Int. J. Plant Sci. 178(9):000–000. 2017. q 2017 by The University of Chicago. All rights reserved. 1058-5893/2017/17809-00XX$15.00 DOI: 10.1086/694186 RECONSTRUCTING THE LOCAL VEGETATION AND SEASONALITY OF THE LOWER EOCENE BLUE RIM SITE OF SOUTHWESTERN WYOMING USING FOSSIL WOOD Sarah E. Allen1,* *Department of Biology, University of Florida, and Florida Museum of Natural History, Gainesville, Florida 32611, USA Editor: Michael T. Dunn Premise of research. Permineralized wood along with fossil leaves, reproductive structures, and dispersed pollen and spores are preserved in the uppermost Lower Eocene (∼49.0 Ma) Blue Rim escarpment of the Bridger Formation in southwestern Wyoming. Each component of the ﬂora provides a slightly different view of the diversity and environment, but the wood assemblage provides additional information on the taxonomic composition, forest structure, and local seasonality and climate. Methodology. Fossil wood specimens, including occasional in situ stumps, were collected from Bridger strata exposed at Blue Rim, north of Green River, Wyoming. Thin sections in the three standard orientations were prepared of each specimen and examined by light microscopy. Speciﬁc gravity and vulnerability index values were calculated for well-preserved specimens. Tree height was estimated from diameter measurements obtained from 10 specimens. Pivotal results. Seven wood types are recognized in the assemblage, including a single specimen of Pinaceae, the only conifer macrofossil recovered at Blue Rim to date, and six angiosperms. Angiosperm woods have afﬁnities to Canellaceae, Fabaceae, and Anacardiaceae. The specimen of Canellaceae is the only Blue Rim wood with scalariform perforation plates, predominately uniseriate rays, and mostly solitary vessels. A new Caesalpinioid species, Peltophoroxylon diversiradii, lacks storied structure and has conspicuous paratracheal axial parenchyma. The most abundant wood type has radial canals and prismatic crystals in the ray cells and is assigned to Edenoxylon parviareolatum (Anacardiaceae). Conclusions. While the wood assemblage has signiﬁcantly lower taxonomic diversity compared to the leaves, reproductive structures, and dispersed palynoﬂora, it includes some taxa not otherwise recognized from Blue Rim. These trees, reaching ∼16–28 m tall, would have supported co-occurring climbers in the same ﬂora. Diffuse porosity, generally absent or indistinct growth rings, rare scalariform perforation plates, and the high vulnerability indices of the Blue Rim woods suggest that climate conditions were warm with limited seasonality. Keywords: Anacardiaceae, Canellaceae, Eocene Bridger Formation (Wyoming), Fabaceae, fossil wood anatomy, Pinaceae. Introduction yielded numerous well-preserved impressions of leaves and reproductive structures, dispersed pollen and spores, and siliciﬁed woods. The woods, preserved as stumps and logs, add to the understanding of the local diversity and environment of this area during the latest Early Eocene. A radiometric age of ∼49 Ma has been estimated from just above the most productive fossil layer at BR (M. E. Smith, personal communication, 2016). In this article, the anatomy of the BR woods is described and used to determine taxonomic afﬁnities. In addition to the taxonomic treatment, wood characters are evaluated to interpret broad paleoclimate and paleoecological parameters. Speciﬁc gravity and vulnerability to freezing conditions are estimated from well-preserved woods, and specimens with complete diameters are used to estimate tree heights. Finally, the taxonomic afﬁnities of the BR wood assemblage are compared with other fossil wood assemblages from the Eocene of western North America. Fossils have been collected in the Eocene Bridger Formation of Wyoming for more than a century, but most attention has been devoted to the fauna (Murphey and Evanoff 2011). The plant remains are known only from a few scattered reports of individual taxa, including Lygodium kaulfussi, Landeenia aralioides, Populus cinnamomoides, Phoenix windmillis, and fruits and leaves of Icacinaceae (Manchester and Zavada 1987; Manchester and Hermsen 2000; Manchester et al. 2006; Allen 2015a; Allen et al. 2015). The Blue Rim (BR) escarpment, exposing the lower Bridger Formation in southwestern Wyoming (ﬁg. 1), has 1 E-mail: firstname.lastname@example.org. Manuscript received December 2016; revised manuscript received April 2017; electronically published October 23, 2017. 000 This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). 000 INTERNATIONAL JOURNAL OF PLANT SCIENCES Material and Methods The Bridger Formation (upper Lower Eocene to lower Middle Eocene) is composed of mostly ﬂuvial and reworked volcanic sediments (Murphey and Evanoff 2011). This formation is well known for its vertebrate fauna and is the stratotype for the Bridgerian North American Land Mammal Age (Krishtalka et al. 1987; Murphey et al. 2001; Robinson et al. 2004). The local stratigraphy of BR was documented by Kistner (1973), who noted the occurrence of various fossils, including permineralized wood. Wood was subsequently collected at BR in the mid-1980s by Jane Landeen (who discovered one of the most productive quarry sites), S. R. Manchester, and others. A larger collection was made in 1995 by S. R. Manchester and S. Hack, with additional specimens acquired in 2003, 2009, 2010, and 2014. These wood specimens (more than 30 in total) come from eight different sites (University of Florida [UF] localities 00341S, 15761, 18289, 18591, 19031, 19225, 19338W, and 19406) along the BR escarpment. Other macrofossils have been collected from or very near to each of these sites with the exception of 18591 and 19406, which yielded only fossil wood. Specimens (and slides) are housed in the paleobotanical collection of the Florida Museum of Natural History in Gainesville. The BR wood specimens were mostly allochthonous, but occasional in situ stumps were encountered. Seven of the eight wood sites were within a ∼2.6-km span of the BR escarpment, but locality UF 18591 was much farther south, and when included, the distance between the northernmost and southernmost wood sites expanded to ∼4 km. Some of the BR wood specimens were the subject of a preliminary study by S. R. Manchester and E. A. Wheeler presented at the 2006 Botanical Society of America annual meeting. They recognized four dicotyledonous, diffuse-porous xylotypes. Hand specimens were cut in transverse, radial, and tangential orientations with a Lortone slab saw with a 10-in diamond blade. Resulting wafers were trimmed and mounted to glass slides with epoxy. Sections were ground to ∼30 mm thick or less (if very dark) using a Buehler PetroThin thin-sectioning machine, and coverslips were mounted with Canada balsam and CitriSolv. Specimens were observed using a Nikon Eclipse E600 microscope and photographed with a Canon EOS Rebel Xsi camera. Images were further edited using Adobe Photoshop CS5. Descriptions follow the terminology and deﬁnitions outlined in the International Association of Wood Anatomists (IAWA) list of microscopic features for softwood and hardwood identiﬁcation (IAWA Committee 1989, 2004). Quantitative measurements are indicated by the average ﬁrst, followed by the standard deviation in parentheses, and then the range (except where speciﬁed otherwise). Unless stated in the text, the number of measurements (n) for the following features is as follows: intervessel pit size: n p 10; mean tangential diameter of vessel lumina: n p 30; vessels per square millimeter: n p 6; ray height and ray width: n p 30; and rays per millimeter: n p 10. When multiple specimens of a xylotype were included in a description, quantitative characters were reported as a range of the calculated lowest and highest mean and standard deviation of all the samples followed by the total range (of raw measurements) observed. Measurements of wood anatomical features were completed on photos using ImageJ (Rasband 1997–). Means, standard deviations, and other relevant calculations were completed using Microsoft Excel. Presence (p) or absence (a) of IAWA hardwood features (from IAWA Committee 1989) were entered into the multiple-entry key of the InsideWood website (http://insidewood .lib.ncsu.edu) to obtain lists of taxa with the combinations of features seen in the fossil woods (InsideWood 2004–; Wheeler 2011). Multiple searches with different combinations of features were conducted on InsideWood for each fossil xylotype. The list of features entered for the most useful search is provided in the discussion of each taxon. The results of each InsideWood search were reviewed to try to narrow down the possible afﬁnities of a particular fossil xylotype. Once a few potential families were identiﬁed using InsideWood, primary literature references were consulted for additional information. Angiosperm xylotypes were ordered using the list provided by the Linear Angiosperm Phylogeny Group (Haston et al. 2009). Two approaches were used to estimate the speciﬁc gravity (SG) of seven different well-preserved wood specimens. The ﬁrst Fig. 1 Location of the Blue Rim site (blue dot) in Sweetwater County in southwestern Wyoming. Thick lines demarcate state boundaries, whereas thin gray lines demarcate county boundaries. Major highways are indicated in red for reference. This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). ALLEN—RECONSTRUCTING AN EOCENE SITE USING FOSSIL WOOD followed the methods of Wheeler et al. (2007b). Using ImageJ (Rasband 1997–), 500 points (100 points repeated ﬁve times) were randomly placed on an image of a transverse section taken at 40# total magniﬁcation. The number of times a point fell on lumen (vs. cell wall) was recorded. This allowed the percentage of wall material to be calculated, which was then multiplied by 1.05 (speciﬁc gravity estimate of swollen cell walls, thought to be similar in both dicotyledons and conifers) as indicated by Wheeler et al. (2007b). The second approach used the work of Martínez-Cabrera et al. (2012) and required vessel diameter, ﬁber wall thickness, and ﬁber lumen diameter measurements, which were also completed in ImageJ (Rasband 1997–). Twenty-ﬁve measurements of ﬁber wall thickness and corresponding ﬁber lumen diameter in the tangential direction were measured in addition to average vessel diameter (n p 30, unless otherwise indicated). These were entered into the ﬁrst equation in table 1 of Martínez-Cabrera et al. (2012): SG p 0.802 1 1.346 (WLR) 2 0.22 (LOGvdiam), where WLR is the average ﬁber wall thickness to lumen diameter ratio and LOGvdiam is the log of the average vessel diameter (Martínez-Cabrera et al. 2012). It is important to note that the ﬁbers and/or vessels were frequently compressed and distorted in the BR specimens during or prior to fossilization, reducing the chance that a point would fall in the lumen space and likely leading to inﬂated speciﬁc gravity estimates. Compression also affects the ability to obtain accurate measurements of the original vessel diameter and the ﬁber lumen diameter. MartínezCabrera et al. (2012) note that their values are consistently greater than those one would obtain from a basic speciﬁc gravity measurement (dry mass/green volume) because the speciﬁc gravity estimates they used from extant vegetation were based on an ovendry speciﬁc gravity measurement (oven-dry mass/oven-dry volume; Barajas-Morales 1987; Williamson and Wiemann 2010). However, the samples used in the source data of Barajas-Morales (1987) were oven dried for 24 h at 1057C, which eliminates one source of error (samples oven dried at !1007C) noted by Williamson and Wiemann (2010). The vulnerability index (V) was calculated for 16 BR wood specimens. This index, which has been used to infer a plant’s ability to withstand water stress, is calculated by simply dividing the mean tangential vessel diameter (in micrometers) by the vessel frequency (Carlquist 1975a, 1975b, 1977; Wheeler 1991). Finally, ﬁve methods (detailed below; McMahon 1973; McMahon and Kronauer 1976; Rich et al. 1986; Brown et al. 1989; Niklas 1994; Feldpausch et al. 2011) were used to estimate tree height from complete stem diameters. Systematics and Results Gymnospermae Division—Pinophyta Cronquist, Takht. & W. Zimm. ex Reveal Family—Pinaceae Spreng. ex Rudolphi Genus—cf. Pinus L. Specimen. UF 19406-61964 (ﬁg. 2). Description. Growth rings intermediate between distinct and indistinct (but difﬁcult to determine; variation may be due to 000 preservational differences). Normal axial resin canals present (ﬁg. 2A, 2B), scattered, distorted, averaging 105 mm in diameter (21 mm, n p 9), 82–148 mm. Uniseriate intertracheary pits on radial walls of longitudinal tracheids; pits large, averaging 18.7 mm in diameter (1.9 mm, n p 5), 16.2–20.9 mm (ﬁg. 2F, 2G). The presence or absence of organic deposits in heartwood tracheids and tracheid length not determined due to poor preservation. Mean tangential diameter of longitudinal tracheids 43 mm (7 mm), 26–55 mm (ﬁg. 2C). Some intercellular spaces present, but these areas distorted due to compression during fossilization. Torus poorly preserved but appears smooth in outline. Helical thickenings and callitroid thickenings not observed. Axial parenchyma not observed in either transverse or longitudinal sections. Ray tracheids present with small circular pits, but poorly preserved (ﬁg. 2E). Cross-ﬁeld pitting not well enough preserved to determine. Ray height averages 198 mm (91 mm), 97–464 mm (ﬁg. 2H). Rays average eight cells high (four cells, n p 30), four to 22 cells. Rays mostly uniseriate but occasionally two to three cells wide (ﬁg. 2H). Ray frequency averages to 10 per millimeter (range: 6–16). Radial canals present (diameter p ∼30 mm, n p 1) but sparse and generally obscured in tangential section by the compressed nature of the specimen (ﬁg. 2I, 2J). No minerals or crystals observed. Remarks. Wood is poorly preserved, compressed, and distorted. Intertracheary pits rarely visible, and cross-ﬁeld pitting type indeterminable. The absence of vessels and presence of large circular bordered pits on the tracheids indicates that this wood is coniferous and thus the only nonangiosperm wood specimen from BR. Discussion of afﬁnities. Among the conifers, only a few genera in Pinaceae (Cathaya, Keteleeria, Larix, Picea, Pinus, and Pseudotsuga) have normal intercellular canals (Phillips 1941; Wu and Hu 1997; Heinz 2004; IAWA Committee 2004). Keteleeria has only axial canals, but Cathaya, Pseudotsuga, Pinus, Picea, and Larix have both normal radial and axial canals as was observed in this specimen (Phillips 1941; Hu and Wang 1984; Heinz 2004; IAWA Committee 2004). The BR specimen has some axial canals in pairs, which is common in Pseudotsuga and Larix. Axial canals in Pinus can generally be distinguished from those of other genera because of their thin-walled epithelial cells and larger diameters (Phillips 1941; Wu and Hu 1997; Heinz 2004; IAWA Committee 2004). However, Hilton et al. (2016) note that thin-walled epithelial cells occur occasionally in Larix, Picea, and Pseudotsuga. It is difﬁcult to distinguish epithelial cells from the surrounding tracheids in the BR specimen; however, their absence may be a consequence of being originally thin-walled, making them less likely to preserve and to be compressed beyond recognition. The diameter of axial resin canals in extant species of Larix, Pseudotsuga, and Picea is small (∼40– 100 mm), whereas species of Pinus often have medium-sized canals (∼100–170 mm; IAWA Committee 2004). The axial resin canals in the BR specimen range from 82 to 148 mm in diameter with an average of 105 mm. These measurements may slightly underestimate the diameter of the canals in life because of distortion in the fossil. In addition, Larix often has biseriate (rather than uniseriate) tracheid pitting in its radial walls (Jagels et al. 2001; Heinz 2004; IAWA Committee 2004), and Cathaya has prominent growth rings and a lower ray frequency (four to eight per millimeter; Hu and Wang 1984) compared to the BR specimen (six to 16 per millimeter; average p 10). This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Fig. 2 Pinaceae, UF 19406-61964. A, Overview of the transverse section with scattered axial resin canals. B, Close-up of axial resin canal, TS. C, Close-up of tracheids, TS. D, Overview of radial section. E, Close-up of ray with ray tracheids along the top of the ray, RLS. F, G, Longitudinal tracheids with large uniseriate pits, RLS. H, Rays mostly uniseriate, TLS. I, Two larger rays, possibly with remnant resin canals, TLS. J, Ray with resin canal, TLS. RLS p radial longitudinal section; TLS p tangential longitudinal section; TS p transverse section. This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). ALLEN—RECONSTRUCTING AN EOCENE SITE USING FOSSIL WOOD Given the uniseriate intertracheary pitting, resin canal sizes, and probable thin-walled epithelial cells, this specimen is most likely a representative of Pinus. Wood of Pinaceae cf. Pinus is present in the Eocene Clarno Formation of Oregon, but that taxon has well-deﬁned distinct growth rings and larger axial canals than the BR specimen (Wheeler and Manchester 2002). Two species of Pityoxylon, Pityoxylon aldersoni Knowlton and Pityoxylon amethystinum Knowlton, have been recovered from the Specimen Ridge site of the Eocene Lamar River Formation in Yellowstone National Park (Knowlton 1899; Conard 1930). Both P. aldersoni and P. amethystinum have numerous large axial resin canals, but they also have well-deﬁned growth rings (Knowlton 1899; Conard 1930), unlike the BR specimen. Fusiform rays are rare in P. aldersoni and the BR specimen, but they are abundant in P. amethystinum (Knowlton 1899; Conard 1930). The rays average eight cells high in the BR specimen (range: four to 22 cells); this is intermediate between P. aldersoni, whose rays range from two to 30 cells high with an average of eight to 15, and P. amethystinum, whose rays range from one to 15 cells high with an average of six to eight (Knowlton 1899; Conard 1930). Knowlton (1899) and Conard (1930) also note that P. aldersoni has three to 10 rows of tracheids between each ray in transverse section, whereas P. amethystinum has three to 25 cells between rays in transverse section; the lower number of P. aldersoni aligns with observations on the BR specimen. Based on this information, P. aldersoni has more features in common with the BR specimen, but the presence of welldeﬁned growth rings is an important difference between the Yellowstone and the BR material. Angiospermae Division—Magnoliophyta Cronquist, Takht. & W. Zimm. ex Reveal Family—Canellaceae Mart. Blue Rim Canellaceae Xylotype Specimen. UF 18591-33036 (ﬁg. 3). Description. Growth boundaries indistinct with occasional tangential lines where cells are more crushed (possible thinwalled earlywood; ﬁg. 3A). Wood diffuse-porous with predominantly solitary vessels (92%) and occasional multiples of two (likely where vessel ends overlap; ﬁg. 3A, 3B). Vessels circular to oval in outline. Perforation plates scalariform with 10–20 bars (counted 13–20 bars on seven different plates; ﬁg. 3C, 3E, 3H, 3I). Intervessel pits not observed (probably due to the high proportion of solitary vessels). Vessel-ray pitting with distinct borders (poorly preserved). Mean tangential diameter of vessel lumina 57 mm (14 mm), 28–87 mm. Mean vessel frequency 12 mm2 (range: 4–18 mm2). Some vasicentric tracheids present (alternately, ﬁbers with distinctly bordered pits on both radial and tangential walls; ﬁg. 3C, 3E, 3F, 3H). Fiber pits visible in radial section (possibly in tangential section as well, but preservation is not as good). Fibers thin- to thick-walled. Axial parenchyma likely rare to absent, as it was not observed in any section. Rays exclu- 000 sively uniseriate (tangential section poorly preserved; ﬁg. 3G). Ray height averages 200 mm (85 mm, n p 23), 91–388 mm. Ray cellular composition poorly preserved, appears to have procumbent body cells and one or two rows of upright and/or square marginal cells. Ray frequency ranges from ﬁve to 20 per millimeter (average p 10). No storied structure, intercellular canals, or crystals observed. Vulnerability index. Specimen UF 18591-33036 has an estimated V of 4.8. Discussion of afﬁnities. This specimen is type III as mentioned by Manchester and Wheeler (2006). They noted that the combination of characters suggested afﬁnities to Ericaceae or Myrtaceae. However, further investigation suggested afﬁnities with Canellaceae. The multiple-entry key of InsideWood was searched for these features: indistinct growth rings (2p), diffuse-porous wood (5p), mostly solitary vessels (9p), scalariform perforation plates with 10–20 bars (14p, 16p), vessel-ray parenchyma pits bordered (30p), and exclusively uniseriate rays (96p). This search yielded six matches: Balanops australiana (Balanopaceae), Canella winterana, Cinnamosma sp., Pleodendron sp. (Canellaceae), Columellia sp. (Columelliaceae), and Tepuia venusta (Ericaceae; InsideWood 2004–). Tepuia venusta, the only species of Ericaceae recovered, has solitary vessels with an angular outline, a feature not observed in the BR material. The character of vasicentric tracheids (60p) was not used, because it is possible that the vasicentric tracheids are instead ﬁbers with distinctly bordered pits. If vasicentric tracheids (60p) is added to the same search list as above in InsideWood, Myrtaceae (the second family suggested by Manchester and Wheeler ) is only recovered if two mismatches are allowed (InsideWood 2004–). However, with the addition of two characters (vessel diameter 50–100 mm [41p] and ﬁve to 20 vessels per square millimeter [47p]) to the original list of features with no mismatches allowed, the results were narrowed to C. winterana and Cinnamosma sp. (InsideWood 2004–). Based on this result, a detailed comparison to both the modern genera in Canellaceae and the fossil species Wilsonoxylon edenense N. Boonchai et Manchester was undertaken. Wilson (1960) examined the wood anatomy of Canellaceae. Comparing his descriptions to the BR Canellaceae Xylotype specimen, two (Cinnamodendron, Warburgia) of the ﬁve currently accepted extant genera can be eliminated from consideration. All extant genera have scalariform perforation plates, but Cinnamodendron has signiﬁcantly more bars (50–100) than the BR specimen (13–20). In addition, the BR specimen has uniseriate rays, yet Wilson (1960) noted that Warburgia has rays ranging from one to four, typically two or three, cells wide. Vessel diameter and frequency is similar between the remaining three genera, Canella, Cinnamosma, and Pleodendron, and the BR specimen (Wilson 1960; InsideWood 2004–). Vessels in Pleodendron are usually solitary with a few pairs (similar to the BR Canellaceae Xylotype), whereas they are usually solitary or occasionally in small clusters in Canella and Cinnamosma (Wilson 1960). However, vessels in Pleodendron tend to be arranged in a diagonal/radial pattern (InsideWood 2004 –), which was not observed in the fossil specimen. Canella (ﬁve to 28), Cinnamosma (11–49), and Pleodendron (15–40) have a larger range of bars in their scalariform perforation plates (which can vary to This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Fig. 3 Canellaceae, UF 18591-33036. A, Mostly solitary vessels with tangential lines of crushed cells (darker), TS. B, Close-up view of transverse section showing rare pairs of vessels, likely where vessel element ends overlap. C, Part of a scalariform perforation plate and pits of vasicentric tracheids and/or ﬁbers, RLS. D, Radial section overview. E, Scalariform perforation plate and pits of vasicentric tracheids and/or ﬁbers, RLS. F, Close-up of pits of vasicentric tracheids and/or ﬁbers, RLS. G, Uniseriate rays, TLS. H, Mostly intact scalariform perforation plate, TLS. I, Close-up of scalariform perforation plate, TLS. RLS p radial longitudinal section; TLS p tangential longitudinal section; TS p transverse section. This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). ALLEN—RECONSTRUCTING AN EOCENE SITE USING FOSSIL WOOD reticulate; Wilson 1960) than the BR specimen (13–20), but bars in the fossil material may be undercounted due to preservational bias and the difﬁculties in observing the ends of vessel elements in sections as opposed to macerations. The BR Canellaceae Xylotype has exclusively uniseriate rays, similar to the predominately uniseriate rays found in Canella, Cinnamosma, and Pleodendron (Wilson 1960). There are a few other differences between the BR Canellaceae Xylotype and extant Canellaceae. The BR Canellaceae Xylotype has thin- to thick-walled ﬁbers, but Canella, Cinnamosma, and Pleodendron have thick- to very thick-walled ﬁbers. Paratracheal axial parenchyma is present in all extant genera, Cinnamosma can have oil cells, and Canella and Cinnamosma have rhombohedral crystals (Wilson 1960), but these features were not observed in the fossil. Based on these comparisons, it is not possible to determine which modern genus of Canellaceae might be the closest to the BR specimen, but it is probably not Cinnamodendron or Warburgia. Wilsonoxylon edenense (Canellaceae), from the nearby Eocene Big Sandy Reservoir (BSR) fossil ﬂora (Boonchai and Manchester 2012), shares with the BR Canellaceae Xylotype the features of mostly solitary vessels, scalariform perforation plates with 10–20 bars, and indistinct growth rings. The BR Canellaceae Xylotype is diffuse-porous, and the BSR taxon ranges from semi-ring to diffuse-porous. The BR specimen and W. edenense have mostly solitary vessels with occasional pairs. These pairs may represent where the section crossed overlapping end walls of two vessel elements. However, vessel density is signiﬁcantly different between W. edenense (40–60 mm2; lowest was 22 mm2 in latewood) and the BR specimen (average p 12 mm2). The mean tangential diameter of vessels in the BR Canellaceae Xylotype is ∼57 mm, whereas vessel diameters average 35 and 40 mm in the BSR specimens (Boonchai and Manchester 2012). It is possible that these variations in vessel diameter and density are due to sampling different parts of the plant (e.g., trunk vs. branch) or varying wood maturities. Ray cellular composition is similar between the BR and BSR material. However, W. edenense is clearly distinguished from the BR specimen because it has two- or three-seriate rays rather than uniseriate. This character, ray width, also precludes the BR specimen from conforming to the generic diagnosis of Wilsonoxylon. It does not seem advisable to emend the diagnosis of Wilsonoxylon to include this specimen, because many of the other diagnostic features of that fossil genus are not preserved in the BR specimen. Furthermore, the BR Canellaceae Xylotype is represented by only a single specimen. Mean ray height is also shorter in the BR specimen (200 mm) than in the BSR specimens (1300 mm; Boonchai and Manchester 2012). This BR wood specimen is not well enough preserved to identify to the generic level, but it does ﬁt the broad characters of the family Canellaceae. Extant Canellaceae are generally small to medium-sized evergreen trees. The ﬁve genera are disjunct and found in tropical and subtropical east Africa (Warburgia), Madagascar (Cinnamosma), and the Americas (Canella, Cinnamodendron, and Pleodendron) including South America, the Caribbean, and south Florida (Salazar 2006; Mabberley 2008; Salazar and Nixon 2008; Müller et al. 2015). However, the range of Canellaceae extended to at least southwestern Wyoming in the Eocene based on the presence of W. edenense and this specimen from BR. 000 Family—Fabaceae Lindl. Subfamily—Caesalpinioideae DC. Genus—Peltophoroxylon Müller-Stoll et Mädel Species—Peltophoroxylon diversiradii S.E. Allen, sp. nov. Diagnosis. Diffuse-porous. Vessels in radial multiples of two or three and solitary; perforation plates simple. Intervessel pits generally alternate, small to medium in size. Vessel-ray parenchyma pitting similar to intervessel pitting. Nonseptate ﬁbers present. Prominent paratracheal axial parenchyma: vasicentric, occasional lozenge-aliform, and conﬂuent. Occasional marginal parenchyma bands present or bands more than three cells wide. Rays mostly biseriate, occasionally uniseriate, rarely wider. Ray cellular composition variable from homocellular procumbent to heterocellular with procumbent body cells and one to three rows of upright/square marginal cells. No storied structure. Holotype, hic designatus. UF 19338W-61966 (ﬁg. 4B, 4D– 4F, 4J). Repository. Florida Museum of Natural History (FLMNH), Gainesville, Florida. Type locality. Blue Rim, Sweetwater County, Wyoming. Stratigraphic position and age. Lower Bridger Formation, latest Early Eocene. Paratypes. UF 15761-56323; 18591-33042; 19338W-61965 (ﬁg. 4A, 4C, 4G–4I). Etymology. The speciﬁc epithet is Latin for “diverse rays.” Extant species of Peltophorum and many fossil species of Peltophoroxylon have homocellular rays, with all cells procumbent. However, this taxon has predominately biseriate rays that are both homocellular (all cells procumbent) and heterocellular (procumbent body cells and one to three rows of upright/square marginal cells). Description. Growth ring boundaries variable from weakly deﬁned by irregular and discontinuous bands of marginal parenchyma to indistinct; wood diffuse-porous (ﬁg. 4A– 4C). Vessel arrangement scattered with a weak radial pattern. Vessels solitary (23%) or grouped in radial multiples of two or three, occasional clusters or radial multiples of four or more (ﬁg. 4A– 4C). Infrequent, short tangential bands of vessels present. Solitary vessel outline rounded; perforation plates simple (ﬁg. 4E). Intervessel pits usually alternate, very rarely opposite with a mean diameter across the samples of 5.3 mm (0.3 mm, UF 19338W61965) to 7.1 mm (0.8 mm, UF 19338W-61966) and a total range of 4.8–8.8 mm (ﬁg. 4J). Intervessel pits usually rounded or with four or fewer sides but occasionally polygonal in outline; pit aperture possibly vestured (ﬁg. 4J). Vessel-ray pits are small, inconspicuous, and similar to intervessel pits (ﬁg. 4I). Helical thickenings not observed. Mean tangential diameter of vessel lumina 45 mm (17 mm, UF 19338W-61966) to 74 mm (20 mm, UF 18591-33042) with a total range of 13–108 mm. Vessel frequency variable (range: 0 – 46 mm2, n p 3). Mean vessel element length 173 mm (40 mm, n p 11, UF 15761-56323) to 213 mm (88 mm, n p 30, UF 19338W-61966) with a total range across all specimens of 79–517 mm. Tyloses and ground tissue pits not observed. Nonseptate ﬁbers present; no septate ﬁbers observed. Fibers vary from thin- to very thick-walled. Paratracheal axial This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Fig. 4 Peltophoroxylon diversiradii (Caesalpinioideae s.l., Fabaceae). A, UF 19338W-61965. Vasicentric and banded axial parenchyma, TS. B, UF 19338W-61966. Vessels grouped in short radial multiples, occasional clusters or solitary, TS. C, UF 19338W-61965. Rays and vessels in radial multiples surrounded by axial parenchyma, TS. D–F, UF 19338W-61966. D, Parenchyma clearly visible adjacent to the vessel, RLS. E, Simple perforation plate and heterocellular rays, RLS. F, Ray with procumbent body cells and a row of upright/square marginal cells, RLS. G, H, UF 19338W-61965. G, Overview of tangential section. H, Close-up of mostly biseriate rays and strands of axial parenchyma, TLS. I, UF 15761-56323. Vessel-ray pits, RLS. J, UF 19338W-61966. Alternate intervessel pits, possibly with vestured apertures, TLS. RLS p radial longitudinal section; TLS p tangential longitudinal section; TS p transverse section. This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). ALLEN—RECONSTRUCTING AN EOCENE SITE USING FOSSIL WOOD parenchyma present, with patterns including vasicentric, occasional lozenge-aliform, and conﬂuent (ﬁg. 4A– 4C). Occasional banded parenchyma more than three cells wide, and marginal parenchyma bands present. Axial parenchyma strand length varies from three to more than eight cells. Rays generally one or two cells wide, very rarely wider, with two-seriate the most common condition (ﬁg. 4G, 4H). Rays average 189 mm (77 mm, UF 19338W-61966) to 263 mm (93 mm, UF 15761-56323) high with a total range of 64–522 mm. Ray cellular composition variable. Rays homocellular procumbent or heterocellular with procumbent body cells and one, two, or three rows of upright and/ or square marginal cells (ﬁg. 4D– 4F). Ray frequency ranges from six to 12 per millimeter (n p 3). No storied structure or intercellular canals observed. Rare prismatic crystals in chambered axial parenchyma cells. Vulnerability index. The value of V averages 5.2 (4.4, n p 3); however, the range (1.9–10.3) is large due to variations in vessel frequency among specimens. Remarks. The prominent vasicentric to lozenge-aliform to conﬂuent axial parenchyma differentiates this wood xylotype from others at BR. The absence of radial canals readily distinguishes this from the co-occurring and most common BR wood species, Edenoxylon parviareolatum Kruse. Peltophoroxylon diversiradii corresponds to wood type IV as identiﬁed by Manchester and Wheeler (2006), who noted that these characters are found in Fabaceae, Sapindaceae, and Oleaceae. Further investigations (discussed below) indicate that these wood specimens correspond to Fabaceae. Discussion of afﬁnities. When the most conspicuous and well-preserved features of this xylotype were entered into InsideWood (5p, diffuse-porous; 13p, simple perforation plates; 22p, 24a, and 27a, intervessel pits alternate and not minute or large; 66p, nonseptate ﬁbers present; 69p, ﬁbers thin- to thick-walled; 79p, 80p, 81p, 83p, 85p, and 89p, axial parenchyma vasicentric, aliform, conﬂuent, in bands more than three cells wide and in marginal bands; 97p, rays one to three cells wide; 104p and 106p, all ray cells procumbent and with one row of upright cells; 118a and 120a, rays, axial parenchyma, and vessel elements not storied) with one mismatch allowed, the results were focused on taxa in Caesalpinioideae and the genus Terminalia in Combretaceae (InsideWood 2004–; Wheeler 2011). Terminalia does not match with the fossil material based on its common tyloses, frequent traumatic intercellular canals, and lack of axial parenchyma bands more than three cells wide (InsideWood 2004–). The InsideWood search returned only two members of Sapindaceae and none of Oleaceae—the other two families considered by Manchester and Wheeler (2006). However, if vestured pits (29p, possible in the fossil material) are added to the same feature list as above, Sapindaceae is no longer recovered. None of the 12 Caesalpinioideae genera suggested on InsideWood match the BR specimens based on the presence of one or more characters. However, further comparison within the legumes, especially to Caesalpinioideae, is warranted. The characters of the other two traditional subfamilies of Fabaceae, Mimosoideae and Papilionoideae, were reviewed ﬁrst to ensure that the fossil material ﬁt best with the Caesalpiniodeae. Although the differences in wood anatomy between the traditional subfamilies of Fabaceae are subtle, they can be helpful (Baretta-Kuipers 1981; Wheeler and Baas 1992; Ogata et al. 2008). Members of the Mimosoideae have exclusively homo- 000 cellular rays, whereas ∼30% of Caesalpinioid genera and ∼20% of genera in Papilionoideae have heterocellular rays (BarettaKuipers 1981; Wheeler and Baas 1992; Ogata et al. 2008). Small, low rays with exclusively procumbent cells are rare in Caesalpinioideae and Papilionoideae, but when present in either of these subfamilies, the rays are storied (Baretta-Kuipers 1981). Approximately 75% of Papilionoid genera have all elements storied, whereas this is absent in Mimosoideae and found in less than 25% of Caesalpinioid genera (Baretta-Kuipers 1981; Wheeler and Baas 1992; Ogata et al. 2008). In addition, members of Caesalpinioideae and Mimosoideae rarely have helical thickenings or ring porosity, but both of these characters are common in Papilionoideae (Wheeler and Baas 1992). Furthermore, Caesalpinioid genera are the least likely of the three subfamilies to have one- or two-celled parenchyma strands; ∼35% of taxa have strands with more than four cells, similar to P. diversiradii (Baretta-Kuipers 1981; Wheeler and Baas 1992). As neither storied structure nor small, low, homocellular rays are present in P. diversiradii, both Papilionoideae and Mimosoideae are less likely to accommodate this fossil. Two of the fossil specimens have prismatic crystals in chambered axial parenchyma cells, a common feature in Caesalpinioideae (Melandri and Espinoza de Pernía 2009). This process of elimination suggests that the fossil specimens represent Caesalpinioideae sensu lato. The traditional Caesalpinioideae subfamily has approximately 2250 species in over 170 genera and four tribes (Lewis et al. 2005–; LPWG 2013a, 2013b). Recent phylogenetic work on Fabaceae has found Caesalpinioideae to be paraphyletic; Papilionoideae and Mimosoideae are monophyletic clades that arose from within Caesalpinioideae (LPWG 2013a, 2013b). The Legume Phylogeny Working Group recently released a new subfamily-level classiﬁcation for Fabaceae (LPWG 2017). In this, the legumes are now composed of six subfamilies; the former Mimosoideae is now a clade in a recircumscribed Caesalpinioideae, Papilionoideae is still a subfamily, and Cercidoideae, Detarioideae, Dialioideae, and Duparquetioideae are newly recognized at the rank of subfamily (LPWG 2017). Here, the use of Caesalpinioideae is in the broader, traditional circumscription (one of three subfamilies within Fabaceae). A brief commentary about the new classiﬁcation and its potential effect on the placement of P. diversiradii is provided at the end of this discussion. Taxonomic afﬁnities within extant Caesalpinioideae. BarettaKuipers (1981) divided the Caesalpinioideae into two groups based on wood anatomy. The ﬁrst group, composed of the tribes Caesalpineae, Cassieae, and Cercideae, is characterized by homocellular rays averaging 450 mm in height. The wood anatomy of eight genera within Caesalpinieae has been explored further, and of these, Dimorphandra shares the most characters with the BR specimens (Espinoza de Pernía and Melandri 2006). Mora is also similar but generally has homocellular rays, whereas the fossils have both homocellular and heterocellular rays. Characters that exclude the other genera studied in Caesalpinieae include regular, very large parenchyma bands; intervessel pits that are too large; mostly or exclusively uniseriate rays; or having storied structure (Espinoza de Pernía and Melandri 2006). The second group within Caesalpinioideae includes the tribes Detarieae and Amherstieae (Amherstieae is not currently recognized as a tribe; Lewis et al. 2005–; LPWG 2013a) and has mostly heterocellular rays with an average height of 650 mm (Baretta-Kuipers 1981). The wood anatomy of 10 genera in the tribe Detarieae This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). 000 INTERNATIONAL JOURNAL OF PLANT SCIENCES has been explored in detail, and four (Copaifera, Elizabetha, Eperua, and Heterostemon) can be eliminated from consideration due to either the presence of axial intercellular canals or mostly/exclusively uniseriate rays, as neither of these characters is present in P. diversiradii (Melandri and Espinoza de Pernía 2009). Many of the genera in both Caesalpinieae and Detarieae have large intervessel pits, but Brownea, Cynometra, Dimorphandra, and Mora tend to have smaller pits that are more consistent with P. diversiradii (Melandri and Espinoza de Pernía 2009). However, these genera have other characters that do not align with the fossil material (InsideWood 2004 –). The similarity of wood characters across Caesalpinioideae s.l. and the inability to observe important features in the fossil specimens that could help narrow down the possible afﬁnities precludes a taxonomic assignment to an extant genus in Caesalpinioideae. Comparison to other fossil wood of Fabaceae. Three fossil legume woods from the middle Eocene Clarno Formation in Oregon were also compared with P. diversiradii (table 1; Wheeler and Manchester 2002). All three Clarno taxa, Dichrostachyoxylon herendeenii, cf. Euacacioxylon, and cf. Mimosoxylon, have a larger average tangential diameter of vessels and differing ray width and axial parenchyma patterns as compared to P. diversiradii (table 1; Wheeler and Manchester 2002). Intervessel pits are larger in two of the Clarno taxa, and average ray height in D. herendeenii is similar to the BR specimens, but it is greater in both cf. Euacacioxylon and cf. Mimosoxylon (table 1; Wheeler and Manchester 2002). Although there is some overlap in characters between each of the Clarno legumes and P. diversiradii, there are enough differences to conclude that they are distinct. Fritz and Fisk (1978) noted a possible fabaceous wood from Yellowstone National Park. Recent examination of “Laurinoxylon pulchrum” Knowlton indicates that it is a legume rather than Lauraceae (InsideWood 2004–; E. Wheeler, personal communication, 2016). However, “L. pulchrum” differs from the BR material in having locally storied parenchyma and semi-ring porosity (Wheeler and Baas 1992; InsideWood 2004–). Legume wood has also been documented from Eocene strata in the Willwood Formation in Park County, Wyoming. Whereas the Willwood material has features similar to P. diversiradii in transverse section, it differs in having well-deﬁned storied structure in longitudinal section (Wheeler and Baas 1992). Assignment to Peltophoroxylon. Müller-Stoll and Mädel (1967) established multiple genera for fossil legume wood. The BR legume is not a perfect match to any of these, but it is most closely aligned with Peltophoroxylon, in part because the rays range to having one row of upright marginal cells rather than being exclusively homocellular procumbent. The main differences are that the ﬁbers are generally septate (rather than nonseptate) and the parenchyma is occasionally in zig-zag-shaped bands in Peltophoroxylon (Müller-Stoll and Mädel 1967), neither of which was observed in the fossil specimens. If “ﬁbers not septate” is selected in Müller-Stoll and Mädel’s (1967) key to the fossil legume genera and the characters matching the fossil are followed (as opposed to following the key to Peltophoroxylon), it terminates at the genera of Berlinioxylon, which usually has uniseriate homocellular rays, or Pahudioxylon, which also has homocellular rays and distinct growth rings with large vessels, unlike the fossil material. Peltophoroxylon includes fossil woods with features similar to extant Cassia, Peltophorum, and Xylia (Müller-Stoll and Mädel 1967; Bande and Prakash 1980; Awasthi 1992). Prakash (1975) later excluded fossil woods from Peltophoroxylon with similarities to extant Cassia and created a new fossil genus, Cassinium. However, the broader diagnosis of Peltophoroxylon is retained here, as these two genera are very similar, and minor differences are challenging to recognize in fossils. Multiple species of Peltophoroxylon have been documented, primarily from the Miocene of southern Asia (Bande and Prakash 1980; Awasthi 1992; InsideWood 2004–; Gregory et al. 2009). Most of these species have signiﬁcantly larger vessel diameters than the BR taxon, and many have exclusively homocellular (all cells procumbent), wider, or storied rays (Awasthi 1992; InsideWood 2004–). In the Western Hemisphere, Peltophoroxylon uruguayensis was recently documented from the late Pleistocene of eastern Argentina (Ramos et al. 2014). Peltophoroxylon uruguayensis shares the feature of predominately biseriate rays with P. diversiradii, but its rays are exclusively homocellular procumbent (Ramos et al. 2014). The taxon most similar to the BR specimens is Peltophoroxylon sp. from the late Eocene marine sands in Helmstedt, Lower Saxony, Germany (Gottwald 1992). This species has many features in common with P. diversiradii, including a lack of clear growth rings, small-diameter vessels, similar-sized intervessel pits, and the presence of crystals (Gottwald 1992). However, the Helmstedt taxon has predominately uniseriate rays that are homocellular and only weakly heterocellular, with some marginal ray cells tending to be square. Extant Peltophorum remains within Caesalpinioideae even with the new subfamily classiﬁcation of legumes (LPWG 2017). A wood anatomy feature that separates the new subfamilies is the presence or absence of vestured pits. Vestured pits are absent in the newly recognized subfamilies of Duparquetioideae, Cercidoideae, and most genera of Dialioideae, whereas they are present in the other three subfamilies (Quirk and Miller 1985; Herendeen 2000; Gasson et al. 2003; Herendeen et al. 2003; Bruneau et al. 2008; LPWG 2017). It is difﬁcult to observe vestured pits with light microscopy in modern woods and even more challenging in fossil woods that are not always well preserved. Even at high magniﬁcation, it was not possible to conﬁdently say whether the pit apertures on the BR specimens are vestured or are simply obscured by debris. Family—Anacardiaceae R. Br. Genus—Edenoxylon Kruse Species—Edenoxylon parviareolatum Kruse emend. N. Boonchai et Manchester Specimens. UF 00341S-61961; 15761-56320; 18289-56305, 56306, 56308; 18591-33008, 33021, 33023, 33030, 33034, 33041 (ﬁgs. 5, 6). Description. Growth ring boundaries indistinct (occasional changes in ﬁber preservation but discontinuous) to absent. Wood diffuse-porous with vessels solitary (27%), in radial multiples of two to four, occasionally in radial multiples of ﬁve or six, or rarely clusters (ﬁgs. 5A, 6A, 6B). Solitary vessel outline rounded. Perforation plates simple. Intervessel pits alternate to rarely opposite, minute to small with an average across specimens of 3.8 mm (0.4 mm, UF 00341S-61961) to 4.8 mm (0.7 mm, This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Note. Diffuse 45–74 5–7 Paratracheal: vasicentric, aliform, occasionally lozenge-aliform, conﬂuent, occasionally banded/ marginal bands 189–263 1–2-seriate, rarely larger Homocellular procumbent or heterocellular with procumbent body cells and one to three rows of upright/square marginal cells Absent Rare, in chambered axial parenchyma Peltophoroxylon diversiradii 166–259 1– 4-seriate Homocellular procumbent or heterocellular with procumbent body cells and one or occasionally two rows of upright/square marginal cells Absent Present, in chambered axial parenchyma Diffuse 123–186 8–10 Paratracheal: vasicentric, aliform, occasionally conﬂuent Dichrostachyoxylon herendeenii Absent Not observed 282–311 Mostly 3– 4-seriate Homocellular procumbent or heterocellular with procumbent body cells and one row of upright/square marginal cells Diffuse 98–147 4–6 Paratracheal: vasicentric, aliform, conﬂuent-banded cf. Euacacioxylon Clarno Formation Absent Not observed 476 1– 4-seriate Heterocellular with procumbent body cells and one or occasionally two rows of upright/square marginal cells Diffuse 162 10–12 Scanty paratracheal to narrow vasicentric cf. Mimosoxylon Clarno Formation data from Wheeler and Manchester (2002). IP p intervessel pit size; RH p mean ray height; VD p mean tangential diameter of vessel lumina. Storied structure Crystals RH (µm) Ray width Ray cellular composition Porosity VD (µm) IP (µm) Axial parenchyma Character Bridger Formation Comparison of the Blue Rim (Bridger Formation, Wyoming) and Clarno Formation (Oregon) Legumes Table 1 Fig. 5 Edenoxylon parviareolatum (Anacardiaceae), UF 00341S-61961 (all except B). A, Vessels grouped in radial multiples and solitary, TS. B, UF 18289-56306. Radial section showing ray with procumbent body cells and upright/square marginal cells. C, Close-up of vessels and ﬁbers, RLS. D, Crystals common in square/upright ray cells, RLS. E, Alternate intervessel pits and crystals in the adjacent ray cells, TLS. F, Vessel elements with alternate intervessel pits interspersed between rays with prismatic crystals, TLS. G, Ray with well-deﬁned radial canal, TLS. H, Broader view of tangential section. RLS p radial longitudinal section; TLS p tangential longitudinal section; TS p transverse section. 000 This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Fig. 6 Edenoxylon parviareolatum. A, UF 18591-33008. Transverse section overview. B, UF 18289-56308. Tyloses present in the vessels, TS. C, UF 18591-33008. Close-up of ﬁber cells, TS. D, UF 18289-56308. Tangential section overview with radial canals. E, UF 18591-33008. Multiseriate ray with radial canal. Ray to the lower right has a prismatic crystal, TLS. F, UF 18591-33030. Double radial canal and vessel elements, TLS. G, UF 18591-33023. Ray with two radial canals, TLS. H, I, UF 18289-56305. H, Ray cells with prismatic crystals, RLS. I, Oblong rounded to horizontal (gash-like) vessel-ray pitting, RLS. J, UF 18289-56306. Septate ﬁbers, RLS. RLS p radial longitudinal section; TLS p tangential longitudinal section; TS p transverse section. 000 This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). 000 INTERNATIONAL JOURNAL OF PLANT SCIENCES UF 18591-33030), with a total range of 3.2–6.6 mm (ﬁg. 5E). Vessel-ray pitting not preserved in many specimens, but pits oblong rounded or horizontal with much-reduced borders to apparently simple when observed (ﬁg. 6I). Vestured pits and helical thickenings not observed. Mean tangential diameter of vessel lumina across all specimens 47 mm (15 mm, UF 18591-33030) to 78 mm (17 mm, UF 18289-56306), with a total range of 14– 113 mm. Vessels per square millimeter ranges from eight to 54. Tyloses present (occasionally appearing sclerotic), often extensive and obscuring vessel element end walls (ﬁgs. 5C, 5H, 6B). Fibers with simple to minutely bordered pits. Helical thickenings in ground tissue ﬁbers not observed. Both septate and nonseptate ﬁbers present (ﬁg. 6J). Fibers occasionally very thin- or very thick-walled, more frequently thin- to thick-walled (ﬁg. 6C). Axial parenchyma present but uncommon, diffuse apotracheal, scanty paratracheal, and in thin, discontinuous marginal bands. Axial parenchyma with approximately three to more than eight cells per strand. Rays mostly one or two cells wide (ﬁgs. 5E, 5F, 5H, 6D), occasionally three (wider in rays with radial canals). When single cells present in center of multiseriate ray, width of uniseriate portion equals width of neighboring multiseriate portion. Ray width (without canals) averages 17 mm (5 mm, UF 1859133008) to 21 mm (6 mm, UF 18591-33030), with a total range of 7–36 mm. Ray height averages 167 mm (48 mm, UF 1828956308) to 253 mm (98 mm, UF 18591-33008), with a total range of 94– 487 mm. Ray cellular composition variable ranging from body cells procumbent with one to more than four rows of upright/square cells to procumbent, square, upright mixed throughout ray (ﬁgs. 5B, 5D, 6H). Rays per millimeter ranges from six to 20. No storied structure. Radial canals present (ﬁgs. 5G, 6D–6G) but variable in frequency among specimens. Across specimens, for rays with a single canal, the average heights are 345 mm (129 mm, n p 6, UF 18289-56308) to 509 mm (288 mm, n p 2, UF 18289-56305), with a total range of 194– 712 mm; their average widths are 60 mm (6 mm, n p 3, UF 18591-33030) to 98 mm (18 mm, n p 14, UF 18591-33023), with a total range of 47–136 mm; the canal openings have average lengths of 50 mm (16 mm, n p 6, UF 18289-56308) to 92 mm (42 mm, n p 2, UF 18289-56305), with a total range of 26– 121 mm, and average widths of 28 mm (4 mm, n p 3, UF 1859133030) to 61 mm (17 mm, n p 14, UF 18591-33023), with a total range of 21–97 mm. Rays with two canals (three examples measured from all specimens; ﬁg. 6F, 6G) range from 437 to 499 mm high by 71 to 123 mm wide. Within these rays, canal size ranges from 35 to 81 mm high by 21 to 70 mm wide. Radial canals generally oblong in shape but occasionally more rounded. Well-preserved canals with a ring of well-deﬁned epithelial cells around the inside of the canal opening. Prismatic crystals present, common in upright/square and procumbent ray cells (ﬁgs. 5D, 6H). Rarely more than one crystal of the same size per cell. Crystals in the ray cells average 16 mm (2 mm, n p 5, UF 18591-33030) to 20 mm (3 mm, n p 13, UF 18591-33021) in the longest dimension, with a total range of 12–29 mm. Occasionally prismatic crystals in axial parenchyma and/or ﬁbers. Speciﬁc gravity. Two approaches were used to estimate the speciﬁc gravity of six specimens of this wood xylotype. Speciﬁc gravity averages 0.75 (0.10, range: 0.58–0.87) using the meth- ods of Wheeler et al. (2007b), whereas using equation (1) of Martínez-Cabrera et al. (2012), the average speciﬁc gravity is 1.05 (0.20, range: 0.82–1.35). Vulnerability index. Values of V were calculated for 10 of the 11 E. parviareolatum specimens (all except UF 1859133041) for an average of 3.0 (1.9, range: 1.4–5.7). Remarks. This is the most common wood type at BR, and it is represented by stems of varying diameters from different sites across the escarpment. Most samples are only small fragments of larger logs or branches (e.g., UF 18591-33008 is ∼3.5 cm in diameter), but they are occasionally much larger (e.g., UF 18591-33023 is ∼30.5 cm in diameter). Although all BR specimens with numerous tyloses, radial canals, and prismatic crystals are considered to be a single xylotype, there are differences. A few specimens (e.g., UF 18289-56306, UF 1859133021, and UF 18591-33023) have signiﬁcantly lower vessel frequency (∼8–20 mm2) as compared to the other specimens (∼24– 44 mm2). This variation might be due to differences in wood maturity or origin in the plant (e.g., trunk vs. branch wood). Fichtler and Worbes (2012) found large differences in vessel frequency within a species and between sites. The frequency of radial canals is also variable among specimens. A few specimens (e.g., UF 18289-56308 and UF 18591-33023) have numerous radial canals, but they are rare in other specimens (e.g., UF 00341S-61961 and UF 18591-33021). The frequency of radial canals has been shown to vary greatly between individuals, the location in the stem (e.g., close to the pith), or the maturity of the wood (Chattaway 1951; Terrazas and Wendt 1995). Discussion of afﬁnities. The most distinguishing feature of this xylotype, radial intercellular canals, is found in several families of the Sapindales, including Anacardiaceae and Burseraceae in addition to some species of Fabaceae, Apiaceae, Araliaceae, Moraceae, Clusiaceae, Dipterocarpaceae, and Euphorbiaceae (InsideWood 2004–; Carlquist 2012). The rhomboidal prismatic calcium oxalate crystals found in many of the rays of these specimens can also be of systematic value (Carlquist 2012). When the characters 2p (growth rings indistinct/absent); 5p (diffuse-porous); 13p (simple perforation plates); 22p, 26a, and 27a (intervessel pits alternate, not medium or large); 30a (vessel-ray pitting different from intervessel pits); 56p (tyloses common); 97p (rays one to three cells wide); 130p (radial canals); and 136p (prismatic crystals present) were entered into InsideWood with 0 mismatches allowed, there were ﬁve results: Melanochyla sp., Myracrodruon urundeuva, Pistacia terebinthus (Anacardiaceae), Canarium indicum (Burseraceae), and Baloghia sp. (Euphorbiaceae; IAWA Committee 1989; InsideWood 2004–; Wheeler 2011). Although none of these taxa were a match to the BR material, Anacardiaceae was the best ﬁt overall, which matches the ﬁndings of Manchester and Wheeler’s (2006) wood type I, which they also identiﬁed as Anacardiaceae. Taxonomic afﬁnities within extant Anacardiaceae. The minute to small (!7-mm) intervessel pits in the BR specimens were used to determine which extant genera of Anacardiaceae to compare with the fossils. This character is usually consistent even if the wood is of variable maturity or from a different part of the plant. Faguetia, Haplorhus, Lithraea, Loxostylis, Micronychia, Rhus, and Trichoscyphya have intervessel pits under 7 mm (Terrazas 1994). However, Faguetia, Lithraea, and Micronychia do not have radial canals, which is a signiﬁcant contrast with the BR specimens, so they were not considered further. This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). ALLEN—RECONSTRUCTING AN EOCENE SITE USING FOSSIL WOOD The remaining genera, Haplorhus, Loxostylis, Rhus, and Trichoscypha, are all members of the tribe Rhoeae (Pell 2004). Haplorhus, an evergreen shrub or small tree found in Peru and Chile, has small intervessel pits (Terrazas 1994), but none of the other quantitative characters examined are in the same range as the BR specimens. In addition, Haplorhus has homocellular rays with very few prismatic crystals (Terrazas 1994), which differs from the BR material. Loxostylis, a small evergreen tree or shrub found in South Africa, is one of a few genera in Anacardiaceae with toothed leaf margins (Terrazas 1994). Vessel frequency and diameter in Loxostylis matches the ranges observed on the BR specimens. Furthermore, ray height in Loxostylis is on the shorter side for Anacardiaceae with an average of 318 mm (Terrazas 1994). Rhus can be evergreen or deciduous and is found in temperate to tropical regions of Asia and North America (Terrazas 1994). Some species of Rhus have rays !300 mm high, and vessel frequency and diameter are also similar to the BR specimens. Trichoscypha, an evergreen tree from tropical West Africa, has silica bodies and signiﬁcantly higher rays (Terrazas 1994) than the BR Edenoxylon specimens. Additional characters that are well preserved in the BR Edenoxylon (tyloses, ﬁbers, axial parenchyma, ray cellular composition, and crystals) were compared to the three remaining genera (Loxostylis, Rhus, and Trichoscypha). Although none of these have characters that speciﬁcally exclude it, Rhus is the best match (of extant genera) with the fossil specimens because it has frequent prismatic crystals in both marginal and procumbent ray cells and similar vessel-ray pits (characters 31 and/or 32; InsideWood 2004 –). Furthermore, Rhus is the most geographically widespread taxon today, and both Loxostylis and Trichoscypha are conﬁned to Africa (Terrazas 1994). Wood of temperate Northern Hemisphere Rhus is typically semi-ringporous to ring-porous (Boonchai and Manchester 2012); however, 10 of the 24 Rhus species in InsideWood are diffuse-porous (InsideWood 2004–). Most of the diffuse-porous species have wider vessels and much larger intervessel pits than observed in the BR specimens (Boonchai and Manchester 2012). Leaves assigned to Rhus nigricans (Lesquereux) Knowlton in the Green River ﬂora have also been found at BR (MacGinitie 1969). However, this wood xylotype might not represent a living genus. At least one extinct anacardiaceous genus, Pentoperculum, which is known from fruits of tribe Spondioideae (Manchester 1994; Collinson et al. 2012), is also found at BR. Comparison to E. parviareolatum. Edenoxylon parviareolatum, established by Kruse in 1954, is an anacardiaceous fossil wood that was ﬁrst described from the Green River Formation Hays’ Ranch site (∼16 mi to the east of Farson) in southwestern Wyoming and later was found to be represented by numerous specimens from the BSR ﬂora, also from southwestern Wyoming (Kruse 1954; Boonchai and Manchester 2012). Edenoxylon parviareolatum from BSR has growth ring boundaries weakly deﬁned by a change in ﬁber radial diameter and marginal parenchyma. The BR specimens have absent to indistinct growth rings, although specimen UF 18591-33034 did have variations in ﬁber preservation across the transverse section. The distinctness of growth rings can vary within an individual tree, as rings tend to be more obvious in trunks than in branches (Tarelkin et al. 2016). Furthermore, coding of growth ring distinctness can vary between researchers, and some of the features that deﬁne growth rings such as changes in ﬁber wall 000 thickness or radial diameter (IAWA Committee 1989; Fichtler and Worbes 2012; Tarelkin et al. 2016) are difﬁcult to see on fossils and may be masked by poor preservation. The BR specimens are diffuse-porous, whereas some of the BSR specimens are semi-ring-porous. All BR specimens have signiﬁcantly lower vessel frequency (∼8–54 mm2) than E. parviareolatum from other sites (56–70 mm2 at BSR and ∼100 mm2 at Hays’ Ranch; Kruse 1954; Boonchai and Manchester 2012). However, vessel frequency was recounted using the images of the BSR E. parviareolatum and the corresponding scale bar (Boonchai 2012; Boonchai and Manchester 2012) with the same ImageJ approach as the BR specimens, and lower numbers that overlapped with the BR material were obtained. Moreover, there can be considerable variation in wood anatomical characters both within species and between sites (Fichtler and Worbes 2012). Vessel arrangement and diameter are comparable in both the BR and the BSR specimens, and both have simple perforation plates. Due to the numerous tyloses obscuring end walls, mean vessel element length could be determined in only two BR specimens (average: 80 and 108 mm; range: 30–195 mm), and they are much shorter than in the BSR specimens (average: 224–362 mm; Boonchai and Manchester 2012) but within the range of the Hays’ Ranch material (100–300 mm; Kruse 1954). It is probable that a few of the measurements for vessel element length in the BR specimens are less than the actual length of the vessel element if a tylose mimicked an end wall. The BSR, Hays’ Ranch, and BR specimens all have numerous tyloses and alternate intervessel pits that are comparable in size (Kruse 1954; Boonchai and Manchester 2012). The BR and BSR specimens have similar axial parenchyma patterns and ray width, and black deposits ﬁll the lumina of some of the vessels in E. parviareolatum from both sites (Boonchai and Manchester 2012). In general, rays without radial canals are shorter in the specimens from BR. However, rays with radial canals are similar or larger in size in the BR specimens as compared to E. parviareolatum from BSR. The BR specimens also have fewer rays per millimeter than the BSR specimens. The BR, BSR, and Hays’ Ranch specimens all have septate ﬁbers and prismatic crystals in the rays and share similar ray cellular composition and vessel-ray pitting type (Kruse 1954; Boonchai and Manchester 2012). Kruse’s (1954) stem of E. parviareolatum from Hays’ Ranch is slightly different from the BSR specimens in that axial parenchyma is rarer and the rays are almost exclusively uniseriate (Boonchai and Manchester 2012). The mean tangential vessel diameter of 50 mm in Kruse’s specimen is closer to the averages of the BR specimens than the BSR material (Kruse 1954; Boonchai and Manchester 2012). However, the differences in vessel diameter could relate to plant maturity. Comparison to other fossil wood of Anacardiaceae. Wood of three species of Anacardiaceae, Maureroxylon crystalliphorum, Tapirira clarnoensis, and Terrazoxylon ductifera, are documented from the middle Eocene Clarno Nut Beds in Oregon and warrant comparison to the BR material (Wheeler and Manchester 2002). All three Clarno species have larger-diameter vessels, larger intervessel pits, and higher rays as compared to the BR specimens. Maureroxylon crystalliphorum further differs from the BR E. parviareolatum specimens in that it has clearly deﬁned growth rings, wider rays, and radial canals are absent (Wheeler and Manchester 2002). Tapirira clarnoensis and Terrazoxylon ductifera have radial canals. All three Clarno taxa This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). 000 INTERNATIONAL JOURNAL OF PLANT SCIENCES have tyloses and crystals, as does the BR material. However, none of the Clarno taxa are similar enough to the BR xylotype to suggest that it is the same taxon. The fossil species, Rhus crystallifera (Anacardiaceae), has been documented from the Eocene Amethyst Mountain locality in Yellowstone National Park (Wheeler et al. 1978). It differs from the BR E. parviareolatum in having distinct growth rings; semi-ring porosity; vessels mostly solitary with occasional clusters, especially in the latewood; larger intervessel pits; and no radial canals (Wheeler et al. 1978). Boonchai and Manchester (2012) also compared the BSR specimens to other Eocene anacardiaceous woods, including Edenoxylon aemulum found in Kent, England (Brett 1966; Wheeler and Manchester 2002). Brett (1966) provided a generic diagnosis for Edenoxylon (unlike Kruse who had a combined genus and species description), which matches the BR material. Furthermore, Brett (1966) noted that E. parviareolatum, so named for the small intervessel pits, matches a few extant genera in the tribe Rhoeae, including Rhus, Faguetia, and Trichoscyphya, to which the similarities with the BR material have already been discussed. Most of these fossil anacardiaceous wood species, including the BR specimens, are diffuse-porous (R. crystallifera is semi-ring-porous), with similar vessel arrangements, heterocellular rays, and simple perforation plates. However, of the Eocene woods assigned to Anacardiaceae, the BR material is most closely aligned with E. parviareolatum. Indet. Family Incertae Sedis Blue Rim Xylotype 1 Specimen. UF 19225-54694 (ﬁg. 7A–7C). Description. Likely diffuse-porous wood with growth rings indistinct or absent. Vessels solitary (40%), in radial multiples of two or occasionally three or four (ﬁg. 7A). Rare clusters of vessels (often three) tangentially arranged. Solitary vessel outline circular to oval. Perforation plates simple. Intervessel pits poorly preserved and challenging to measure but appear alternate, averaging to 5.8 mm (1.0 mm), 4.4–7.3 mm. Vessel-ray pitting not observed (rays ﬁlled with dark deposits). Mean tangential diameter of vessel lumina 57 mm (15 mm, n p 24), 23–86 mm. Transverse view poorly preserved, but vessel frequency averages 14 mm2 (range: 0–25 mm2). Some of the vessels in transverse and radial section are ﬁlled with black deposits (ﬁg. 7A); it is unclear whether these are tyloses or other deposits (e.g., gum residue), as vessels are not obscured in tangential section. Fibers very thickwalled (ﬁg. 7B). No axial parenchyma observed (but it might be due to the poor preservation). Rays frequently two-seriate, occasionally one or three cells wide (ﬁg. 7C). Rays vary from exclusively uniseriate to having alternating uniseriate and multiseriate portions. In some rays, the uniseriate and multiseriate portions are equal in width. Rays average 227 mm in height (85 mm), 92– 414 mm. Ray cellular composition not clear; many procumbent cells, possibly all procumbent. Black deposits present in many of the ray cells. Rays closely spaced in tangential section with an average of 22.8 rays per millimeter (range: 16–28). Rays not storied. No intercellular canals observed. Remarks. This specimen, BR Xylotype 1, is poorly preserved but does not have the diagnostic characters of any of the taxonomically identiﬁed wood types at BR. The features that could be coded with conﬁdence were entered into InsideWood (with 0 mismatches): 2p (growth rings indistinct/absent), 5p (diffuseporous), 9a (vessels not exclusively solitary), 13p (simple perforation plates), 22p (intervessel pits alternate), 41p (vessel diameter 50–100 mm), 70p (ﬁbers very thick-walled), 97p (rays one to three cells wide), 104p (all ray cells procumbent), 116p (≥12 rays per millimeter), and 118a (rays not storied). This search yielded 19 results in 10 families (InsideWood 2004–); these were reviewed, but none was found to be a good match to the fossil specimen. BR Xylotype 1 will be compared to the other two unidentiﬁed specimens, BR Xylotype 2 (UF 18591-33017) and BR Xylotype 3 (UF 19225-57353), following the discussion of BR Xylotype 3. Incertae Sedis Blue Rim Xylotype 2 Specimen. UF 18591-33017 (ﬁg. 7D–7F). Description. Growth ring boundaries indistinct to absent (areas with fewer and smaller vessels, but these are not continuous). Wood diffuse-porous, with vessels tending to be weakly arranged in short diagonals or discontinuous tangential bands. Vessels solitary (38%) and in radial multiples of two to three, rarely four, or occasional clusters (ﬁg. 7D). Solitary vessels oval to rounded in outline. Perforation plates not observed but likely simple. Intervessel pits alternate (ﬁg. 7E), apertures frequently slit-like, small to medium, averaging 6.8 mm across (0.4 mm), 6.3–7.5 mm. Mean tangential diameter of vessel lumina 75 mm (29 mm), 29–136 mm. Vessel frequency averages 11 mm2 (range: 6–16 mm2). Mean vessel element length 169 mm (62 mm, n p 15), 85–337 mm (ﬁg. 7F). Axial parenchyma could not be distinguished from ﬁbers due to the poor preservation of the specimen (compressed and distorted). Ray start and end points difﬁcult to distinguish but average 209 mm in height (62 mm, n p 19), 163–314 mm (ﬁg. 7F). Rays mostly one or two cells wide, occasionally wider, often biseriate. Rays frequent and closely spaced, averaging 14 per millimeter (range: 10–16). Large black spherical structures present in ray cells. Storied structure, intercellular canals, and crystals not observed. Vulnerability index. BR Xylotype 2 has an estimated V of 6.8. Remarks. This specimen (and possibly others) was described as type II by Manchester and Wheeler (2006). BR Xylotype 2 is poorly preserved, but it lacks the diagnostic features of the previously described BR woods assigned to Canellaceae, Fabaceae, and Anacardiaceae. The following characters were entered into InsideWood: 2p (growth rings indistinct to absent), 5p (wood diffuse-porous), 13p (simple perforation plates), 22p (intervessel pits alternate), 25p (intervessel pits small), 41p (mean tangential diameter of vessel lumina 50–100 mm), 47p (ﬁve to 20 vessels per square millimeter), 52p (mean vessel element length ≤350 mm), and 97p (ray width one to three cells), and modern wood was searched with no mismatches allowed. This recovered 282 results spanning 46 families (InsideWood 2004–). The characters of this wood will be compared to the other two unknown xylotypes at BR following the discussion of BR Xylotype 3. This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). Fig. 7 Incertae sedis specimens. A–C, Blue Rim Xylotype 1, UF 19225-54694. A, Vessels generally grouped in short radial multiples, TS. B, Very thick-walled ﬁbers, TS. C, Biseriate rays, TLS. D–F, Blue Rim Xylotype 2, UF 18591-33017. D, Vessels grouped in radial multiples, clusters, or solitary, TS. E, Alternate intervessel pits, TLS. F, Rays and vessels, TLS. G–I, Blue Rim Xylotype 3, UF 19225-57353. G, Vessels solitary, grouped in short radial multiples, or occasionally in clusters, TS. H, Rays often biseriate, TLS. I, Close-up of rays; enlarged white cells may represent idioblasts, TLS. TLS p tangential longitudinal section; TS p transverse section. This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). 000 INTERNATIONAL JOURNAL OF PLANT SCIENCES Incertae Sedis Blue Rim Xylotype 3 Specimen. UF 19225-57353 (ﬁg. 7G–7I). Description. Growth ring boundaries indistinct, possibly deﬁned by irregular areas with no vessels alternating with areas with irregular rows of vessels. Wood diffuse-porous with vessels solitary (39%) and in radial multiples of two or three, occasionally four or ﬁve, or rarely clusters (ﬁg. 7G). Solitary vessels oval to circular in outline; perforation plates likely simple, as no scalariform observed. Intervessel pits alternate, averaging 8.7 mm in diameter (0.4 mm), 8.3–9.3 mm. Mean tangential diameter of vessel lumina 66 mm (17 mm), 28–98 mm. Vessel frequency averages 13 mm2 (range: 8–20 mm2). Tyloses not obvious in transverse section, but some vessels obscured in longitudinal section. Parenchyma not distinguishable from ﬁbers in transverse section. Fibers nonseptate, possibly rare septate. Occasional strands of diffuse apotracheal axial parenchyma observed in tangential section—strand length variable from two? to nine? cells. Rays mostly two-seriate, occasionally three-seriate or uniseriate (ﬁg. 7H, 7I). If uniseriate and biseriate in same ray, uniseriate portion usually narrower than biseriate portion. Rays average 152 mm in height (70 mm), 60–337 mm. Ray cellular composition heterocellular, but speciﬁc arrangement could not be determined. Rays per millimeter averages to 15 (range: 12–18). Ray cells often with large open lumens and thin walls (ﬁg. 7I). Storied structure and intercellular canals not observed. Frequent enlarged white cells in rays—possibly idioblasts. Dark brown to black rounded deposits also present in rays. Vulnerability index. This specimen has an estimated V of 5.1. Remarks. Although specimen UF 19225-57353 was found in vertical position within the UF 19225 leaf quarry (estimated to be ∼17 m tall; table 3), poor preservation (most areas are compressed and distorted) did not allow its taxonomic afﬁnities to be determined, because many critical diagnostic features could not be seen. However, it is unlike any other BR wood. This specimen differs from the BR Canellaceae Xylotype because it has vessels in radial multiples, and rays are generally biseriate rather than uniseriate. BR Xylotype 3 does not have the conspicuous axial parenchyma of P. diversiradii. It also differs from E. parviareolatum because it lacks radial canals and rhomboidal prismatic crystals. The characters that could be scored with high conﬁdence (5p, wood diffuse-porous; 22p, intervessel pits alternate; 26p, intervessel pits medium; 41p, mean tangential diameter of vessel lumina 50–100 mm; 47p, ﬁve to 20 vessels per square millimeter; 97p, rays one to three cells wide; 102a, ray height !1 mm; 116p, ≥12 rays per millimeter; 118a, rays not storied) were entered into InsideWood (2004–; Wheeler 2011) with 0 mismatches, and searching only modern wood revealed 93 results in 29 families. Sixteen matches were in the Myrtaceae, whereas 18 results were in the Sapotaceae. When additional characters (2p, growth ring boundaries indistinct or absent; 43a, mean tangential diameter of vessel lumina not ≥200 mm; 76p, axial parenchyma diffuse; 92p, four cells per parenchyma strand; 115p, four to 12 rays per millimeter) were included and modern wood was searched with 0 mismatches, there were eight results: Guiera senegalensis J.F. Gmel. (Combretaceae), Geissois (Cunoniaceae), Tamarindus indica L. (Fabaceae, Caesalpinioideae), Henriettella (Melastomataceae), Eucalyptus dealbata Schauer, Eucalyptus macrorhyncha F. Muell. ex Benth., Syncarpia (Myrtaceae), and Nothofagus subgrp. Brassospora (Nothofagaceae; InsideWood 2004–). These results were reviewed, and none were found to be a good match with the fossil. This wood might not belong to a modern genus, however. The angiosperm megafossil compressions found in the same quarry as this wood specimen were almost exclusively fruits of the extinct sapindalean genus Landeenia (Manchester and Hermsen 2000) and leaﬂets called “Cedrela” schimperi (Meliaceae; MacGinitie 1974). The association of “C.” schimperi leaves with Landeenia fruits and ﬂowers at Kisinger Lakes (MacGinitie 1974) and with the same kind of fruits in Paleocene Eocene Thermal Maximum (∼56 Ma) quarries in the Bighorn Basin (S. L. Wing and S. R. Manchester, personal communication, 2016) suggest that the leaves and fruits represent part of the same extinct plant. Although it cannot be proven because of the lack of a physical connection, the close proximity of these organs around the base of the stem described here leads to the hypothesis that it might represent the wood of Landeenia. If this hypothesis is correct, better-preserved fossils of this wood type may provide additional characters to resolve the systematic position of Landeenia, whose fruit, ﬂower, and pollen characters are consistent with Sapindales but are not a match to any modern genus (Manchester and Hermsen 2000). Comparison of incertae sedis xylotypes. Although the three incertae sedis specimens have some anatomical characters in common, differences include the very thick-walled ﬁbers in BR Xylotype 1. The intervessel pits in BR Xylotype 1 are not well preserved, and the measurements likely have a larger error than normal, but the average pit size (5.8 mm) is smaller than either BR Xylotype 2 (6.8 mm) or BR Xylotype 3 (8.7 mm). Whereas ray width is similar among these specimens, BR Xylotype 1 often has rays where the uniseriate portion equals the multiseriate portion, which is not as common in the other two specimens. In addition, ray frequency is higher in BR Xylotype 1, as compared to the other two incertae sedis specimens. Strands of axial parenchyma were observed in BR Xylotype 3 (UF 19225-57353) and not in the other two unknowns, but that may be due to variations in preservation. BR Xylotype 3 might have idioblasts in the ray cells, which were not observed in either BR Xylotype 1 (UF 19225-54694) or BR Xylotype 2 (UF 18591-33017). All three incertae sedis specimens share multiple characters, including absent to indistinct growth rings, diffuse porosity, similar vessel grouping, simple perforation plates, similar vessel diameters and frequency, ray height, and the presence of black structures in the ray cells. However, many of these characters are common across angiosperm woods. Taphonomy and Size of Specimens Some of the fossilized wood samples were found upright and in situ (e.g., BR Xylotype 3, UF 19225-57353) or with minimal transport. However, others appeared to have traveled a more signiﬁcant distance (e.g., cf. Pinus, UF 19406-61964). An in situ stump (not sampled for wood anatomy due to poor preservation) was 25 cm in diameter with a lateral root trace extending approximately 67 cm from the edge of the trunk. The largest fossilized wood piece observed in the ﬁeld at BR measured ap- This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). ALLEN—RECONSTRUCTING AN EOCENE SITE USING FOSSIL WOOD proximately 33 cm in diameter by 42 cm in length. This specimen (not sampled for wood anatomy) was not in situ and appeared to have rolled down one of the badland ﬁngers; it was resting near a gully at the bottom of the escarpment along with many other pieces of petriﬁed wood. The majority of the wood specimens at BR were prostrate and not in situ. However, some specimens (including UF 18591-33014 through UF 1859133018 from the 1995 ﬁeld season) were upright. Whereas most of the wood samples suggest relatively small trees or shrubs, a tree over 30 cm in diameter is quite large and suggests that larger and/or more mature trees were present on the landscape. Specimens whose full diameter was measured in the ﬁeld were used to estimate tree height (tables 2, 3). Estimated tree height. Complete trees are almost never preserved in the fossil record; stumps and logs are more common. The diameter of these tree fragments can be used to estimate tree height (tables 2, 3). McMahon (1973) and McMahon and Kronauer (1976) used the record trees (largest known individuals) of over 500 North American species to plot height and diameter. They also drew a line representing the buckling height of a hypothetical uniform cylinder with a consistent ratio of E/r, with E representing the elastic modulus and r the mass density. Although McMahon (1973) and McMahon and Kronauer (1976) did have a large data set, using the record specimen of a species is not a reasonable representative of the average size of that taxon. Record specimens are often found in ﬁelds or in areas with optimal growing conditions and little competition. Furthermore, the ratio of E/r is not consistent across species. For these reasons, the values calculated using the work of McMahon (1973) and McMahon and Kronauer (1976) were excluded from the second average calculation in table 3. Rich et al. (1986) used a forest of dicots and palms from a tropical wet forest in Costa Rica to measure diameter at breast height and total height. The relationship between diameter and height was plotted for all dicot species together and for individual palm and dicot species. Rich et al. (1986) found that diameter varies in relation to height, but they also noted that the tallest individuals of a few of the species they examined exceeded the buckling height as determined in McMahon (1973). This provides additional support for being cautious when using McMahon (1973) and McMahon and Kronauer’s (1976) regression to estimate height. 000 Brown et al. (1989) developed regression equations (table 2) using height and diameter measurements from ∼4000 trees from 35 sites in Venezuela, Papua New Guinea, and Puerto Rico. Brown et al. (1989) provided equations for estimating height in both moist and wet life zone conditions (Holdridge 1967). Holdridge’s (1967) life zone deﬁnitions were examined to determine which of the two conditions (moist or wet) were more aligned with the climate conditions inferred for the BR locality in the early Eocene. The tropical wet life zone has a mean annual precipitation of 400–800 cm, and the tropical moist life zone is deﬁned by 200– 400 cm of mean annual precipitation (Holdridge 1967). The subtropical wet life zone has a mean annual precipitation of 200–800 cm, with the subtropical moist life zone mean annual precipitation ranging from 100 to 200 cm (Holdridge 1967). Based on this information and the precipitation estimates from leaf area analysis and the Climate Leaf Analysis Multivariate Program for the leaf fossils at BR (S. E. Allen, unpublished data), the moist life zone is more aligned with the conditions at BR ~49.0 Ma. Therefore, the regression equation developed from plants growing in moist conditions, rather than wet, was used. Niklas (1994) also developed correlations between stem length and diameter. Data from 265 self-supporting tree or shrub species from multiple plant clades were used to create equations to predict the height of fragmented fossil plants (Niklas 1994). A subset of these specimens were classiﬁed as woody, and that equation (table 2) was used to estimate the height of the BR specimens (table 3). Feldpausch et al. (2011) compiled an extensive database of 39,955 diameter and height measurements of trees from 283 tropical sites in 22 countries. Feldpausch et al. (2011) noted numerous factors that impact the height/diameter relationship, including forest type and geographic region. Whereas many of the equations presented by Feldpausch et al. (2011) contained variables that could not be calculated with the limited information from fossil logs, an equation detailed in the appendix of that work removed these coefﬁcients and proved useful for estimating tree height (table 2). The formulas of Brown et al. (1989), Niklas (1994), and Feldpausch et al. (2011) all give height estimates within 2 m of each other for an individual stump or log (table 3). Estimates using the Rich et al. (1986) approach tend to be slightly lower Table 2 Equations Used to Estimate Tree Height from Diameter Reference Formula Variables Notes McMahon 1973; McMahon and Kronauer 1976 log(H) p m(log(D)) 1 log(b) or H p bDm m: slope p 2/3; b: value of H where D p 1 p 25 Rich et al. 1986 log(H) p m(log(D)) 1 log(b) or H p bDm H p exp(1.0710 1 .5688 ln D*) log(H) p 1.59 1 .39(logD) 2 .18(logD)2 ln(H) p b0 1 b1ln(D*) m: slope p .58; b: value of H where D p 1 p 33 ... m and b calculated from line in ﬁg. 5 of McMahon and Kronauer 1976 (p. 456) m and b calculated from line in ﬁg. 1 (p. 243) Moist life zone eq. (5) from table 2 (p. 886) Woody species equation from table 2 (p. 1240) Pantropical equation from table A2 (p. 1100) Brown et al. 1989 Niklas 1994 Feldpausch et al. 2011 Note. ... b0 p 1.2229; b1 p .5320 D p measured diameter in meters; D* p measured diameter in centimeters; H p estimated height in meters. This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). INTERNATIONAL JOURNAL OF PLANT SCIENCES 000 Table 3 Estimated Tree Heights of 10 Blue Rim Fossil Wood Specimens Calculated from the Diameter Estimated height (m) from the ﬁve methods in table 2 Blue Rim specimen FO (7/8/2014) FO (7/8/2014) UF 18591-33014 UF 18591-33015 UF 18591-33016 UF 18591-33017 UF 18591-33018 UF 18591-33022 UF 18591-33023 UF 19225-57353 Diameter (m) A B C D E .25 .33 .46 .31 .2 .55 .25 .2 .31 .23 9.92 11.94 14.90 11.45 8.55 16.78 9.92 8.55 11.45 9.38 14.77 17.35 21.03 16.73 12.98 23.33 14.77 12.98 16.73 14.07 18.14 21.24 25.65 20.50 15.99 28.39 18.14 15.99 20.50 17.31 19.50 22.93 27.42 22.13 16.96 29.96 19.50 16.96 22.13 18.52 18.83 21.82 26.04 21.11 16.72 28.64 18.83 16.72 21.11 18.01 Average (SD) 16.23 19.06 23.01 18.39 14.24 25.42 16.23 14.24 18.39 15.46 (3.97) (4.50) (5.13) (4.38) (3.56) (5.45) (3.97) (3.56) (4.38) (3.81) Average (SD) excluding A 17.81 20.84 25.04 20.12 15.66 27.58 17.81 15.66 20.12 16.98 (2.10) (2.43) (2.77) (2.36) (1.84) (2.92) (2.10) (1.84) (2.36) (2.00) Note. A p McMahon 1973; McMahon and Kronauer 1976; B p Rich et al. 1986; C p Brown et al. 1989; D p Niklas 1994; E p Feldpausch et al. 2011; FO p ﬁeld observation. than those three, with the McMahon (1973) and McMahon and Kronauer (1976) estimates being the shortest (table 3). As previously mentioned, the McMahon (1973) regression is likely biased (it estimates shorter heights using the same diameter) by a data set based on the largest specimens of a species. Furthermore, both the McMahon and Kronauer (1976) and Rich et al. (1986) equations were manually calculated from graphs in the original manuscripts. This likely added a small level of additional error not present in the estimates using Brown et al. (1989), Niklas (1994), and Feldpausch et al. (2011) because those equations were directly provided in the original manuscripts. Field observations suggest that trees were not closely spaced and were generally small, perhaps representing an early successional forest or an environment less favorable to massive trees. However, this assumption does not account for taphonomic effects that may have preferentially preserved smaller woody stems. Alternatively, it could be that wood of the dominant tree type on the landscape, such as Populus, which is well represented in the leaf ﬂora, is not preserved because Populus wood decays rapidly and is less likely to enter the fossil record (Barrett 1995). Overall, based on the fossils recovered, trees on the BR landscape ~49.0 Ma were in the vicinity of 28 m or 92 ft tall (table 3). Discussion Density and Speciﬁc Gravity Wood density and wood speciﬁc gravity are related but are not interchangeable terms (even though they are often treated as the same thing in the literature). Wood density is the mass of the wood per unit volume—this can be measured at any moisture content (Williamson and Wiemann 2010). By contrast, wood speciﬁc gravity is the density of the wood (using ovendry mass) relative to the density of water (Williamson and Wiemann 2010). Wood density can be thought of as a measure of how much of a stem is cell wall versus open space; this is directly connected to plant form and function (Swenson and Enquist 2007). Speciﬁc gravity was estimated for a few of the well- preserved BR woods (six specimens assigned to Edenoxylon parviareolatum). Although speciﬁc gravity estimates could not be calculated for any of the Peltophoroxylon diversiradii specimens, abundant axial parenchyma (and therefore fewer ﬁbers) frequently correlates with a lower-density wood (Zheng and Martínez-Cabrera 2013). Work on modern ﬂoras suggests that leaf size decreases with increasing wood density (Chave et al. 2009). Whereas it is not possible to know for sure whether any of the leaf morphotypes at BR represent the same species as a wood type, E. parviareolatum (estimated speciﬁc gravity 0.58–0.87 using the methods of Wheeler et al. [2007b]) may be aligned with leaves assigned to Rhus nigricans, which are on the smaller side. The speciﬁc gravity estimates from BR (0.58–0.87 using the methods of Wheeler et al. [2007b] and 0.82–1.35 using the methods of Martínez-Cabrera et al. ) seem high. The average speciﬁc gravity of 156 tree species in the United States that comprise over 95% of the total volume of trees is 0.48 (SD p 0.11, range: 0.29–0.80; Miles and Smith 2009). The average speciﬁc gravity is 0.41 (SD p 0.08, range: 0.29–0.68) for the 56 species of gymnosperms and 0.52 (SD p 0.10, range: 0.31–0.80) for the 100 angiosperm species (Miles and Smith 2009). These numbers are much lower than speciﬁc gravity estimates from the BR woods, but the values obtained from the fossils are likely inﬂated due to some of the cells being crushed, lowering the percentage of lumen visible. Tropical lowlands, which are dominated by angiosperms, have more variation in wood density than temperate, high-elevation, or stressful environments, which are often dominated by gymnosperms (Swenson and Enquist 2007; Chave et al. 2009). Taxa with low speciﬁc gravity wood are likely fast-growing, lightdemanding colonizers or pioneers, whereas taxa with high speciﬁc gravity wood are more likely to be shade-tolerant, slowgrowing members of the subcanopy (Wiemann and Williamson 2002; Baker et al. 2004). Species with intermediate speciﬁc gravities are typically mature forest species or emergents. Some studies have found that wood density is positively correlated with climate conditions including mean annual temperature, maximum monthly temperature, and mean annual precipitation (Wiemann and Williamson 2002; Swenson and Enquist This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). ALLEN—RECONSTRUCTING AN EOCENE SITE USING FOSSIL WOOD 2007; Chave et al. 2009). By contrast, mean speciﬁc gravity has been shown to be negatively correlated with soil fertility; sites with higher speciﬁc gravity woods had poorer soils (Muller-Landau 2004). Yet, studies examining the correlation between various environmental conditions and wood speciﬁc gravity have often contradicted each other; these differences might be due in part to how speciﬁc gravity is calculated (Muller-Landau 2004). Vulnerability Index Vulnerability index (V) was also calculated for all BR wood specimens with both average vessel frequency (mm2) and average tangential vessel diameter (mm; Carlquist 1975a, 1975b, 1977; Wheeler 1991). A low V (below 1) suggests that a plant is capable of withstanding freezing or water stress conditions, because narrow vessels can withstand lower negative pressures, and in plants with high vessel frequency, the effect of each embolism on total conductivity is minimized (Carlquist 1977, 2012). For example, desert and arctic shrubs have an average V of 0.08 and 0.10, respectively (Carlquist 1975a, 1977). By contrast, vines and lianas have an average V of 8.22 (Carlquist 1975a, 1977). The BR woods have V values ranging from 1.43 to 10.30 with a median of 3.51 and an average of 4.22 (SD p 2.52, n p 16). Extant tropical trees tend to have high V values, more than 2 or 3 (Wheeler 1991; Burnham and Johnson 2004). No BR specimens have V ! 1, whereas nine specimens have V 1 3. All three of the fossil angiosperm wood types identiﬁed to family have specimens with values falling in the modern tropical range. The nine specimens with V 1 3 include four assigned to E. parviareolatum, two legumes, the specimen of Canellaceae, BR Xylotype 2 (UF 18591-33017), and BR Xylotype 3 (UF 19225-57353). High V values suggest that the BR woods did not experience signiﬁcant water stress (Wheeler 1991). This observation is supported by the local geology. The Bridger Formation in the BR area (lower Bridger) preserves ﬂuvial, deltaic, and ﬂoodplain environments interbedded with lacustrine sandstones representing transgressions of the Laney Member of the Green River Formation (M. E. Smith, personal communication, 2014). However, Wheeler (1991) cautions that V does not account for the amount of active, conducting sapwood, which can vary greatly. Ecology and Climate Even though only seven wood types have been recognized, the BR wood ﬂora is dominated by angiosperms, with only a single Pinaceae sample present. It is suggested that one of the ways angiosperms may have been able to diversify and dominate so many ecosystems, especially in the lowland tropics, is the development of vessels. These structures permit higher maximum conductivity and therefore higher maximum productivity compared to tracheids alone (Chave et al. 2009). Many wood anatomical features, including vessel characters, have been shown to be closely correlated with environmental conditions. Woody species with narrow-diameter vessels (!100 mm), a high vessel frequency (≥40 per square millimeter), and vessels grouped in clusters, features that provide safeguards against embolism, are more common in cool temperate, arctic, tropical montane, alpine, and very dry environments. By contrast, woody taxa 000 found in frost-free lowland tropical forests often have characters that provide less resistance to hydraulic ﬂow, including largediameter vessels (1200 mm), a low vessel frequency (!5 per square millimeter), and simple perforation plates (Baas 1983; Wheeler and Baas 1993; Baas et al. 2004; Ewers et al. 2007; Chave et al. 2009; Carlquist 2012). Within a genus, generally the widest vessels are found in species growing in tropical lowlands, the narrowest in species at high latitudes, and those of intermediate diameter in tropical alpine species (Baas et al. 2004). Many BR specimens are diffuse-porous with narrow (!100-mm) vessels. Baas (2004) noted that diffuse porosity with narrow vessels is characteristic of evergreen tropical montane, evergreen temperate, and many deciduous temperate trees and shrubs. Plant size, in addition to environmental factors, also contributes to variations in vessel diameter. For example, very wide vessels are rare in small trees and shrubs (Wheeler and Baas 1993; Baas et al. 2004; Olson et al. 2014). Many of the BR wood specimens are small hand samples, possibly from plants with a smaller stature, but they could also represent branch rather than trunk wood. This factor could account for the smaller vessel diameters in the BR woods. Alternatively, it has been shown that trees growing in more stressful environments tend to have significantly smaller vessel diameters than trees growing in less stressful sites (Fichtler and Worbes 2012). Shrubs and branches often have a higher vessel frequency than mature tree trunk wood (Wheeler and Baas 1993). In general, the BR woods have an intermediate vessel frequency (averages range between seven and 40 per square millimeter from 17 specimens), which may indicate that they represent mostly branch or shrub wood of tropical lowland species. However, multiple specimens were measured at over 1 ft in diameter, clearly indicating that at least some were trees. One of these specimens (UF 18591-33023, 31 cm) was assigned to E. parviareolatum, but most of the other large wood samples documented in the ﬁeld were too poorly preserved to describe or identify. One exception is BR Xylotype 2 (UF 18591-33017), which had a stem diameter of over 55 cm and was described, but the taxonomic afﬁnity could not be determined. Conspicuous axial parenchyma patterns including aliform, conﬂuent, and bands more than three cells wide, such as those observed in P. diversiradii, are generally found in tropical, nondrought-tolerant taxa (Baas 1983; Wheeler et al. 2007a; Zheng and Martínez-Cabrera 2013; Morris et al. 2016). Ample axial parenchyma has been correlated with higher conduction capacity, in part because axial parenchyma may help prevent cavitation or repair vessels when embolisms occur (Zheng and MartínezCabrera 2013; Morris 2016; Morris et al. 2016). Other wood characters also provide ecological and climate signals. For example, regular growth rings are a consequence of periodic dormancy, a strategy many trees use to withstand conditions including drought, cold, or photoperiodic events, and are therefore less common in plants from wet tropical lowlands (Carlquist 2012). Even though only seven wood types were recovered from BR, most have indistinct growth rings, indicating likely tropical conditions (nonseasonal, equable warm, and wet; Wheeler and Baas 1993). There are no BR woods with ring porosity or helical thickenings, characters more common in temperate environments. Randomly arranged vessels is the most frequently encountered pattern both globally and in the BR woods; This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). 000 INTERNATIONAL JOURNAL OF PLANT SCIENCES clusters (rare at BR) are less common in the tropics than in other environments (Wheeler and Baas 1993). Crystals (observed in the species of Anacardiaceae and Fabaceae) and silica bodies (possible in P. diversiradii and a few of the unidentiﬁed specimens) are also more common in the tropics than in temperate zones (Wheeler et al. 2007a). The taxonomically identiﬁed woods also provide clues about the climate and ecological conditions at BR in the latest Early Eocene. For example, legumes have a worldwide distribution, but the Caesalpinioideae are mainly tropical and subtropical woody trees, shrubs, or lianas with species diversity centered in Southeast Asia, Africa, and South America (Klitgård and Lewis 2010; LPWG 2013a). Edenoxylon parviareolatum is thought to be subtropical, in part because of the taxonomic afﬁnities of the other woods recovered at BSR, including Palmoxylon macginitiei Tidwell, a species of palm (Boonchai and Manchester 2012). The nearest living relatives of the BR Canellaceae Xylotype are from subtropical and tropical environments. However, both extant and fossil taxa of Canellaceae, including the BR specimen, have scalariform perforation plates. Scalariform perforation plates provide more resistance to hydraulic ﬂow, as the end walls of vessel elements account for ∼50% of the total resistance in a vessel; they are thought to help localize any air embolisms that might occur (Baas 1983; Baas et al. 2004; Chave et al. 2009; Carlquist 2012). The percentage of species with scalariform perforation plates is higher in cool temperate to arctic (23%–53%) and tropical high montane (15%–33%) environments but lower in tropical lowland and arid environments (0%–8%; Wheeler and Baas 1993). The taxonomic composition of the BR ﬂora, in conjunction with temperature estimates from leaf physiognomic data (S. E. Allen, unpublished data), is more suggestive of tropical montane conditions rather than a cool temperate or arctic environment. The specimen of Pinaceae is not very informative, as pines are distributed from subtropical to temperate areas throughout the Northern Hemisphere and many species are tolerant of extreme conditions including high elevation and/or high latitudes (Farjon and Styles 1997). Overall, the wood characters observed at BR, including diffuse porosity, the absence of distinct growth rings, and rarity of plants with scalariform perforation plates, suggest warm to tropical conditions with limited seasonality in southwestern Wyoming in the latest Early Eocene. This agrees with paleoclimate estimates from leaf physiognomic methods (S. E. Allen, unpublished data) and the presence of frost-intolerant taxa such as Phoenix windmillis (Allen 2015a). Comparison with Other Western Interior North American Eocene Wood Floras Available data from the limited number of Eocene wood localities studied from the Western Interior of North America suggest that there was variation in taxonomy and wood anatomy across the region (Wheeler and Michalski 2003). Most early and middle Eocene dicotyledonous woods in the United States lack growth rings (Wheeler 2001), which is similar to what was observed at BR. The BR assemblage is distinctive in some respects. For example, most middle Eocene wood assemblages documented from the United States have a plane tree relative, Plataninium or Platanoxylon, and Quercinium (cf. evergreen oaks; Wheeler et al. 1978; Wheeler 2001; Gregory et al. 2009). Yet, neither of these taxa is represented at BR. Wheeler and Michalski (2003) documented siliciﬁed woods from Paleocene and Eocene strata in the Denver Basin of Colorado. Among those, the Eocene sites were dominated by wood of Paraphyllanthoxylon (probable Lauraceae; Wheeler and Michalski 2003). The characters of the Denver Basin woods, including diffuse porosity (which is shared with the BR woods), are most similar to the wood anatomy of taxa growing in the frost-free lowland tropics today (Wheeler and Michalski 2003). However, there is no taxonomic overlap between the identiﬁed BR and Denver Basin woods. Although leaf ﬂoras from the Eocene of Yellowstone National Park have more angiosperms than conifers, the woods preserve a higher percentage of conifer taxa (Knowlton 1899; Wheeler et al. 1977; Fritz and Fisk 1978). The BR leaf ﬂora is similarly dominated by angiosperms, with a few ferns and no gymnosperms (Allen 2015b); however, the wood assemblage differs from that of Yellowstone in having only one coniferous specimen. More than 30 dicotyledonous wood types have been found from the Yellowstone localities of the Lamar River Formation (Wheeler et al. 1977). Families represented by the wood include Betulaceae, Lauraceae, Magnoliaceae, and Platanaceae (Wheeler et al. 1977, 1978). Growth rings in the Yellowstone woods range from indistinct to distinct (Wheeler et al. 1977, 1978), whereas none of the BR specimens have well-preserved distinct growth rings (however, it is difﬁcult to determine in the Pinaceae specimen). The presence of species in the Yellowstone ﬂora with well-deﬁned growth rings may be due to either its slightly younger age (and therefore further from the peak of early Eocene warming) or the higher elevation of Yellowstone contributing to more seasonality than BR. Chadwick and Yamamoto (1984) examined 119 wood specimens from the Specimen Creek area of the Gallatin Petriﬁed Forest, Yellowstone National Park, Montana. They found representatives of three gymnosperm families, including Pinaceae, and 15 genera in 13 angiosperm families, including wood they assigned to Salix (Salicaceae; Yamamoto and Chadwick 1982; Chadwick and Yamamoto 1984). The potential presence of wood assignable to Salicaceae is notable because there are deﬁnitive leaves and fruits of Populus at BR. Populus wood should be easily recognizable by its medium to large polygonally shaped alternate intervessel pits and exclusively uniseriate rays with all ray cells procumbent (Metcalfe and Chalk 1950; InsideWood 2004–), but this combination of characters is not present in any of the BR woods. Given the abundance of Populus leaves and fruits at BR and other Eocene sites in Wyoming, Utah, and Colorado (MacGinitie 1969, 1974), it might seem surprising that wood of Salicaceae has not been recovered. However, this absence might be due to taphonomic bias and the ease with which Populus wood degrades due to its low speciﬁc gravity and extractive content (Barrett 1995; E. Wheeler, personal communication, 2016). Furthermore, Elisabeth Wheeler (personal communication, 2016) has examined Yamamoto and Chadwick’s original slides from Specimen Creek, in addition to woods from other areas of Yellowstone, and has not found any evidence of Salix or Populus. The middle Eocene (∼44 Ma) wood assemblage from the Clarno Formation in Oregon is extremely diverse, with 66 genera. Representatives of Anacardiaceae, Fabaceae, and Pinaceae are present in both the BR and the Clarno ﬂoras, but there are This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). ALLEN—RECONSTRUCTING AN EOCENE SITE USING FOSSIL WOOD no shared species (Wheeler and Manchester 2002). The latest Eocene Florissant Fossil Beds National Monument (∼34 Ma) in central Colorado preserves ring-porous woods belonging to the Ulmaceae, Sapindaceae, and Fabaceae (Wheeler 2001). This suggests distinct seasonality in the later part of the Eocene as compared to the early Eocene. 000 probable leaves (R. nigricans) assigned to Anacardiaceae in the BR ﬂora (MacGinitie 1969; Manchester 1994; Collinson et al. 2012). Multiple organs assigned to Anacardiaceae provide the potential for whole-plant reconstruction. Concluding Remarks Comparison with Other Blue Rim Floral Elements The wood assemblage, comprised of seven xylotypes from over 30 specimens, along with the rest of the macroﬂora from BR, is dominated by angiosperms. Some broad taxonomic overlap between the woods and other elements of the macroﬂora and microﬂora is present. The occurrence of a wood assignable to Pinaceae is signiﬁcant, as no other conifer macrofossils have been recovered from BR. Coniferous megafossils are also very rare in the adjacent Green River Formation (MacGinitie 1969; Wilf 2000). Yet, pollen assignable to Pinaceae (including Pinus) occurs at multiple stratigraphic horizons at BR, indicating that the family was present in the regional ﬂora with possible rare local trees based on the presence of the pinaceous wood specimen. However, the cf. Pinus specimen was found at the bottom of the BR escarpment in a gully, with no evidence that it originated nearby. Furthermore, both the hand specimen and microscopic views suggest that the cf. Pinus fossil wood was transported from its original growing location and was subjected to compression and distortion during or prior to the permineralization process. No other macrofossils of Canellaceae have been recovered at BR to date. However, representatives of Canellaceae might be overlooked in the leaf assemblage because they have evergreen, entire-margined, pinnately veined leaves lacking special characters, which makes them difﬁcult to distinguish from various other entire-margined dicotyledonous taxa. By contrast, the ﬂowers are rather distinctive, with three sepals, ﬁve to 12 petals, two to six connate carpels, and six to many stamens fused into a tube (Judd et al. 2008). This combination of characters has not been observed in any of the BR ﬂower morphotypes. In addition to the wood, a few legume leaﬂets have been recovered at BR. The leaves of this large family are often pinnately compound with entire leaﬂets and pulvini. There are many (often poorly preserved) laminae with entire margins at BR. However, only one specimen had an obvious pulvini (which was matched to a few other specimens), allowing for a conﬁdent assignment to Fabaceae. Additional specimens could easily represent leaves or leaﬂets of Fabaceae or other families with entire margins. Legume ﬂowers have also not been recognized from BR, even though there are still multiple unidentiﬁed ﬂower morphotypes. Legumes—both leaves and fruits—are well represented in the nearby Green River ﬂora (MacGinitie 1969; Johnson and Plumb 1995; Grande 2013), and fruits have also been recovered at the Kisinger Lakes site in northwestern Wyoming. Along with the anacardiaceous wood, there are also fruits of Pentoperculum minimus (Reid and Chandler) Manchester and This survey found only seven different wood types from the BR localities. Hence, BR preserves a low-diversity wood ﬂora dominated by Edenoxylon parviareolatum (Anacardiaceae). Specimens assigned to Fabaceae, Canellaceae, and Pinaceae are also present, along with a few unidentiﬁable but distinct xylotypes. Other elements of the same fossil ﬂora, including leaves, reproductive structures, and dispersed pollen and spores, suggest that overall richness is much higher than is represented by the wood. Stem diameter measurements obtained from some specimens yield tree height estimates of ∼16–28 m. Self-supporting taxa would have been essential for climbers such as Iodes, Vitis, and Lygodium present in the same ﬂora (Manchester and Zavada 1987; Allen et al. 2015). The angiosperm woods are diffuseporous with absent or indistinct growth rings and rare scalariform perforation plates. These characters suggest that climate conditions were warm with limited seasonality, which agrees with estimates from leaf physiognomic methods (S. E. Allen, unpublished data) and the presence of frost-intolerant taxa such as Phoenix windmillis (Allen 2015a). Acknowledgments Thank you to Steven R. Manchester, my advisor, for his valuable comments and support. Work on this article was also assisted by discussions and comments from Nareerat Boonchai, Nathan Jud, Chris Nelson, and Elisabeth Wheeler. Hongshan Wang’s curatorial assistance at the FLMNH is greatly appreciated. Recent wood slides were prepared by Chris Nelson. My participation in an NSF-funded short course on Plant Anatomy (microMORPH) at the Arnold Arboretum of Harvard University in the summer of 2015 (instructors: Pieter Baas, Pam Diggle, William [Ned] Friedman, Peter Gasson, Elisabeth Wheeler) greatly facilitated my knowledge of wood anatomy and allowed me to complete this aspect of my dissertation project. Thank you also to those who helped me with ﬁeldwork in 2012 and 2014: Jim Barkley, Sahale Casebolt, Ellen Currano, Xiaoyan Liu, Terry Lott, Steve Manchester, Paul Murphey, Mike Smith, and Greg Stull. Thanks are also extended to two anonymous reviewers who provided valuable suggestions that greatly improved the article. Funding was provided by the NSF (DEB-1404895), the Evolving Earth Foundation, the Paleontological Society, and the UF Department of Biology. Finally, thank you to my friends and family for their continued support. Literature Cited Allen SE 2015a Fossil palm ﬂowers from the Eocene of the Rocky Mountain region with afﬁnities to Phoenix L. (Arecaceae: Coryphoideae). Int J Plant Sci 176:586–596. ——— 2015b Leaf fossils from the Blue Rim ﬂora of the Eocene Bridger Formation, southwestern Wyoming. Paper presented at Botany 2015, Edmonton, Alberta, July 25–29. This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). 000 INTERNATIONAL JOURNAL OF PLANT SCIENCES Allen SE, GW Stull, SR Manchester 2015 Icacinaceae from the Eocene of western North America. Am J Bot 102:725–744. Awasthi N 1992 Indian fossil legumes. Pages 225–250 in PS Herendeen, DL Dilcher, eds. Advances in legume systematics. Pt 4. The fossil record. Royal Botanic Gardens, Kew. Baas P 1983 Ecological patterns of xylem anatomy. Pages 327–352 in TJ Givnish, ed. On the economy of plant form and function. Proceedings of the Sixth Maria Moors Cabot Symposium, Evolutionary Constraints on Primary Productivity: Adaptive Patterns of Energy Capture in Plants. Cambridge University Press, Cambridge. Baas P, FW Ewers, SD Davis, EA Wheeler 2004 Evolution of xylem physiology. Pages 273–295 in AR Hemsley, I Poole, eds. The evolution of plant physiology. Elsevier, London. Baker TR, OL Phillips, Y Malhi, S Almeida, L Arroyo, A Di Fiore, T Erwin, et al 2004 Variation in wood density determines spatial patterns in Amazonian forest biomass. Glob Change Biol 10:545–562. Bande MB, U Prakash 1980 Fossil woods from the Tertiary of West Bengal, India. Geophytology 10:146–157. Barajas-Morales J 1987 Wood speciﬁc gravity in species from two tropical forests in Mexico. IAWA Bull 8:143–148. Baretta-Kuipers T 1981 Wood anatomy of Leguminosae: its relevance to taxonomy. Pages 677–705 in RM Polhill, PH Raven, eds. Advances in legume systematics. Pt 2. Royal Botanic Gardens, Kew. Barrett JW, ed 1995 Regional silviculture of the United States. Wiley, New York. Boonchai N 2012 Systematic afﬁnities and paleoenvironment of Eocene petriﬁed woods from southwestern Wyoming, USA. PhD diss. Jilin University, Changchun, China. Boonchai N, SR Manchester 2012 Systematic afﬁnities of early Eocene petriﬁed woods from Big Sandy Reservoir, southwestern Wyoming. Int J Plant Sci 173:209–227. Brett DW 1966 Fossil wood of Anacardiaceae from the British Eocene. Paleontology 9:360–364. Brown S, AJR Gillespie, AE Lugo 1989 Biomass estimation methods for tropical forests with applications to forest inventory data. For Sci 35:881–902. Bruneau A, M Mercure, GP Lewis, PS Herendeen 2008 Phylogenetic patterns and diversiﬁcation in the Caesalpinioid legumes. Botany 86:697–718. Burnham RJ, KR Johnson 2004 South American paleobotany and the origins of Neotropical rainforests. Phil Trans R Soc Lond B 359: 1595–1610. Carlquist S 1975a Ecological strategies of xylem evolution. University of California Press, Berkeley. ——— 1975b Wood anatomy of Onagraceae, with notes on alternative modes of photosynthate movement in dicotyledon woods. Ann Mo Bot Gard 62:386–424. ——— 1977 Ecological factors in wood evolution: a ﬂoristic approach. Am J Bot 64:887–896. ——— 2012 How wood evolves: a new synthesis. Botany 90:901–940. Chadwick A, T Yamamoto 1984 A paleoecological analysis of the petriﬁed trees in the Specimen Creek area of Yellowstone National Park, Montana, U.S.A. Palaeogeogr Palaeoclim Palaeoecol 45:39–48. Chattaway M 1951 The development of horizontal canals in rays. Aust J Biol Sci 4:1–11. Chave J, D Coomes, S Jansen, SL Lewis, NG Swenson, AE Zanne 2009 Towards a worldwide wood economics spectrum. Ecol Lett 12:351–366. Collinson ME, SR Manchester, V Wilde 2012 Fossil fruits and seeds of the middle Eocene Messel biota, Germany. Abh Senckenberg Ges Naturforsch 570:1–249. Conard HS 1930 A Pityoxylon from Yellowstone National Park. Am J Bot 17:547–553. Espinoza de Pernía N, JL Melandri 2006 Wood anatomy of the tribe Caesalpinieae (Leguminosae, Caesalpinioideae) in Venezuela. IAWA J 27:99–114. Ewers FW, JM Ewers, AL Jacobsen, J López-Portillo 2007 Vessel redundancy: modeling safety in numbers. IAWA J 28:373–388. Farjon A, BT Styles 1997 Pinus (Pinaceae). Flora Neotrop 75:1–291. Feldpausch TR, L Banin, OL Phillips, TR Baker, SL Lewis, CA Quesada, K Affum-Baffoe, et al 2011 Height-diameter allometry of tropical forest trees. Biogeosciences 8:1081–1106. Fichtler E, M Worbes 2012 Wood anatomical variables in tropical trees and their relation to site conditions and individual tree morphology. IAWA J 33:119–140. Fritz WJ, LH Fisk 1978 Eocene petriﬁed woods from one unit of the Amethyst Mountain “fossil forest.” Northwest Geol 7:10–19. Gasson P, C Trafford, B Matthews 2003 Wood anatomy of Caesalpinioideae. Pages 63–93 in BB Klitgaard, A Bruneau, eds. Advances in legume systematics. Pt 10. Royal Botanic Gardens, Kew. Gottwald H 1992 Hölzer aus marinen sanden des oberen Eozän vos Helmstedt (Niedersachsen). Palaeontogr Abt B 225:27–103. Grande L 2013 The lost world of Fossil Lake: snapshots from deep time. University of Chicago Press, Chicago. Gregory M, I Poole, EA Wheeler 2009 Fossil dicot wood names: an annotated list with full bibliography. IAWA J 6(suppl.):1–220. Haston E, JE Richardson, PF Stevens, MW Chase, DJ Harris 2009 The linear angiosperm phylogeny group (LAPG) III: a linear sequence of the families in APG III. Bot J Linn Soc 161:128–131. Heinz I 2004 Systematische Erfassung und Dokumentation der mikroanatomischen Merkmale der Nadelhölzer aus der Klasse der Pinatae. PhD diss. Technische Universität München, Munich. Herendeen PS 2000 Structural evolution in the Caesalpinioideae (Leguminosae). Pages 45–64 in PS Herendeen, A Bruneau, eds. Advances in legume systematics. Pt 9. Royal Botanic Gardens, Kew. Herendeen PS, A Bruneau, GP Lewis 2003 Phylogenetic relationships in Caesalpinioid legumes: a preliminary analysis based on morphological and molecular data. Pages 37–62 in BB Klitgaard, A Bruneau, eds. Advances in legume systematics. Pt 10. Royal Botanic Gardens, Kew. Hilton J, JB Riding, GW Rothwell 2016 Age and identity of the oldest pine fossils. Geol Forum 44:e400–e401. Holdridge LR 1967 Life zone ecology. Tropical Science Center, San Jose, Costa Rica. Hu YS, FH Wang 1984 Anatomical studies of Cathaya (Pinaceae). Am J Bot 71:727–735. IAWA (International Association of Wood Anatomists) Committee 1989 IAWA list of microscopic features for hardwood identiﬁcation. IAWA Bull 10:219–332. ——— 2004 IAWA list of microscopic features for softwood identiﬁcation. IAWA J 25:1–70. InsideWood 2004– The InsideWood Database. http://insidewood.lib .ncsu.edu. Jagels R, BA LePage, M Jiang 2001 Deﬁnitive identiﬁcation of Larix (Pinaceae) wood based on anatomy from the middle Eocene Axel Heiberg Island, Canadian High Arctic. IAWA J 22:73–83. Johnson KR, C Plumb 1995 Common plant fossils from the Green River Formation at Douglas Pass, Colorado, and Bonanza, Utah. Pages 121–130 in WR Averett, ed. The Green River Formation in the Piceance Creek and eastern Uinta Basins. Grand Junction Geological Society, Grand Junction, CO. Judd WS, CS Campbell, EA Kellogg, PF Stevens, MJ Donoghue 2008 Plant systematics: a phylogenetic approach. Sinauer, Sunderland, MA. Kistner FB 1973 Stratigraphy of the Bridger Formation in the Big Island– Blue Rim area, Sweetwater County, Wyoming. MSc diss. University of Wyoming, Laramie. Klitgård BB, GP Lewis 2010 Neotropical Leguminosae (Caesalpinioideae). In W Milliken, B Klitgård, A Baracat, eds. Neotropikey: interactive key and information resources for ﬂowering plants of the Neotropics. http://www.kew.org/science/tropamerica/neotropikey/families /Leguminosae_(Caesalpinioideae).htm. Knowlton FH 1899 Fossil ﬂora of the Yellowstone National Park. Government Printing Ofﬁce, Washington, DC. This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). ALLEN—RECONSTRUCTING AN EOCENE SITE USING FOSSIL WOOD Krishtalka L, RM West, CC Black, MR Dawson, JJ Flynn, WD Turnbull, RK Stucky, et al 1987 Eocene (Wasatchian through Duchesnean) biochronology of North America. Pages 77–117 in MO Woodburne, ed. Cenozoic mammals of North America. University of California Press, Berkeley. Kruse HO 1954 Some Eocene dicotyledonous woods from Eden Valley, Wyoming. Ohio J Sci 54:243–268. Lewis G, B Schrire, B Mackinder, M Lock 2005– Legumes of the World Online. http://www.kew.org/science-conservation/research-data/resources /legumes-of-the-world. LPWG (Legume Phylogeny Working Group) 2013a Legume phylogeny and classiﬁcation in the 21st century: progress, prospects and lessons for other species-rich clades. Taxon 62:217–248. ——— 2013b Towards a new classiﬁcation system for legumes: progress report from the International Legume Conference. S Afr J Bot 89:3–9. ——— 2017 A new subfamily classiﬁcation of the Leguminosae based on a taxonomically comprehensive phylogeny. Taxon 66:44–77. Mabberley DJ 2008 Mabberley’s plant-book. Cambridge University Press, New York. MacGinitie HD 1969 The Eocene Green River ﬂora of northwestern Colorado and northeastern Utah. University of California Press, Berkeley. ——— 1974 An early middle Eocene ﬂora from the YellowstoneAbsaroka Volcanic Province, northwestern Wind River Basin, Wyoming. University of California Publications in Geological Science, Berkeley. Manchester SR 1994 Fruits and seeds of the middle Eocene Nut Beds ﬂora, Clarno Formation, Oregon. Palaeontogr Am 58:1–205. Manchester SR, EJ Hermsen 2000 Flowers, fruits, seeds, and pollen of Landeenia gen. nov., an extinct sapindalean genus from the Eocene of Wyoming. Am J Bot 87:1909–1914. Manchester SR, WS Judd, B Handley 2006 Foliage and fruits of early poplars (Salicaceae: Populus) from the Eocene of Utah, Colorado, and Wyoming. Int J Plant Sci 167:897–908. Manchester SR, EA Wheeler 2006 Siliciﬁed woods from the early middle Eocene Bridger Formation in southwestern Wyoming, USA. Paper presented at Botany 2016, Chico, CA, July 28–August 2. Manchester SR, MS Zavada 1987 Lygodium foliage with intact sorophores from the Eocene of Wyoming. Bot Gaz 148:392–399. Martínez-Cabrera HI, E Estrada-Ruiz, C Castañeda-Posadas, D Woodcock 2012 Wood speciﬁc gravity estimation based on wood anatomical traits: inference of key ecological characteristics in fossil assemblages. Rev Palaeobot Palynol 187:1–10. McMahon T 1973 Size and shape in biology. Science 179:1201–1204. McMahon TA, RE Kronauer 1976 Tree structures: deducing the principle of mechanical design. J Theor Biol 59:443–466. Melandri JL, N Espinoza de Pernía 2009 Wood anatomy of tribe Detarieae and comparison with tribe Caesalpinieae (Leguminosae, Caesalpinioideae) in Venezuela. Rev Biol Trop 57:303–319. Metcalfe CR, L Chalk 1950 Anatomy of the dicotyledons. Oxford University Press, London. Miles PD, WB Smith 2009 Speciﬁc gravity and other properties of wood and bark for 156 tree species found in North America. Research Note NRS-38. US Department of Agriculture, Forest Service, Newtown Square, PA. Morris H 2016 The structure and function of ray and axial parenchyma in seed plants. PhD diss. Ulm University, Ulm, Germany. Morris H, L Plavcová, P Cvecko, E Fichtler, MAF Gillingham, HI Martínez-Cabrera, DJ McGlinn, et al 2016 A global analysis of parenchyma tissue fractions in secondary xylem of seed plants. New Phytol 209:1553–1565. Müller S, K Salomo, J Salazar, J Naumann, MA Jaramillo, C Neinhuis, TS Feild, S Wanke 2015 Intercontinental long-distance dispersal of Canellaceae from the New to the Old World revealed by a nuclear single copy gene and chloroplast loci. Mol Phylogenet Evol 84:205–219. 000 Muller-Landau HC 2004 Interspeciﬁc and inter-site variation in wood speciﬁc gravity of tropical trees. Biotropica 36:20–32. Müller-Stoll WR, E Mädel 1967 Die fossilen Leguminosen-Hölzer: eine Revision der mit Leguminosen verglichenen fossilen Hölzer und Beschreibungen älterer und neuer Arten. Palaeontogr Abt B 119: 95– 174. Murphey PC, E Evanoff 2011 Paleontology and stratigraphy of the middle Eocene Bridger Formation, southern Green River Basin, Wyoming. Proceedings of the 9th Conference on Fossil Resources, Kemmerer, WY. Murphey PC, LL Torick, ES Bray, R Chandler, E Evanoff 2001 Taphonomy, fauna, and depositional environment of the Omomys Quarry, an unusual accumulation from the Bridger Formation (middle Eocene) of southwestern Wyoming (USA). Pages 361–402 in GF Gunnell, ed. Eocene biodiversity: unusual occurrences and rarely sampled habitats. Kluwer, New York. Niklas KJ 1994 Predicting the height of fossil plant remains: an allometric approach to an old problem. Am J Bot 81:1235–1242. Ogata K, T Fujii, H Abe, P Baas 2008 Identiﬁcation of the timbers of Southeast Asia and the Western Paciﬁc. Kaiseisha, Tsukuba, Japan. Olson ME, T Anfodillo, JA Rosell, G Petit, A Crivellaro, S Isnard, C León-Gómez, LO Avarado-Cárdenas, M Castorena 2014 Universal hydraulics of the ﬂowering plants: vessel diameter scales with stem length across angiosperm lineages, habits and climates. Ecol Lett 17: 988–997. Pell SK 2004 Molecular systematics of the cashew family (Anacardiaceae). PhD diss. Louisiana State University and Agricultural and Mechanical College, Baton Rouge. Phillips EWJ 1941 The identiﬁcation of coniferous woods by their microscopic structure. Bot J Linn Soc 52:259–320. Prakash U 1975 Fossil woods from the Lower Siwalik beds of Himachal Pradesh, India. Palaeobotanist 22:192–210. Quirk JT, RB Miller 1985 Vestured pits in the tribe Cassieae Bronn (Leguminosae). IAWA Bull 6:200–212. Ramos RS, M Brea, R Pardo 2014 A new fossil wood of Peltophoroxylon (Leguminosae: Caesalpinioideae) from the El Palmar Formation (late Pleistocene), Entre Ríos, Argentina. IAWA J 35:199–212. Rasband WS 1997– ImageJ. US National Institutes of Health, Bethesda, MA. http://imagej.nih.gov/ij. Rich PM, K Helenurm, D Kearns, SR Morse, MW Palmer, L Short 1986 Height and stem diameter relationships for dicotyledonous trees and arborescent palms of Costa Rican tropical wet forest. Bull Torrey Bot Club 113:241–246. Robinson P, GF Gunnell, SL Walsh, WC Clyde, JE Storer, RK Stucky, DJ Froehlich, I Ferrusquia-Villafranca, MC McKenna 2004 Wasatchian through Duchesnean biochronology. Pages 106–155 in MO Woodburne, ed. Late Cretaceous and Cenozoic mammals of North America. Columbia University Press, New York. Salazar J 2006 Systematics of Neotropical Canellaceae. PhD diss. Cornell University, Ithaca, NY. Salazar J, KC Nixon 2008 New discoveries in the Canellaceae in the Antilles: how phylogeny can support taxonomy. Bot Rev 74:103– 111. Swenson NG, BJ Enquist 2007 Ecological and evolutionary determinants of a key plant functional trait: wood density and its communitywide variation across latitude and elevation. Am J Bot 94:451– 459. Tarelkin Y, C Delvaux, M De Ridder, T El Berkani, D Cannière, H Beeckman 2016 Growth-ring distinctness and boundary anatomy variability in tropical trees. IAWA J 37:275–294. Terrazas T 1994 Wood anatomy of the Anacardiaceae: ecological and phylogenetic interpretation. PhD diss. University of North Carolina at Chapel Hill, Chapel Hill. Terrazas T, T Wendt 1995 Systematic wood anatomy of the genus Tapiria Aublet (Anacardiaceae)—a numerical approach. Brittonia 47:109–129. This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). 000 INTERNATIONAL JOURNAL OF PLANT SCIENCES Wheeler EA 1991 Paleocene dicotyledonous trees from Big Bend National Park, Texas: variability in wood types common in the late Cretaceous and early Tertiary, and ecological inferences. Am J Bot 78: 658–671. ——— 2001 Fossil dicotyledonous woods from Florissant Fossil Beds National Monument, Colorado. Pages 187–203 in E Evanoff, KM Gregory-Wodzicki, KR Johnson, eds. Fossil ﬂora and stratigraphy of the Florissant Formation, Colorado. Vol 1. Denver Museum of Nature and Science, Denver. ——— 2011 InsideWood—a web resource for hardwood anatomy. IAWA J 32:199–211. Wheeler EA, P Baas 1992 Fossil wood of the Leguminosae: a case study in xylem evolution and ecological anatomy. Pages 281–301 in PS Herendeen, DL Dilcher, eds. Advances in legume systematics. Pt 4. The fossil record. Royal Botanic Gardens, Kew. ——— 1993 The potentials and limitations of dicotyledonous wood anatomy for climatic reconstructions. Paleobiology 19:487–498. Wheeler EA, P Baas, S Rodgers 2007a Variations in dicot wood anatomy: a global analysis based on the InsideWood database. IAWA J 28:229–258. Wheeler EA, SR Manchester 2002 Woods of the middle Eocene Nut Beds ﬂora, Clarno Formation, Oregon, USA. IAWA J 3(suppl.):1–188. Wheeler EA, TC Michalski 2003 Paleocene and early Eocene woods of the Denver Basin, Colorado. Rocky Mt Geol 38:29–43. Wheeler EA, RA Scott, ES Barghoorn 1977 Fossil dicotyledonous woods from Yellowstone National Park. J Arnold Arbor 58:280–306. ——— 1978 Fossil dicotyledonous woods from Yellowstone National Park. II. J Arnold Arbor 59:1–31. Wheeler EA, MC Wiemann, JG Fleagle 2007b Woods from the Miocene Bakate Formation, Ethiopia: anatomical characteristics, estimates of original speciﬁc gravity and ecological inferences. Rev Palaeobot Palynol 146:193–207. Wiemann MC, GB Williamson 2002 Geographic variation in wood speciﬁc gravity: effects of latitude, temperature, and precipitation. Wood Fiber Sci 34:96–107. Wilf P 2000 Late Paleocene–early Eocene climate changes in southwestern Wyoming: paleobotanical analysis. Geol Soc Am Bull 112: 292–307. Williamson GB, M Wiemann 2010 Measuring wood speciﬁc gravity . . . correctly. Am J Bot 97:519–524. Wilson TK 1960 The comparative morphology of the Canellaceae. I. Synopsis of genera and wood anatomy. Trop Woods 112:1–27. Wu H, Z Hu 1997 Comparative anatomy of resin ducts of the Pinaceae. Trees 11:135–143. Yamamoto T, A Chadwick 1982 Identiﬁcation of fossil wood from the Specimen Creek area of the Gallatin Petriﬁed Forest, Yellowstone National Park, Montana, U.S.A. II. Angiosperms. J Saniku Gakuin Jr Coll 11:49–66. Zheng J, HI Martínez-Cabrera 2013 Wood anatomical correlates with theoretical conductivity and wood density across China: evolutionary evidence of the functional differentiation of axial and radial parenchyma. Ann Bot 112:927–935. This content downloaded from 130.064.011.153 on October 28, 2017 14:04:09 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).