocarpus group and would be better placed in Quercinium Unger or Lithocarpoxylon Petrescu. Castanoxylon philipii Crawley (2001) from the Paleocene Thanet Sands, England, is a semi-ring-porous to diffuse-porous Castanopsis-like wood. There are mummified woods from the upper Oligocene of South China assigned to Lithocarpoxylon and Castanopsis (Huang et al. 2018); the Castanopsis woods had a combination of distinctly ring-porous woods, but their growth ring boundaries were not delineated by thick-walled imperforate elements, a combination suggested to be confined to the tropics.

Given that oak woods are relatively easy to recognize, even without prepared thin sections, it would be possible to examine hand specimens from localities with good age control to see whether the timing of the appearance of ring porosity in the Castaneoideae could be better defined. The diffuse to semi-ring-porous condition seen in Late Cretaceous and Paleogene Castaneoideae wood is apparently a precursor to the more strongly ring-porous woods seen in the Neogene (e.g., Selmeier 1971; Suzuki and Terada 1996; Wheeler and Dillhoff 2009; Iamandei et al. 2011). Present-day ring-porous woods are deciduous (Boura and DeFranceschi 2007), so comparing changes in porosity to changes in leaf characters would provide information on the history of the relationship between leaf physiognomy and xylem physiognomy.

Juglandaceae

Among the useful studies of the wood anatomy of the Juglandaceae are those by Heimsch and Wetmore (1939), Müller-Stoll and Mädel (1960), Miller (1976), Manchester (1983), and Dupéron (1988), as well as that by Withner (1941) on Rhoiptelea. Rhoiptelea, native to China and northern Vietnam, was formerly treated in its own family, Rhoipteleaceae, distinguished by its bisexual flowers, but recent phylogenetic analyses treat it as a genus of the Juglandaceae and the sole member of the subfamily Rhoipteleoideae. Juglandaceae are the only extant Fagales with compound leaves. Rhoiptelea wood differs from other Juglandaceae woods because it has exclusively scalariform perforation plates, usually with 3–12 bars (fig. 8B, 8C; Tippo 1938; Withner 1941; table 1). We had access to only a tangential section of juvenile wood, but its rays are markedly taller than those of other Juglandaceae (cf. fig. 8C with figs. 8E, 9E, 9F). One of the four samples that Withner described also had exceptionally tall rays. The pore distribution pattern is consistent with that of other Juglandaceae (table 1). There are no reports of fossil wood of Rhoiptelea.

Manos and Stone (2001) investigated relationships within extant members of Juglandaceae using ITS, chloroplast DNA, morphological, and chemical data. Their results supported the continued recognition of two major clades within the family, Engelhardioideae and Juglandoideae. In the Engelhardioideae (Alfaroa Standl., Engelhardia Lesch. ex Blume, Oreomunnea Oerst.), there are predominantly simple perforation plates accompanied by occasional scalariform perforation plates and vessel-ray parenchyma pits of various shapes (fig. 8G). Recognition of Alfaropsis Iljinsk., with the single species Alfaropsis roxburghiana (Wall.) Ilinsk., has received support for its segregation from Engelhardia (Stone 2010). However, we did not find any anatomical features that distinguish the wood of Engelhardia roxburghiana Lindl. from other Engelhardia species (table 1).

In the Juglandoideae (Carya Nutt., Cyclocarya Iljinsk., Juglans L., Platycarya Siebold and Zucc., Pterocarya Kunth), there are exclusively simple perforation plates and vessel-ray parenchyma pits that are similar in shape to the intervessel pits (although at times with the pit borders reduced; table 1; fig. 9). The Juglandoideae are further divided into two tribes, Juglandeae and Platycaryoideae (Platycarya). Platycarya wood has vascular/vasicentric tracheids, vessels with helical thickenings, and crystals in idioblastic ray parenchyma cells, features not found in Juglandeae wood (table 2; fig. 9H, 9I). Manos and Stone (2001) proposed the recognition of two subtribes within the Juglandeae, Juglandinae (Juglans, Pterocarya, Cyclocarya) and Caryinae (Carya). Septate pith, a feature not found elsewhere in the Juglandaceae or other Fagales, characterizes Juglans, Pterocarya, and Cyclocarya. Miller (1976) found that woods of the different sections of Juglans could be distinguished by aspects of their wood anatomy. The butternuts (sections Cardiocaryon and Trachycaryon) were observed to have procumbent ray cells that were “flattened”/oval when viewed in tangential section, whereas black walnuts (section Rhysocaryon) had ray cells that were circular when viewed in tangential section. Crystals in the axial parenchyma were observed in long chains (tropical black walnuts) or short chains (temperate black walnuts) or were absent (butternuts and English walnuts).

There is overlap in the wood anatomical features of the subtribes Juglandinae and Caryinae, for example, in the anatomy of the semi-ring-porous species of Carya (e.g., Carya illinoinensis (Wangenh.) K. Koch) and some Juglans species (e.g., Stark 1953; Dupéron 1988). Annamocarya A. Chev. is considered a monotypic genus in the Flora of China (http://www.efloras.org/florataxon.aspx?flora_id=2&taxon_id=10460) but is treated as Carya sinensis Dode in Kew’s Plants of the World (http://www.plantsoftheworldonline.org). Its wood anatomy is consistent with placement in the Juglandeae (tables 1, 2), and it has rhomboidal crystals in idioblasts of axial parenchyma—a distinctive feature of Carya—and rarely in ray parenchyma. Its differences from the temperate species of Carya are as would be expected from its geographic origin in tropical and subtropical southwest China and Vietnam (Leroy 1953): growth ring boundaries are not as distinct, and marginal parenchyma is lacking. Within the Juglandeae, Annamocarya also stands out because its multiseriate rays are exclusively heterocellular, a feature shared with the Engelhardioideae.

Fossil woods of Engelhardioideae assigned to Engelhardioxylon Manchester occurred in the middle Eocene of Oregon and Texas (Manchester 1983), the late Eocene of western Europe (Gottwald 1992), and the Oligocene of the Russian Far East (Blokhina et al. 2002). The earliest known Juglandoideae woods occurred in the early middle Eocene (Yellowstone National Park: Wheeler et al. 1977; Denver Basin: Wheeler and Michalski 2003) and are referable to Pterocaryoxylon Müller-Stoll & Mädel, a genus used for woods that have characteristics seen in both Pterocarya and the “soft walnuts” (Juglans, section Trachycaryon and section Cardiocaryon).

Dupéron (1988) reviewed fossil wood of the Juglandaceae, updated the work of Müller-Stoll and Mädel (1960), emended the diagnosis of Eucaryoxylon Müller-Stoll & Mädel, and agreed that it should be used for fossil woods with clear affinities to Carya. He also proposed a new genus, Rhysocaryoxylon Dupéron, for woods resembling the “black walnut” group, Juglans, section Rhysocaryon. Rhysocaryoxylon was subsequently recognized from the early Eocene of Denmark (Sakala and Gryc 2011), which is significant in that extant black walnuts are confined to the Americas.

Strongly ring-porous woods of Juglandaceae, like those seen today in species of Carya and Platycarya, did not appear until after the Eocene. This post-Eocene timing coincides with the appearance of the same condition in Fagaceae.

Myricaceae

The position of the Myricaceae with respect to the other families of the Fagales is considered somewhat uncertain (Stevens 2001–), as is the number of species within this geographically widespread family (Huguet et al. 2005). Li et al.’s (2004) analysis recovered Myricaceae as sister to Juglandaceae (including the Rhoipteleaceae), but there was not strong support for this relationship. Myricaceae (fig. 10) and Juglandaceae (figs. 8, 9) share few wood anatomical features (tables 1, 2). Carlquist (2001) studied the wood anatomy of the Myricaceae’s four genera (Canocomyrica Guillaumin, Comptonia Banks ex Gaertn., Myrica L., and Morella Lour.) and found them to be distinct from one another.

Huguet et al. (2005) used ITS and rbcL sequence data to analyze Myrica species. They found that Myrica gale L. (the lectotype of Myrica) and Myrica hartwegii S. Watson formed a clade distinct from other Myrica. The “other” Myrica species are now treated as the separate genus Morella, leaving Myrica sensu stricto as a relatively depauperate genus. Wood anatomy supports this distinction: both M. gale and M. hartwegii are semi-ring-porous, with an earlywood that includes vessel multiples and a latewood with very few vessels (fig. 8C), whereas Morella species are diffuse-porous and have predominantly solitary vessels (table 1; fig. 10A, 10B).

Fossil woods assigned to Myrica are reported from the early Eocene of Eden Valley (Myrica scalariforme Kruse (1954)) and the early middle Eocene of Yellowstone National Park, Wyoming (Myrica absarokensis Wheeler, Scott & Barghoorn; Wheeler et al. 1978). However, these Eocene fossil woods have features of Morella rather than Myrica, as they have exclusively solitary vessels, multiseriate rays are as common as uniseriate rays, and crystals occur in chambered axial parenchyma. Myrica sensu stricto, as exemplified by M. gale and M. hartwegii, is distinguished by radial multiples, mostly uniseriate rays with rare partially multiseriate rays, and a lack of crystals. Consequently, we propose the following new combinations.

Morella Lour.

Morella scalariforme (Kruse) Wheeler & Manchester comb. nov.

Basionym. Myrica scalariformis Kruse 1954, Ohio Journal of Science 54(4):243–244.

Morella absarokensis (Wheeler, Scott, Barghoorn) Wheeler & Manchester comb. nov.

Basionym. Myrica absarokensis Wheeler, Scott, & Barghoorn 1978, Journal of the Arnold Arboretum 59(1):10–11.

There is one putative Myrica wood from the Paleocene of South Carolina (Melchior 1998) based on a small axis with a stellate pith and fewer than three growth rings. The wood shares the following features with Myricaceae: scalariform perforation plates, alternate to opposite intervessel pitting, vessel-ray parenchyma pits similar to intervessel pits, diffuse to diffuse-in-aggregates axial parenchyma, and narrow heterocellular rays. However, it is not a “match” for any of the modern genera; Canocomyrica, Comptonia, and Morella have exclusively solitary vessels, and this wood has some radial multiples. Myrica has some radial multiples, but it has exclusively uniseriate rays, semi–ring porosity, and very thin-walled fibers (table 1). It is possible that there is some relationship to Myricaceae, but the same combination of features occurs in other families, including the Cornaceae, Cunoniaceae, Curtisiaceae, Pentaphylacaceae, Styracaceae, Symplocaceae, and Viburnaceae, although these families usually have more bars per perforation plate than does Melchior’s fossil, which has fewer than 20 bars. Wood of Comptonia has not been identified in the fossil record despite the widespread occurrence of this genus from the Eocene onward in North America, Europe, and Asia on the basis of distinctive leaves (Liang et al. 2010).

Nothofagaceae

Nothofagus Blume was formerly placed in the Fagaceae, but it has long been recognized that it has distinct wood anatomy (Metcalfe and Chalk 1950), and it is positioned as sister to the other Fagales in phylogenetic analyses (e.g., Li et al. 2004; Larson-Johnson 2016). This genus is characterized by radial multiples of vessels (fig. 9A–9C) and narrow heterocellular rays (fig. 11D–11F, 11H, 11I). These features do not characterize the Castaneoideae or Fagoideae, although there are rare reports of narrow heterocellular rays in Lithocarpus. Within Nothofagus there are differences between the tropical subgenus Brassospora Philipson & M.N. Philipson from New Guinea and New Caledonia and the temperate clades from South America, New Zealand, and Australia. Brassospora has irregularly spaced conspicuous parenchyma bands and relatively wide vessels in low densities (table 1; fig. 9B). Temperate species have narrow and numerous vessels, rare axial parenchyma that is largely restricted to growth ring boundaries, and, in some species, very narrow latewood vessels intergrading with vascular tracheids (Dadswell and Ingle 1954). Septate fibers occur in varying frequencies together with nonseptate fibers in the temperate group, but they are virtually absent from tropical Brassospora. Nothofagus wood differs from that of other Fagales, and it might be difficult to recognize that it belongs in the same clade only on the basis of wood.

Fossil woods similar to Nothofagus are usually assigned to Nothofagoxylon Gothan. Poole (2002) reviewed the fossil record of woods of the Nothofagaceae as well as the characteristics of the modern species. In extant species of this family, there is variation in intervessel pitting (scalariform to opposite to alternate; fig. 11D–11F), ray width (exclusively uniseriate or one- to three-seriate; fig. 11E, 11F, 11I), and fiber types (septate [fig. 11I] and/or nonseptate [fig. 11E, 11F]; tables 1, 2). Nothofagaceae is the only fagalean family in which some septate fibers occur (13 of the 20 InsideWood Nothofagus descriptions). It is uncertain whether this is a retained primitive feature or a novelty (adaptation).

According to Poole (2002), anatomically distinct groups, that is, groups that resemble the extant subgenera Lophozonia and Nothofagus, occurred by the Late Cretaceous (Campanian to Maastrichtian). No Cretaceous or early Tertiary wood with characteristics of the tropical subgenus Brassospora is known.

Ticodendraceae

In the same volume in which Ticodendraceae was first described (Gómez-Laurito and Gómez 1991), Carlquist (1991) provided a detailed description of its wood. He considered it “exceptionally rich in primitive features” (Carlquist 1991, p. 101), that is, primitive in the Baileyan sense (e.g., Wheeler and Baas 2019). The monotypic genus Ticodendron Gómez-Laur. & L.D. Gómez, distributed in montane cloud forests from southern Mexico to Costa Rica and Panama, has numerous narrow vessels that are predominantly solitary and tend to be angular in outline (fig. 12A–12D), vessel element lengths of >1 mm, scalariform perforation plates with numerous bars (fig. 12F, 12H), imperforate elements with distinctly bordered pits (fig. 12M), diffuse axial parenchyma and no obvious paratracheal parenchyma (fig. 12A–12D), and heterocellular rays (tables 1, 2). No fossil wood of Ticodendron has been reported. However, additional comparative work is needed to establish whether Ticodendron wood can be reliably distinguished from that of genera in families outside the Fagales that have these primitive characters. A search of InsideWood’s database indicates that Ticodendron’s suite of characters also occurs in these families (order): Clethraceae (Ericales), Cornaceae (Cornales), Cunoniaceae (Oxalidales), Daphniphyllaceae (Saxifragales), Ericaceae (Ericales), Hamamelidaceae (Saxifragales), Pentaphylacaceae (Ericales), Stachyuraceae (Crossosomatales), Symplocaceae (Ericales), Theaceae (Ericales), and Viburnaceae (Dipsacales). These families are spread throughout the eudicot phylogeny in the rosids and asterids. It is somewhat surprising that its wood does not more closely resemble that of Betulaceae or Casuarinaceae, which are similar in pollen morphology and phylogenetically adjacent (fig. 1). The wood anatomical similarities to Myricaceae may indicate retention of plesiomorphic characters in both clades 2 and 3 of figure 1.

Woods with these general characteristics but not assigned to a family are known from the Late Cretaceous onward in both the Northern and Southern Hemispheres (e.g., Page 1981; Gottwald 2000; Meijer 2000; Terada et al. 2011; Pujana et al. 2018). Leaves from the middle and late Eocene of Oregon and anatomically preserved fruits from the Eocene of Oregon and southern England resemble those of Ticodendron (Ferrignocarpus Manchester; Manchester 2011).

Comments on the Fossil Record

Although more than eight genera of small fagalean flowers and fruits have been recognized on the basis of superbly preserved charcoalified remains from the Late Cretaceous of Europe and North America (e.g., Herendeen et al. 1995; Schönenberger et al. 2001; Friis et al. 2003), we do not have good records for the co-occurring vegetative organs. We do not know how the corresponding woods might have compared to extant woods of the families thought to be related to these remains, that is, Juglandaceae, Myricaceae, and Betulaceae. Fruits and flowers of a presumed fagaceous affinity are also known from the Late Cretaceous of the US state of Georgia (Herendeen et al. 1995), but, again, we lack information on any co-occurring fossil woods.

Nevertheless, there are a few indications of fagalean diversity during the Cretaceous based on wood. These include Carpinoxylon ostryopsoides from the Campanian Panoche Formation of California (Page 1970), Quercinium centenoae (Estrada-Ruiz et al. 2007) from the late Campanian/early Maastrichtian Olmos Formation of Mexico, and Paraquercinium cretaceum from the Maastrichtian McNairy Formation of southern Illinois (Wheeler et al. 1987). Moreover, Page (1970) described a wood that had fagaceous features (sample CASG 60426 of her group VIB). Nothofagaceae woods occurred in the Late Cretaceous of Patagonia and Antarctica, with subgenera beginning to diverge by the Campanian (Poole 2002).

Despite Paleocene occurrences of foliage and fruits of Juglandaceae, the woods diagnostic of this family are first known from the Eocene. These include species of Pterocaryoxylon with features of present-day Pterocarya and Juglans species that have similar wood anatomy; Engelhardioxylon, accommodating woods similar to extant Alfaroa, Engelhardia, and Oreomunnea; and the extinct Clarnoxylon and Manchesteroxylon, with combinations of features that indicate juglandaceous affinities but that do not occur in a single present-day genus.

Myricaceae wood is reported from the Eocene of Wyoming; reliable reports of Casuarinaceae wood are Miocene. There are no known fossil Ticodendron woods.

Wood Anatomy and Segregate Genera

The recognition of segregate genera during the past few decades is supported in some cases, but not in others, by wood anatomy. The distinction between Morella and Myrica is well supported by wood anatomy. There is some support for the recognition of Notholithocarpus as separate from Lithocarpus. On the other hand, wood anatomy does not support the segregation of the genera Colombobalanus and Formanodendron from Trigonobalanus or the recognition of Engelhardia roxburghiana as the monotypic Alfaropis. Annamocarya wood is very similar to that of Carya, and the differences from other species of Carya, such as diffuse versus ring porosity, are consistent with it being a tropical outlier of the genus. Its narrow, wavy tangential bands of axial parenchyma occur in several genera of Juglandaceae (table 1).

Wood Anatomical Character Evolution: Hypotheses of Ancestral States

To underpin hypotheses on character evolution and as a basis for discussing repeating patterns of wood anatomical syndromes, we plotted some characters from tables 1 and 2 on the Li et al. (2004) phylogeny (fig. 13).

Perforation plate type (fig. 13A). Most Fagales have simple perforations with a low percentage of scalariform perforations. Exclusively simple perforations characterize the Castaneoideae and Juglandoideae, whereas exclusively scalariform perforation plates characterize the Betuloideae, the Rhoipteleoideae, and the genera Canacomyrica, Morella, and Myrica (Myricaceae), as well as the Ticodendraceae (many bars). Plotting vessel perforation type on the molecular phylogeny would reconstruct mixed simple and scalariform perforations as the ancestral condition, with multiple derivations of exclusively scalariform or exclusively simple perforations and some reversals. An alternative interpretation would posit exclusively scalariform perforations as the ancestral condition. Lens et al. (2016), Carlquist (2018), Wheeler and Baas (2019), and Olson (2020) have presented arguments and evidence to support the hypothesis that scalariform to simple perforation plate transitions are largely irreversible (“Carlquist’s ratchet”) but that parallel transitions from scalariform to simple perforations were rampant. In this alternative interpretation, ancestral Fagales had exclusively scalariform perforation plates with multiple (approximately five) independent origins of simple perforations. Lens et al. (2016) reconstructed the vessel element evolution of the asterids, with almost all basal nodes along the spine of the molecular phylogeny of this superorder characterized by exclusively scalariform plates and multiple convergent origins of simple perforation plates.

However, in Fagales, the situation seems more complicated because Fagales are nested among clades with exclusively simple perforation plates (Cucurbitales, Rosales, Fabales) in the APG phylogeny. The few known Late Cretaceous fagalean woods have exclusively simple perforations (P. cretaceum, Q. centenoae) or both simple and scalariform perforations (C. ostryopsoides, Nothofagoxylon corrugatus, Nothofagoxylon menendezii, Nothofagoxylon scalariforme); we do not know of any Cretaceous fagalean woods with exclusively scalariform perforations, although such woods, that is, early Myricaceae or Ticodendraceae, which have a suite of primitive features found in other families, likely would be difficult to recognize as fagalean. Yet scalariform perforations appear to be in the early divergent branches within some families, including Juglandaceae and Fagaceae.

Vessel-ray pitting (fig. 13B). Vessel-ray pitting is typically simple or with strongly reduced borders and of various shapes in Nothofagaceae, Fagaceae, the Engelhardioideae of the Juglandaceae, and Ticodendraceae. The other taxa have bordered vessel-ray pits similar to the intervessel pits in shape and size, with some Juglandaceae tending toward slightly reduced borders in their vessel-ray parenchyma pits (table 1). Because vessel-ray pits that are simple or with reduced borders characterize the basalmost clades of the order, we hypothesize that they are the ancestral condition. This is relevant for recognizing Cretaceous and Paleogene fossils as potential members of the Fagales.

Vessel grouping and vessel patterns (fig. 13C, 13D). Exclusively solitary vessels occur in Ticodendraceae, Myricaceae (except Myrica), most Fagaceae (only Fagus has some small multiples in addition to solitary vessels), and Casuarinaceae. It is tempting to reconstruct solitary vessels as ancestral, but the basalmost genus Nothofagus has vessels in radial multiples. Vessel grouping shows considerable homoplasy within the order.

Diagonal to dendritic vessel patterns occur more frequently in Fagales than elsewhere in the angiosperms: in Fagaceae (except Fagus), some Casuarinaceae, coryloid Betulaceae, and some Juglandaceae (see table 1). These occurrences considered in light of the phylogenetic topology presented in figure 1 indicate that the condition has arisen convergently within the order at least three times. There are two Late Cretaceous fagalean woods with a diagonal/dendritic vessel arrangement (Page 1970; Wheeler et al. 1987).

Porosity. Most Fagales are diffuse-porous, and this can safely be assumed to be the ancestral character state for the order, as robustly supported by both the molecular phylogenies and the fossil record. Semi–ring porosity occurs in four families, and well-developed ring porosity is limited to the deciduous Castaneoideae of the Fagaceae and two extant genera of the Juglandaceae (Carya, Platycarya), apparently as part of convergent evolution in response to more seasonal climates. Almost all Cretaceous and Paleogene fagalean woods are diffuse-porous, with one report of a possible latest Eocene ring-porous Quercus (Brezinova 1981).

Imperforate tracheary elements. Imperforate tracheary elements in Fagales show great diversity: there are vascular tracheids intergrading with narrow vessel elements in low frequencies in some Betulaceae (especially in the latewood) and temperate Nothofagus; vasicentric tracheids in Castaneoideae and trigonobalanoid Fagaceae and Casuarinaceae; ground tissue fibers with distinctly bordered pits (fiber tracheids or true tracheids sensu Carlquist) in Fagus, Ticodendron, and Myricaceae; fibers with simple to minutely bordered pits (libriform fibers) in Betulaceae and Juglandaceae (Juglandoideae and Rhoipteleoideae); and a mix of libriform fibers and fiber tracheids or an intermediate fiber type (Castaneoideae and Trigonobalanoideae, Casuarinaceae, and Engelhardia of the Juglandaceae) in some Fagales. Septate libriform fibers are unique to Nothofagus p.p.

Various wood anatomical characters considered in light of the Li et al. (2004) phylogeny indicate considerable homoplasy. Vasicentric tracheids are reconstructed as convergent specializations in Casuarinaceae, Castaneoideae, trigonobalanoid Fagaceae, and some Betulaceae. Vessel grouping is strongly associated with the type of background imperforate tracheary elements (Carlquist 1984, 2001), and this also holds true for Fagales. All clades with exclusively (>90%) solitary vessels have ground tissue fiber tracheids and/or vasicentric tracheids as backup hydraulic tissue. Fagus, with its high percentage of solitary vessels, also has fiber pits that are distinctly bordered as well as with reduced borders.

Aggregate rays. Aggregate rays—a “signature feature” of the Fagales—are in fact limited to Castaneoideae p.p. and trigonobalanoid Fagaceae, Casuarinaceae, and Betulaceae p.p., where they occur as aggregations either of uniseriate rays or of broader rays and can be interpreted as “dissected broad rays” or compound rays. Plotting this feature on the Li et al. (2004) phylogeny suggests that they arose twice in convergent evolution.

Parenchyma distribution. Axial parenchyma in Fagales is predominantly apotracheal, ranging from diffuse to diffuse-in-aggregates to narrow bands. Narrow marginal parenchyma bands occur in temperate Nothofagus, Betulaceae, and Juglandaceae. Relatively broad (more than three-seriate) and conspicuous bands are limited to tropical Nothofagus.

Paratracheal parenchyma is limited to scantily paratracheal in the Myricaceae to narrowly vasicentric in Juglandaceae; nonetheless, in these families and all other fagalean families, apotracheal parenchyma is the most common type. We hypothesize diffuse and diffuse-in-aggregates axial parenchyma as the ancestral state for the order, with paratracheal, banded, and marginal parenchyma as convergent and/or divergent specializations.

Recurring Patterns

If the Li et al. (2004, fig. 1) phylogeny is accepted and Nothofagus is ignored as sister to the whole order, there are three major clades in the order: (1) Fagaceae (Fagus + Castaneoideae + Trigonobalanus), (2) Casuarinaceae + Ticodendraceae + Betulaceae, and (3) Myricaceae + Juglandaceae. Wood anatomically, Casuarinaceae is most similar to Castaneoideae, but its other features place it in a different clade. It is striking that in these three clades partially similar patterns have evolved. The clades in more or less basal positions have diffuse porosity and scalariform or mixed scalariform and simple perforation plates (Fagus, Ticodendron, Myricaceae), while those in more derived positions have simple perforations, diagonal vessel patterns and/or ring porosity, and/or vasicentric tracheids and/or libriform fibers and/or aggregate rays (Fagaceae p.p., Casuarinaceae p.p., carpinoid Betulaceae, and Juglandaceae p.p.).

We hypothesize that wood anatomy in each of these three major clades shows convergent adaptive specializations toward (1) hydraulic efficiency, associated with simple perforation plates and wide vessels, as found in the earlywood of ring-porous woods (e.g., Spicer 2016; Robert et al. 2017), and (2) safety, associated with sectored xylem networks as caused by broad or aggregate rays and/or vessel patterning in radial, diagonal, or dendritic sectors that may facilitate bypassing areas with embolized vessels and also confine infections (Carlquist 1987, 2009; Zanne et al. 2006; Morris et al. 2016). However, not all of these adaptations have evolved in each of the three clades. It has been suggested that sectorality is associated with longevity (e.g., Lanner 2002; Bernard et al. 2020). Quercus, which has highly sectored wood, has some of the longest-lived eudicot trees, with multiple reliable reports of trees over 400 years in age (Di Filippo et al. 2015).

Ecological and Functional Adaptations

Herein, we discuss only the following aspects: alternative hydraulic strategies to optimize safety and efficiency and contrasts between tropical and temperate anatomies. Fagales include wood anatomical types that have been associated with contrasting strategies for hydraulic safety and efficiency, as follows: diffuse-porous woods with numerous relatively narrow vessels, ring-porous woods with efficient but vulnerable wide earlywood vessels, and hydraulic safety provided by vasicentric tracheids and narrow latewood vessels. It is noteworthy that these contrasting wood types often grow together in successful species assemblages. For instance, in northwestern Europe and the eastern United States, oak (Quercus), beech (Fagus), hornbeam (Carpinus), hazel (Corylus), alder (Alnus), and chestnut (Castanea) co-occur despite their very divergent wood anatomical structures (Braun 1950; Weeda et al. 1985; Dyer 2006). Apparently, the contrasting combinations of typically safe or efficient features function equally well in ecological settings that are almost identical with regard to water availability and temperature regime.

Fagales with exclusively scalariform plates (Alnus, Betula, Myricaceae, Ticodendron) tend to grow in mesic sites partly subjected to winter frosts, confirming well-documented ecological trends (Baas 1976, 1986; Wheeler and Baas 2019). Rhoiptelea (Juglandaceae) is native to mountainous regions of southern China and Vietnam, regions where scalariform perforation plates with relatively few bars may not be a hydraulic disadvantage. Ticodendron, from montane cloud forests, even fits the so-called xerophobic wood syndrome recognized by Feild et al. (2009).

Although most Fagales are restricted to the temperate zone and the subtropics, a few genera extend into the equatorial tropics. Nothofagus, with subgenus Brassospora, occurs in New Guinea and New Caledonia, and the other two subgenera are restricted to temperate South America, Australia, and New Zealand. The wood anatomical differences fit well-known ecological trends: the tropical clade has wider and fewer vessels and more axial parenchyma (cf. Wheeler et al. 2007; Morris et al. 2015). It would be interesting to analyze critically whether the differences in vessel diameter among Nothofagus species might be explained by differences in tree size and diameter, that is, by larger trees generally having wider vessels (e.g., Olson and Rosell 2013; Olson et al. 2020). The maximum heights reported for the tropical Nothofagus species that Dadswell and Ingle (1954) examined are 45 m; the maximum heights for some higher-latitude Nothofagus are very similar, 40–45 m (https://treesandshrubsonline.org).

In the Fagaceae, the tropical species of Castanopsis, Lithocarpus, and Trigonobalanus are diffuse-porous, while the deciduous Quercus species of the temperate zone are ring-porous. In Quercus, there is a correlation with phenology: deciduous oaks are ring-porous, while evergreen oaks are diffuse-porous to semi-ring-porous.