Late Paleocene woods from Cherokee Ranch, Colorado, U.S.A.

Little is known about the wood anatomy of Paleocene angiosperm trees, with fewer than 90 wood types described worldwide, and fewer than 25 reported for North America (InsideWood, 2004-onwards; Wheeler, 2011). Therefore, any new locality for Paleocene angiosperm wood is important for shedding light on past plant diversity and vegetation types. The wood anatomical traits of the logs found at Cherokee Ranch, Douglas County, Colorado, provide insights into the environment of the Paleocene of the southern Denver Basin. Mustoe and Viney (2017) described the mineralogy and fossilization process of the Cherokee Ranch woods. This study complements the considerable amount of previous work on the paleontology of the Denver Basin, done mostly for leaves and pollen (summaries in Johnson et al., 2003; Raynolds and Johnson, 2003).

Almost all angiosperm trees have vessel elements for water conduction, fibers for support, parenchyma for storage, radial transport (rays), and wound response. Variations in the proportions, sizes, arrangements, and connections between these cells are used in wood identification (Wheeler and Baas, 1998). For fossil wood, it can be difficult to be more specific than determining the family to which a wood sample belongs. Sometimes this is because some genera within a family overlap in their wood anatomical features or because features important for distinguishing genera may not be preserved in the fossil. Regardless of the affinities of the fossils, wood anatomical features, especially those related to water conduction (vessel diameter and frequency), provide general information on environments (e.g., Baas, 1986; Carlquist, 1975, 2001).

The objectives of this paper are to: (1) describe the geologic setting of the Cherokee Ranch wood locality; (2) describe and compare the Cherokee Ranch fossil woods to previously described Denver Basin woods; and (3) give the basis for proposing a new genus, Ubiquitoxylon.

The 13.7-square-km Cherokee Ranch (Fig. 1) is one of the best locations in the south Denver Basin for Paleogene petrified wood. Approximately 40 logs (up to 1.25 m in diameter and up to 10 m in length) have been found scattered over 600 square meters, the area within the upper D1 stratigraphic sequence that contains logs. The logs appear to have been carried by paleochannels, rather than an in-place fall, perhaps during high-water events, and deposited on and covered by sandbars composed of coarse sand. The log-bearing sandbars were then submerged below the water table. Logs appear stripped of branches and bark from transit. No vertical trees or stumps are present. Iron-oxide-cemented crusts of sandstone that coat and encase the petrified trees testify to the fact logs lay where originally deposited (Fig. 2). Rapid petrification by silica from the enclosing tuffaceous, feldspathic sandstone resulted in wood anatomical features being preserved in many logs. Many of the logs preserve enough detail to be identified as to family and genus. Topographic relief on Cherokee Ranch is 233 m. Cherokee Mountain (2,024 m) is clearly visible from U.S. 85 near Sedalia. All strata on the ranch are Paleogene in age. The history of private ownership of Cherokee Ranch and the rugged terrain of the site have helped preserve the wood from human predation over the years.

Paleogene strata are all mostly flat-lying with a regional 1–2° dip to the east. Formations exposed on the ranch include the Denver Formation (D1), Dawson Arkose (D2), and Castle Rock Conglomerate (Tcr) (Fig. 3). The lower 133 m (436 ft) of Denver Formation strata are believed to be Paleocene D1 sequence (Raynolds, 2002). Only the upper half of the 133 m (436 ft) of D1 is well exposed. The Denver Basin paleosol overlies the D1 and comprises the base of the Dawson Arkose, D2 sequence, of Eocene age. The Denver Basin paleosol on the ranch varies in thickness from about 5 m to 30 m. The D2 sequence is composed of up to 100 m of mostly conglomerate and sandstone. The Castle Rock Conglomerate (Tcr) of late Eocene age is a very coarse fluvial deposit, 83 m thick, that cuts into the underlying D2 and trends diagonally (NW–SE) across the south end of the ranch (Koch et al., 2018). The Castle Rock Conglomerate is older than the end of the Eocene (33.9 ma) because it contains bones of titanotheres (Thorson, 2011). In the D1, logs occur over a 47-m interval mostly near the top of D1. Three quarters of the logs are found in a 12-m interval, elevation ~1,899–1,911 m (~6,230–6,270 ft.). (Note: both the metric system, as preferred by Rocky Mountain Geology, as well as measurements in feet are used in various places of the paper because all of the older data that are referenced on Cherokee Ranch topographic maps and Castle Pine core holes are in feet.)

In 2015, A.J. Koch geologically mapped the Cherokee Ranch at 1:5,000 scale (unpublished data available at the Philip S. Miller Library, Douglas County Libraries’ Archives and Local History). The mapping shows that the Denver Basin paleosol is clearly exposed in multiple locations and can be correlated around the ranch and with the nearby Castle Pines core hole 27995F, 2.5 km to the east. The Denver Basin paleosol can be differentiated on the ranch from four other reddish, oxidized horizons that were also mapped. We believe that the Denver Basin paleosol mapped on the ranch is a dependable stratigraphic horizon to help understand the overlying and underlying stratigraphy. Limited radiometric dates suggest that the base of the Denver Basin paleosol is approximately 55 Ma (Raynolds, 2002). Eocene leaves have been confirmed in D2 on the ranch, ~90 ft (~30 m) above the base of the paleosol (I. Miller, personal communication, 2010; see Fig. 4 this paper).

No palynological work or age dating has been done on the ranch prior to one zircon laser ablation sample taken in 2015 (Koch et al., 2018). The nearest age control is in the Castle Pines core hole 27995F. Figure 4 shows the stratigraphic relationship between the Castle Pines core location, 2.5 km to the east, and a measured section in the area of the petrified wood on the ranch. The Denver Basin paleosol in the Castle Pines core overlies Paleocene age strata with palynological ages of P1 (oldest) through P3 (youngest) and one radiometric date of 63.94 ± 28 Ma (Obradovich, 2002; see Fig. 4 this paper). No younger palynological zones (P4–P6) were found in the Castle Pines core. Because late Paleocene age strata are rare in the Denver Basin (Nichols and Fleming, 2002), one might expect a similar age (P3) below the paleosol on nearby Cherokee Ranch; however, new age evidence plus more recent fieldwork argues that some of the ranch strata may be much younger than the P3 age in the Castle Pines core. More recent palynology work by Nichols, reported by Thorson (2011), indicates that P6-age strata may not be as rare as previously thought. Multiple samples taken by Thorson within 33 m of the base of the Dawson Arkose yielded P5 and P6 ages, leading Thorson to conclude that the age of the Dawson Arkose is still not completely understood (Thorson, 2011, p. 37). As discussed in the next section, our recent zircon laser ablation data does not agree with the Castle Pines core or conventional regional wisdom predicting older Paleocene as an age for our petrified wood horizons.

Koch et al. (2018) utilized laser ablation dating of zircons to separate lithofacies within the Castle Rock Conglomerate. As part of this work, a sample was taken on Cherokee Ranch in D1 sandstone adjacent to the petrified wood 2 m below the base of the Denver Basin paleosol. The distribution and age of the zircons in the sample show an undeniable geometric concentration within the Paleocene (65–55 Ma). The data are shown in Figure 5. The youngest zircon age in the distribution is 56 Ma. The strata cannot be older than the youngest zircon. Further, 90% of the 64 zircons in the distribution have Paleocene ages younger than 64 Ma, the youngest Paleocene in the Castle Pines core. Therefore, the sandstone in the D1 on Cherokee contains zircon ages that suggest palynological ages as young as P6. More sampling is warranted to support the laser ablation evidence.

The most probable provenance for the Paleocene zircons found on the ranch is the Laramide igneous province in the northern part of the Colorado mineral belt (40–80 km to the north). Koch et al. (2018) reached similar conclusions for the Paleocene zircons found in the Castle Rock Conglomerate. The zircon northern provenance is further supported by D1 conglomerates above and below the sampling site (Fig. 5). The conglomerates contain Precambrian Coal Creek Quartzite, derived from Coal Creek Canyon, which lies 60 km north of the ranch (Koch et al., 2018). As shown in Figure 4, the 56-Ma sample is at the same elevation and stratigraphically correlative with much of the petrified wood located 0.2 km directly to the west. On Figure 4, the enlargement of the log-bearing interval, a dashed jagged line shows a proposed lithofacies relationship between the northern provenance and western provenance ranch stream systems. The south-flowing streams from the north, carrying Coal Creek quartzite and zircons from the mineral belt, are interbedded and mixed with east-flowing stream detritus containing coarse arkosic sandstones derived from the Pikes Peak Granite to the west. During very high-volume flow, the confluence of these two stream systems was probably conducive to creating obstructions that could favor log jams.

1. Zircon laser-ablation dating evidence from a sample 2 m below the Denver Basin paleosol indicates an age no older than 56 Ma. So, uppermost D1 strata on the Cherokee Ranch appear much younger than the P3 palynology age in the Castle Pines core.

2. P5–P6 palynology ages have been found in four other south Denver Basin locations (Thorson, 2011).

3. North provenance conglomerates on the Cherokee Ranch can be correlated physically northward (10 km) to conglomerates in the Highlands Ranch quadrangle that underlie the Denver Basin paleosol.

4. D1 strata on the Cherokee Ranch are predominantly arkosic and sandy, lacking volcanic detritus that is typical of lower Paleocene stratigraphy.

5. The 56-Ma maximum age of the laser ablation sample agrees with the assumed 55-Ma age for the base of the Denver Basin paleosol.

Although our evidence is limited to one zircon age and some stratigraphic evidence, we interpret that at least part, and perhaps all, of the petrified wood discussed in this report is late Paleocene in age. More sampling and palynological work is needed to confirm how much late Paleocene section may be on the ranch. The new information on sediment supply from the north may help explain why younger Paleocene strata are deposited on the ranch as compared to much of the southern Denver Basin.

Pieces of fossil wood were examined in the field with a hand lens. Eighteen samples from 18 distinct logs that appeared to have reasonably good preservation were selected for thin sectioning. Transverse sections (TS), radial longitudinal sections (RLS), and tangential longitudinal sections (TLS) were prepared for each sample. Samples that were too compressed for accurate measurements of quantitative vessel features or too poorly preserved to show minute anatomical details are not discussed herein. The hand samples and slides are archived at the Denver Museum of Nature & Science in Denver, Colorado. The specimen archive at the museum has the catalog designation ‘DMNH’ (the institution was formerly called the Denver Museum of Natural History), and the specimens have been assigned DMNH numbers.

The descriptions of the woods use terminology consistent with the International Association of Wood Anatomists’ (IAWA) “List of Microscopic Features for Hardwood Identification” (IAWA Committee, 1989). Tyloses in vessels commonly obscured vessel element end walls so we could measure only a few vessel element lengths. Definitions of the IAWA hardwood list terms are available on the InsideWood website (http://insidewood.lib.ncsu.edu). For quantitative features, we present mean vessel diameters with the standard deviation in parentheses and ray heights as minimum–mean–maximum.

The possible affinities for the fossil woods were initially investigated using the multiple entry key of the InsideWood website (InsideWood, 2004-onwards; Wheeler, 2011), followed by a literature review. Appendix A gives the criteria used when searching for fossil woods with features similar to the Cherokee Ranch woods. The InsideWood database contains more than 9,000 coded descriptions of both modern and fossil dicotyledonous angiosperms and 46,000-plus images.

We computed the vulnerability index (VI = mean vessel tangential diameter divided by the mean number of vessels per square mm) for each wood. Carlquist (1975, 1977) proposed using VI values as a means of expressing the riskiness of a plant’s hydraulics. Low VI values (< 1) as found in woods with many narrow vessels suggest that these woods can tolerate environments with water stress. A plant with high VI values (> 3) is one that is unlikely to survive in environments subject to drought or freezing.

All the Cherokee Ranch samples analyzed to date are dicotyledonous angiosperms with characters commonly seen in previously described Denver Basin woods (Wheeler and Michalski, 2003). These common characters are: (1) wood is diffuse porous; (2) vessels are solitary and in radial multiples of 2–3, without a pronounced pattern of arrangement; (3) perforation plates between vessel elements are simple; (4) intervessel pits are medium to large in horizontal width and crowded alternate, polygonal in outline; (5) vessel-ray parenchyma pits, when observed, have much reduced borders or are apparently simple, and oval to horizontally elongate; (6) rays are heterocellular with procumbent body cells and 1–2 marginal rows of upright cells (the widest rays are 3–5 cells wide; one-cell-wide rays are rare; rays are < 1 mm high); (7) axial parenchyma is rare or scanty paratracheal with strands usually of four cells; and (8) fibers are non-septate and/or septate; fiber pits are not obvious. Moreover, the following features are absent: storied structure, helical thickenings in vessel elements, sheath cells, perforated ray cells, rays of two distinct size classes, radial canals, laticifers, and cambial variants.

This combination of features characterizes many Late Cretaceous to late Eocene fossil woods from the Northern Hemisphere. If oil cells are present as well, then the wood can be assigned to the Lauraceae (Laurel family) (Metcalfe and Chalk, 1950; Richter, 1987; Mantzouka et al., 2016; Jud et al., 2017). If the diagnostic feature of oil cells is absent, then it is usually impossible to confidently assign such woods to a single family because this combination of features occurs not only in the Lauraceae (order Laurales), but in other families in other orders, e.g., Anacardiaceae, Burseraceae, Kirkiaceae (order Sapindales), Lamiaceae, and Verbenaceae (order Lamiales).

In tangential sections, three of the Cherokee Ranch woods described herein have somewhat enlarged marginal ray cells, but in radial sections we could not confirm that these cells were oil cells. Nevertheless, we believe it most likely that these woods are Lauraceae. These enlarged cells are sparse, so it is possible the radial sections did not intersect them.

Naming fossil dicotyledonous woods having generalized features or having poor preservation so that some diagnostic features cannot be seen is problematic. Wheeler and Michalski (2003) referred to Denver Basin woods lacking oil cells and having the common anatomical characters described above as phyllanthoid type woods and did not assign them to genus. Although these woods share some features with ParaphyllanthoxylonBailey (1924), which is common in the Late Cretaceous and Paleocene of North America, they differ in ray features. Recently, one of us (EAW) examined slides of the holotype of Paraphyllanthoxylonarizonense and additional samples collected at and near the type locality. The rays in all samples have thin-walled cells, lack obvious marginal rows of upright and square cells, and usually have square cells intermixed with procumbent cells in the body of the ray (Wheeler and Lehman, 2009; Jud et al., 2017). Rays in the “phyllanthoid” Denver Basin woods, including these Cherokee Ranch woods, do not have those ray features.

The Denver Basin woods do not resemble Phyllanthaceae woods. Woods of this family have abundant uniseriate rays, and the multiseriate rays have well-defined uniseriate margins usually with more than four rows of upright cells (InsideWood, 2004-onwards). We suggest that it would be useful to have a genus for the common combination of features described above, and later in this paper we propose such a genus.

Dicotyledoxylon (Gottwald, 1992) and Dryoxylon Schleiden in Schmid (1853, p. 28) have been used for a variety of dicot woods of uncertain affinity, but are problematic. There is no diagnosis of Dicotyledoxylon. It is possible this was a typographical error and that Gottwald intended to use Dicotyloxylon (Chitaley and Patel, 1971). Dicotyloxylon was established for a young stem without secondary xylem, so it is not appropriate for woody stems. Dryoxylon includes woods with relatively few features in common, other than that they have vessels. Edwards (1931) noted, however, that the only species of Dryoxylon that Schleiden described was probably not a dicotyledon.

Unger (1842) was the first to describe a fossil wood with features of Lauraceae. Unfortunately, he used the name Ulminium for this wood, implying affinities with Ulmus (elm). Page (1967) pointed out that this name has priority over LaurinoxylonFelix (1883), which also was proposed for woods with lauraceous features. Unger’s original sample was located and used to provide an emended diagnosis for Laurinoxylon (Unger) Felix (Dupéron et al., 2008). Dupéron et al. (2008) and Doweld (2017) proposed rejecting the name Ulminium and conserving the use of Laurinoxylon for fossil lauraceous woods because Laurinoxylon has been more commonly used for lauraceous fossil woods. We are following their proposals and using Laurinoxylon. We are not formally proposing specific epithets for the Cherokee Ranch lauraceous woods, but informally designating them as Laurinoxylon Cherokee Ranch sp. A and Laurinoxylon Cherokee Ranch sp. B. We also relate them to the Laurinoxylon groups of Mantzouka et al. (2016), groups that are defined by the location of the idioblasts (oil cells).

Growth rings indistinct (Fig. 6A). Tangential diameter of vessels 95 (18) μm, 17–19 vessels per mm²; 42% solitary vessels; intervessel pits crowded alternate and polygonal in outline (Fig. 6B), 8–11 μm across. Vessel-ray parenchyma pits not found. Thin-walled tyloses present. Rays mostly 3–5 cells wide (Fig. 6C), ray height 256–420–660 μm; some rays appear homocellular composed of all procumbent cells, more commonly heterocellular with a single row of square to upright cells; occasional idioblasts in marginal rows of the rays (Fig. 6D); 4–5 rays/mm. Medium-thick-walled to thick-walled fibers, non-septate.

Material: DMNH EPI.40935

Comments:Mantzouka et al. (2016) used Werff and Richter’s (1996) classification scheme for Lauraceae tribes because it included wood anatomical characters. Mantzouka et al. (2016) suggested that modern genera resembling their ‘Group 1’ included members of Tribe Laureae (Laurus, the Litsea chinensis group) and Tribe Perseae: Dicypellium, North American Persea, Systemonodaphne, and Urbanodendron. We did not observe septate fibers in DMNH EPI.40935, so it is unlikely to be in Tribe Perseae, which has “ubiquitous septate fibers” (Werff and Richter, 1996). Information in InsideWood suggests that Litsea is most similar because rays in this genus can be wider than three cells (InsideWood, 2004-onwards; Wheeler, 2011). Litsea is primarily Asian, but there are a few species in Australia as well as some ranging from North America to subtropical South America (Huang et al., 2008). More recently, Huang et al. (2018) created the genus Litseoxylon for an Oligocene Chinese wood they considered to belong to the Litsea complex. Their generic diagnosis specified presence of helical thickenings throughout the vessel elements and occasional scalariform perforation plates; neither feature occurs in the Cherokee Ranch woods.

Wheeler and Michalski (2003) described but did not assign species names to some D1 sequence woods with oil cells. Both DB.D1 Xylotype 4a (six specimens) and DB.D1 Xylotype 4b (one specimen) have oil cells only in the ray margins and thus also belong to Mantzouka et al.’s (2016) Laurinoxylon Type 1. DB.D1 Xylotype 4a differs from the Cherokee Ranch wood as it has a few scalariform perforation plates. DB.D1 Xylotype 4b differs in having wider vessels (mean tangential diameter 134 μm), a higher vessel frequency (23–30 per mm²), and more frequent oil cells.

We searched InsideWood’s fossil wood database to determine if there were other fossil woods with a similar combination of features (Appendix A). There are 93 reliable records of fossil Lauraceae woods in the InsideWood database. Of these, Laurinoxylonrennerae (Estrada-Ruiz et al., 2018) from the Campanian McRae Formation of New Mexico is the most similar, but it differs in having some rarely occurring scalariform perforation plates.

Growth rings distinct, boundaries defined by changes in fiber radial diameters and a slight change in vessel diameter (Fig. 6E–F). Tangential diameter of vessels 99 (17) μm, 17–20 vessels per mm²; tyloses present, obscuring intervessel and vessel-ray parenchyma pits. Rays mostly 3–4 cells wide; 127–331–574 μm high, 4–5 / mm (Fig. 6G). Occasional idioblasts in ray margins (Fig. 6G) and amongst the ground tissue fibers (Fig. 6H). Medium-thick-walled fibers, septate and non-septate.

Material: DMNH EPI.40937; DMNH EPI.40941

Comments: Quantitative features primarily from DMNH EPI.40941. These specimens represent a new type of lauraceous wood for the Denver Basin, D1 sequence. The two features that set this wood type apart are: (1) idioblasts within the ground tissue fibers as well as in ray margins; and (2) distinct growth ring boundaries defined by changes in the fibers’ radial diameter and a slight change in vessel diameter.

This wood fits Laurinoxylon Type 2b of Mantzouka et al. (2016). According to them, Type 2b’s combination of features occurs in only three extant genera: (1) Actinodaphne p.p. (tropical and subtropical Asia); (2) Nectandra p.p. (tropical America); and (3) Neolitsea p.p. (Indo-Malaysia to eastern Asia).

Only one report of Laurinoxylon seemanianum Mädel (Selmeier, 1984) had the combination of features given above (Appendix A).

Generic diagnosis: Growth rings indistinct or distinct. Diffuse porous. Vessels solitary and in short radial multiples, mean tangential diameter between 50–150 μm; fewer than 40 vessels per sq. mm; perforations exclusively simple; intervessel pitting crowded alternate, usually polygonal in outline, medium to large; vessel-ray parenchyma pits with reduced borders, oval to horizontally elongate; average vessel element lengths less than 500 μm; axial parenchyma not common, sometimes scanty paratracheal; fibers non-septate and/or septate, medium-thick-walled to thick-walled, without distinctly bordered pits; multiseriate rays heterocellular with procumbent body cells and 1–2 marginal rows of square/upright cells, slightly inflated ray cells present or absent; uniseriate rays rare, multiseriate rays less than six cells wide; mean height of multiseriate rays less than 1 mm (usually < 500 μm). Storied structure, sheath cells, tile cells, rays of two distinct sizes, aggregate rays, normal axial and radial canals, laticifers, and cambial variants all absent.

Derivation of generic name:Ubiquitoxylon for the common occurrence of this combination of anatomical features in fossil and present-day woods.

Comments: As mentioned at the beginning of the ‘Results’ section, the combination of features of the Cherokee Ranch woods occurs in more than one family of more than one order, most commonly Anacardiaceae and Burseraceae (Sapindales), Lamiaceae, and Verbenaceae (order Lamiales), as well as Lauraceae (Laurales).

We searched the fossil wood database of InsideWood looking for fossil wood genera that might accommodate the combination of features given above (see Appendix 1 for coding).

Not surprisingly, our search of the fossil wood database returned records of woods suggested to have affinities with the Anacardiaceae, Burseraceae, and Lauraceae. Other suggestions included Euphorbiaceae, Phyllanthaceae, species of Paraphyllanthoxylon (including P. marylandense found in association with lauraceous fruits [Herendeen, 1991]), Carlquistoxylon (unknown affinities), and wood types (xylotypes) with distinctive anatomy, but of unknown affinities and not assigned to genus. BurseroxylonPrakash and Tripathi (1973 [issued 1975]) differs from the Cherokee Ranch woods because it has only septate fibers, ray cells that are frequently crystalliferous; vessel-ray parenchyma pits were not described. The genus Euphorbioxylon Felix emend. Mädel (1962) seems to have been misused by some authors. The diagnosis of this genus mentions abundant apotracheal parenchyma, which is not a feature of these Cherokee Ranch woods and of some other Euphorbioxylon species. Some fossil woods thought to belong to the Euphorbiaceae were assigned to Euphorbioxylon, even though their features did not match the generic diagnosis. For instance, axial parenchyma is rare in Euphorbioxylon hernenseCrawley (2001) and E. ortenburgenseSelmeier (1988). Differences with ParaphyllanthoxylonBailey (1924) have already been noted. Phyllanthaceae commonly have uniseriate rays, while uniseriate rays are rare in these Cherokee Ranch woods.

This wood is similar to Carlquistoxylon Wheeler, McClammer & LaPasha emend. Nunes, Pujana, Escapa, Gandolfo, Cúneo (Wheeler et al., 1995; Nunes et al., 2018) in vessel grouping, diameter, and frequency; perforation plate type, and parenchyma distribution. Carlquistoxylon differs in having longer vessel elements (500 and 800 μm) and narrower rays. Its diagnosis specifies only non-septate fibers, and slightly inflated ray cells do not occur in the genus. Samples assigned to Ubiquitoxylon and to Carlquistoxylon are of mature wood so the differences in vessel element length cannot be attributed to differences in age of the cambia that produced these woods.

Ubiquitoxylon raynoldsii Wheeler sp. nov. (Fig. 6I–L, Fig. 7A–O)

Specific diagnosis: Same as for genus

Specific epithet recognizes Dr. Robert (Bob) Raynolds for his extensive work on Denver Basin geology.

Holotype: DMNH EPI.40934 Paratype: DMNH EPI.40940

Additional material: DMNH EPI.40936; DMNH EPI.40938; DMNH EPI.40939

Repository: Denver Museum of Nature & Science.

There are some differences between DMNH EPI.40934, DMNH EPI.40938, DMNH EPI.40939, and DMNH EPI.40940 in vessel diameter, frequency, and groupings (Table 1). The vessels in DMNH EPI.40934 are more commonly solitary than in the other four samples, and the vessels in DMNH EPI.40938 show a slight tendency to a diagonal arrangement. We do not consider these differences sufficient to create additional species because this type of variation has been documented within some present-day species.

DMNH EPI.40936 differs from other Cherokee Ranch woods and from the D1 sequence phyllanthoid woods described by Wheeler and Michalski (2003) because its growth ring boundaries are more distinct. Changes in fiber radial diameter and vessel diameter are more obvious, and rays tend to be noded (inflated) at the growth ring boundaries. Otherwise, its anatomy is similar to DMNH EPI.40934, DMNH EPI.40938, DMNH EPI.40939, and DMNH EPI.40940. Growth ring distinctiveness can vary within a species, so we do not consider this difference enough to consider it a separate wood type and are including it in this species.

The VI values of Ubiquitoxylon samples range from 3.4 to 12.4, averaging 7.5.

The Denver Basin Project—an interdisciplinary, cooperative research effort to study the evolution of the Denver Basin—investigated 149 Upper Cretaceous and Paleogene sites and documented not only changes in megafloral composition through time, but also variation in coeval floras related to their proximity to the Front Range (Johnson et al., 2003). Different plant parts and families vary in the likelihood they will enter the fossil record. Invariably, when leaves and woods co-occur, the leaf assemblage is more diverse, yet some taxa are unique to the wood assemblage (e.g., Wheeler and Manchester, 2002; Allen, 2017). Consequently, fossil wood assemblages contribute to a fuller understanding of ancient vegetation.

The Cherokee Ranch D1 woods are a low diversity assemblage of large angiosperm trunks, probably all Lauraceae. The woods have similar hydraulic strategies, i.e., few and relatively wide vessels, a strategy only successful where there is little water stress (e.g., Carlquist, 1975; Baas, 1986; Beeckman, 2016). The high VI values of the Cherokee Ranch woods indicate that these trees were not subjected to frost or drought.

There is debate over how to recognize growth rings or growth interruptions in tropical trees (e.g., Silva et al., 2019); however, there is consensus that the lack of well-defined growth ring boundaries, as occurs in all but one of the Cherokee Ranch woods, indicates lack of pronounced seasonality in water availability or temperature (e.g., Poole and van Bergen, 2006). For the Cherokee Ranch woods, the features traditionally used for inferring paleoclimate complement the leaf physiognomic analyses of Denver Basin Paleocene leaf assemblages (Johnson et al., 2003).

To the best of our knowledge, no Paleocene angiosperm woods resembling the woods described herein or the previously described Denver Basin woods (Wheeler and Michalski, 2003) have been reported from Wyoming or Montana (InsideWood, 2004-onwards; Wheeler, 2011).

1. All Cherokee Ranch woods are dicotyledonous angiosperms and further document latitudinal variation in Paleocene floras and provide additional evidence for the Denver Basin vegetation being dominated by angiosperms.

2. A zircon age and stratigraphic evidence indicate that the Cherokee Ranch woods are late Paleocene in age.

3. There are two types of Lauraceae woods, and they have features of Laurinoxylon.

4. The combination of features seen in the majority of the Cherokee Ranch woods warrants establishing a new genus, Ubiquitoxylon.

5. The hydraulic features (vessel diameter and frequency) and lack of distinct growth rings imply that these trees did not experience water stress or pronounced dormancy.

6. We suggest that continued exploration for fossil woods in the Western Interior will help document regional variation in vegetation types and what might be the dominant tree species of the Paleocene.

John McKinney discovered and mapped most of the logs and helped select samples for analysis. National Petrographic Service Inc. of Rosenberg, Texas, prepared the thin sections. Archiving is provided by the Denver Museum of Nature & Science. Cherokee Ranch & Castle as well as the Cherokee Ranch Science Institute provided financial support for travel, thin sections, and zircon laser ablation. We thank Ian Miller for his extremely helpful and detailed review of our manuscript. We also thank Tom Michalski and Emilio Estrada-Ruiz for their constructive reviews. Rocky Mountain Geology (RMG) Co-Editor Arthur Snoke, RMG Managing Editor Brendon Orr, and RMG copy editor Robert Waggener of Waggener Editorial Services deserve thanks for their patience and thorough editorial work. George Mustoe gave permission to use Figure 1 from Mustoe and Viney (2017).

Features used: Growth rings indistinct (2p), wood diffuse porous (5p), vessels not tangentially or radially arranged (6a 7a 8a), vessels not exclusively solitary or commonly in radial multiples of four or more or commonly in clusters (9a 10a 11a), 13p (perforation plates simple), 22p 24a (alternate intervessel pits that are not minute), 40a 43a (vessel mean tangential diameter between 50 and 200 μm), 47p (5–20 vessels per sq. mm), 61p (fibers with simple to minutely border pits), 78p 79a 80a 83a 84a (axial parenchyma scanty paratracheal, not vasicentric or aliform or confluent or banded), 98p (larger rays commonly 4–10 seriate), 102a (rays < 1 mm high), 106p (heterocellular rays with procumbent body cells and one marginal row of square / upright cells), 124p 125a 126a (oil cells associated with rays, not with axial parenchyma or among the fibers).

Results of search: Two records in InsideWood’s fossil wood database had a similar combination of features. DB.D1 Xylotype 4a (Paleocene of Denver Basin), which differs in having some scalariform perforation plates, some samples with > 20 vessels per sq. mm, and slightly narrower rays (Wheeler and Michalski, 2003). Laurinoxylon rennerae from the late Campian McRae Formation, New Mexico, which differs in having some scalariform perforation plates (Estrada-Ruiz et al., 2018).

Features used: Growth rings distinct (1p), wood diffuse porous (5p), vessels not tangentially or radially arranged (6a 7a 8a), vessels not exclusively solitary or commonly in radial multiples of four or more or in clusters (9a 10a 11a), 13p (perforation plates simple), 22p 24a (alternate intervessel pits that are not minute), 40a 43a (vessel mean tangential diameter between 50 and 200 μm), 47p (5–20 vessels per sq. mm), 61p (fibers with simple to minutely border pits), 65p (septate fibers present), 78p 79a 80a 83a 84a (axial parenchyma scanty paratracheal, not vasicentric or aliform or confluent or banded), 96e 99e (rays not exclusively uniseriate or > 10 cells wide), 102a (rays < 1mm high), 106p (heterocellular rays with procumbent body cells and one marginal row of square / upright cells), 304r 124p 125a 126p (oil cells associated with rays and among the fibers, but not with axial parenchyma).

Results of search: Only one report of Laurinoxylon seemanianum Mädel, synonym: Cinnamomonoxylon seemanianum (Mädel) Gottwald (Selmeier, 1984) had the combination of features given above, albeit with variation in the presence/absence of oil cells associated with axial parenchyma or among the fibers.

Features used: Wood diffuse porous, vessels not arranged in a well-defined pattern, vessels solitary and in short radial multiples (5p 6a 7a 8a 9a 10a 11a), exclusively simple perforation plates (13r 14e), alternate intervessel pits that are not minute (22r 24a), vessel-ray parenchyma not the same as intervessel pits (30a), mean vessel diameter > 50 μm and < 200 μm (40a 43a), fewer than 40 vessels per sq. mm (49a 50a), mean vessel element length < 800 μm (54a), fibers without distinctly bordered pits (62a), diffuse-in-aggregate, vasicentric, aliform, confluent, banded, and marginal parenchyma all absent (77a 79a 80a 83a 85a 86a 89a), rays not exclusively uniseriate, not commonly > 10-seriate, not aggregate, not > 1 mm high, not of two distinct sizes (96a 99a 101a 102a 103a), rays heterocellular with < 4 marginal rows of square/upright cells (104a 105a 108a), sheath cells, tile cells, storied structure, oil cells, normal axial and radial canals all absent (110a 111e 120a 303a 304a 130a 305a).