We report exceptional preservation of fossil wood buried deeply in a kimberlite pipe that intruded northwestern Canada’s Slave Province 53.3±0.6 million years ago (Ma), revealed during excavation of diamond source rock. The wood originated from forest surrounding the eruption zone and collapsed into the diatreme before resettling in volcaniclastic kimberlite to depths >300 m, where it was mummified in a sterile environment. Anatomy of the unpermineralized wood permits conclusive identification to the genus Metasequoia (Cupressaceae). The wood yields genuine cellulose and occluded amber, both of which have been characterized spectroscopically and isotopically. From cellulose δ 18 O and δ 2 H measurements, we infer that Early Eocene paleoclimates in the western Canadian subarctic were 12–17°C warmer and four times wetter than present. Canadian kimberlites offer Lagerstätte-quality preservation of wood from a region with limited alternate sources of paleobotanical information.

Copyright: © Wolfe et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

The Panda kimberlite is one of approximately 150 pipes in the Lac de Gras field with emplacement ages ranging from 45 to 75 Ma [8] . The Panda pipe has a small diameter (∼200 m) and relatively simple geometry ( Fig. 1 ), with no evidence of multiple phreatomagmatic events. Two additional pipes, Koala and Koala North, are emplaced immediately to the southwest of Panda; all three have been exploited by open-pit diamond mining. The age of the Panda intrusion has been dated precisely by Rb-Sr determinations on kimberlitic phlogopite (n = 7), yielding a robust isochron of 53.3±0.6 Ma (Early Eocene, Ypresian) [9] . Wood is common in the upper 300 m of the kimberlite body, and small rounded fragments (<10 cm) are often floated during rinsing of crushed ores. Larger wood fragments (>50 cm) have been retrieved directly from the ore before crushing; these are typically encrusted in reworked volcaniclastic kimberlite ( Fig. 1D ). We focus on one such specimen obtained from BHP Billiton and originating from the 315 m bench of the mine. The specimen is exceptional because of its size and the preservation of an amber nodule revealed within the xylem upon splitting ( Fig. 1E ). This offers the rare opportunity to conduct parallel geochemical investigations of the wood and associated amber, and to diagnose the wood anatomically. We envisage that the source tree collapsed into the diatreme at the time of kimberlite emplacement. The great depth of burial suggests that it entered a narrow marginal boundary layer between the blast zone and the wall rock before becoming entombed. We consider the wood to be representative of the Early Eocene forest growing at the site at the time of magmatism. The lack of permineralization suggests that burial was rapid, and that little post-eruptive thermal or tectonic alteration has occurred at the locality.

A. Location of the Ekati diamond mine. B. Situation of the Panda kimberlite in relation to other pipes that comprise the property. C. Morphology of the Panda kimberlite pipe [8] and occurrence of wood. D. Fossil wood encrusted in olivine-rich volcaniclastic kimberlite. E. Photograph of the specimen characterized in this study. The wood was split when removed from the ore, revealing a sliver of opaque amber (9.5 cm long by 0.5 cm wide) in the xylem. F. RLS in transmitted light showing uniseriate and biseriate bordered pits and cross-fields. G. TLS showing rays stacked 3–26 cells high. H. SEM (TS) of ring boundary with earlywood (left) and latewood (right). I. Close-up of tracheids in TS and calcite crystals within cells (arrows). J. Cross-section of ray with cross-field pits. K and L. Close-ups of cross-field pits. M. TLS close-up of rays. N. Radial longitudinal section showing four contiguous rows of ray parenchyma cells with smooth end walls and no separation between the individual rows of cells.

Kimberlites are volatile-rich volcanic systems that ascend from the mantle episodically in Earth history as explosive phreatomagmatic events [1] . Fossils associated with kimberlites have been recognized for decades [2] , [3] . Two modes of preservation are possible: (1) collapse or entrainment into the diatreme at the time of emplacement, resulting in the emtombment of fossiliferous xenoliths within the kimberlite body; and (2) accumulation in the crater following magmatism (kimberlite maar sedimentation). Both preservation types occur in kimberlite pipes of northwestern Canada’s Slave Province [4] – [7] . Although abundant wood has been recovered at depth from the Panda pipe (64.73°N, 110.59°W; Fig. 1 ) during mining of the diamondiferous ore body on BHP Billiton’s Ekati property, these remarkable botanical fossils have not yet been described in detail. We report here on the preservation and identification of unpermineralized wood recovered from the Panda kimberlite, and present an array of stable isotopic measurements that are used to derive a provisional reconstruction of regional paleoclimate during the Early Eocene.

Results and Discussion

Wood Anatomy and Identity Wood from the Panda kimberlite has pristine preservation (Fig. 1). Only the exterior of the specimen (<1 mm) is fusinized, implying that little free oxygen was present at the time of burial. Tracheids measure up to 62 µm in tangential diameter. In radial longitudinal section (RLS), earlywood tracheids exhibit uniseriate (58%) and biseriate (36%) bordered pits, with the remainder unpitted. Pits are circular (diameter: 15–20 µm) and are characterized by circular apertures (5–6 µm) arranged in contiguous chains of 1–8 cells. Multiseriate bordered pits have dominantly opposite arrangement (98%). Crassulae are present. Rays possess thin horizontal walls composed of cells up to 248 µm long, 10 µm high, and 12 µm wide. Cross-fields have 1–3 oppositely arranged taxodioid-cupressoid pits per field. Axial parenchyma consists of vertically stacked cells up to 160 µm in height, with both smooth (65%) and nodular (35%) end-walls. Axial tracheid walls are sparsely pitted in tangential longitudinal section (TLS). Rays are 3–26 cells high and dominantly uniseriate (97%). In transverse section (TS), growth rings are narrow (1.5–2.0 mm) with marked boundaries between latewood (35–49 µm) and earlywood (10–15 µm) cells. Near the ring boundary, rays are up to 2.5 mm long and spaced on average 70 µm apart (i.e. 12–14 rays per mm). Both horizontal and axial resin ducts are absent. These features collectively identify the Panda wood as Taxodioxylon Hartig 1848 [10]–[12]. Moreover, fine anatomical details, including axial parenchyma with alternately nodular and smooth end-walls, single rows of taxodioid-cupressoid cross-field pits lacking separation between ray cells, and abrupt ring boundaries narrow the identification to Metasequoia Miki ex. Hu & Cheng 1948 [13]. Metasequoia was common in southern Alaska in the Late Paleocene and Early Eocene, producing a rich record of foliage and cones [14]. The genus is also abundant in younger (Middle Eocene, ca. 40 Ma) post-eruptive kimberlite maar sediments from the Slave Province, where it has been described as M. occidentalis (Newberry) Chaney [7]. The latter taxon is likely conspecific with the only extant congener, M. glyptostroboides Hu & Cheng, at present native only to isolated montane tracts in south-central China [13].

Cellulose Preservation Cellulose preservation in fossil conifers varies tremendously given the labile nature of constituent polysaccharides, mandating the need for quality control prior to isotopic analysis [15]. When viewed under scanning electron microscopy (SEM), extracts from Panda Metasequoia yield fibrous white material having the characteristic texture of wood cellulose (Fig. 2A–B). Fourier transform infrared (FTIR) spectra of the extracts support this contention: major absorption peaks including CH 2 deformation (900 cm−1) and CH 3 skeletal vibration (1375 cm−1) are common to both the Panda extracts and laboratory standard α-cellulose (Fig. 2C), yielding spectra that are fundamentally different from hemi- and holocellulose [16]. Furthermore, the removal of hemicellulose with NaOH results in markedly more crystalline x-ray diffraction patterns (Fig. 2D), which lends support to the contention that these extracts are primarily α-cellulose [17]. To our knowledge, this is the oldest verified instance of α-cellulose preservation to date, testifying to the remarkable preservation potential of kimberlite-hosted wood. PPT PowerPoint slide

PowerPoint slide PNG larger image

larger image TIFF original image Download: Figure 2. Cellulose extracted from Panda Metasequoia. A. SEM of cellulose fibers at low magnification. B. SEM at higher magnification showing surface texture. C. Duplicate FTIR spectra (800–1500 cm−1) of the same material, shown in relation to Middle Eocene cellulose from Axel Heiberg Island Metasequoia [15] and laboratory standard (Sigma-Aldrich) α-cellulose. D. X-ray diffraction traces of Panda cellulose extracts before (grey) and after (black) treatment with NaOH. Peaks associated with α-cellulose crystallinity are indicated by vertical dashed lines, whereas the idealized spectrum of pure α-cellulose is shown above the Panda traces [17]. https://doi.org/10.1371/journal.pone.0045537.g002

Amber Spectroscopy and Thermal Alteration FTIR spectra of amber fragments from the Panda wood indicate that they are dominantly composed of polylabdanoid diterpenes in the absence of succinic acid; the material is therefore classified tentatively as a Class 1b amber, consistent with other deposits attributed to cupressaceous conifers [18]. Indeed, FTIR spectra conform to modern and unaltered fossil Metasequoia resins in several regards (Fig. 3). The out-of-plane C = H deformation bands typical of cupressaceous conifers (887 and 975 cm−1) are well expressed in both materials, as are the positions of peaks associated with C–O stretching (1030 cm−1), CH 2 (2847 cm−1), and CH 3 (2870 cm−1) [19]. However, the Panda specimens have stronger broad-band O–H (3300–3400 cm−1), markedly reduced C = O in COOH (1693 cm−1), and greatly enhanced aromatic C = C (1520 cm−1) absorption bands relative to other congeneric resins. These features are well expressed by the spectroscopic difference between Panda amber and modern Metasequoia resins (Fig. 3E), and are interpreted to reflect a diagenetic sequence involving decarboxylation of cyclic hydrocarbons to more aromatic compounds, as observed elsewhere in the thermal maturation of conifer resins [20]. Abiotic processes operating under anoxic conditions are responsible for these chemical transformations. Additional observations are consistent with some degree of thermal alteration in the Panda material, including the opacity of the amber owed to microscopic bubble inclusions (Fig. 3), the fusinization of wood outer surfaces, and the presence of calcite (CaCO 3 ) crystals in wood cells (Fig. 1I), as confirmed by SEM-based energy dispersive spectroscopy. Calcite crystals are interpreted as in situ pseudomorphs following calcium oxalate (CaC 2 O 4 ) initially produced by secondary metabolism in the living tree prior to burial [21]. PPT PowerPoint slide

PowerPoint slide PNG larger image

larger image TIFF original image Download: Figure 3. Complete FTIR spectra (650–4000 cm−1) of fossil and modern Metasequoia resins. A–B. duplicate analyses of the amber from Panda wood. C. Middle Eocene amber from post-eruptive sediments in the Giraffe kimberlite pipe, which has not been thermally altered [7], [19]. Insets show typical fragments of Panda and Giraffe Metasequoia ambers (scale bars are 5 mm). D. Modern resin from M. glyptostroboides cultivar (Washington DC, USA) is visually identical to the Giraffe material. E. The spectroscopic differences between Panda and modern Metasequoia resins reveals features associated with thermal maturation (grey area is±1 SD of the difference). https://doi.org/10.1371/journal.pone.0045537.g003 Although it remains difficult to constrain eruptive temperatures during kimberlite emplacement due to the potential variability of volatile content and proximity to the local water table at the time of eruption, these are likely to have been in the 800–1200°C range initially, with the potential for considerable chilling (to 90–140°C) associated with adiabatic expansion during ascent [1], [22]. Thermal maturation of organic macerals from a range of Slave Province kimberlite diatreme facies has been evaluated by vitrinite reflectance, revealing maximum diagenetic temperatures of 350–450°C [23]. Wood from the Panda pipe has no cellular damage associated with devolatilization, whereas resin FTIR spectra provide no evidence for dehydration, confirming that the material was not exposed to exceedingly high temperatures (i.e. >500°C) upon burial. Despite the relatively subtle features attributed to thermal alteration noted above, we find little evidence that either the quality of cellulose preservation or the isotopic signatures of the various analyzed fractions have been overprinted. We thus envisage that cooling of the igneous body following emplacement in the diatreme was extremely rapid, potentially near-instantaneous [1], and surmise that any chemical changes to the entombed organic matter occurred in a closed system.