Early Jurassic silicified woods from Carapace Nunatak, South Victoria Land, Antarctica

The Jurassic vegetation of Antarctica remains poorly known and, while there have been several reports of large fossil trees from that time period across the continent, detailed anatomical studies of their wood are extremely scarce. Here we describe new silicified woods of Early Jurassic (probably Toarcian) age from Carapace Nunatak, South Victoria Land. The genera Agathoxylon and Brachy-oxylon are formally recognized for the first time in the Jurassic of Antarctica. The preservation of the woods is imperfect, which is likely explained by the presence in some of the specimens of fungi, whose anatomical structures are described in detail. Combined with previous reports of pollen, leaves, and cones from South and North Victoria Land, these new specimens support the presence of several conifer families in the Early Jurassic floras of the region.


Introduction
Here we describe fossil woods collected at the locality during the austral summer 2014-2015. We discuss their affinities and how they fit within our knowledge of conifer diversity in Antarctica during the Early Jurassic.

Materials and methods
Locality and age of the specimens The fossil woods described in this study were collected from the moraine located on the E-NE side of Carapace Nunatak, southern Victoria Land, Central Transantarctic Mountains, Antarctica (76°53'S, 159°25'E, elevation 2150 m, Convoy Range Quadrangle; Fig. 1

Specimen preparation and imaging
Of the 10 putative wood specimens collected, four (#270, 271, 272, 274) were preserved well enough to observe all the characters important for wood taxonomy. Specimens 273, 275, 276 and 279 are more distorted and, while qualitative characters can be observed, obtaining a significant number of measurements proved more difficult. Finally, specimen 277 is insufficiently preserved to allow its anatomy description and taxonomic analysis. Specimen 278 lacked anatomical preservation: it consisted of a wood impression on a piece of sedimentary rock containing small fusains.
The specimens are heavily silicified and were prepared as thin-sections in the transverse, tangential, and radial planes following the classical technique (Hass and Rowe 1999). Images were taken in reflected light using ArchiMed software (Microvision Instruments, Evry, France) with Sony XCD-U100CR digital cameras attached to an Olympus SZX12 stereomicroscope and to an Olympus BX51 compound microscope, except for Fig

Statistical analyses
Statistical analyses were performed to describe ray height and its variability across the specimens, to better distinguish between wood morphospecies. They were performed using Rstudio software (version 4.0.3; R Core Team 2020, Boston, USA). For each specimen, thirty ray heights were measured on tangential or radial sections. Data normality was tested by a Shapiro-Wilk test and homoscedasticity, with a Levene test (package car, version 3.0-10, Fox and Weisberg 2019). Since normality and variance homogeneity hypotheses were most of the time not respected, only non-parametric tests were used to compare ray heights. Differences between specimens were evaluated using Mann-Whitney-Wilcoxon. Kruskal-Wallis and Multiple Comparison Kruskal-Wallis tests (package agricolae, version 1.3-3, de Mendiburu 2020) were used to compare specimens (Fig. 5). For the different tests, a significance level, α, of 0.05 was chosen.

Results
With the exception of specimen 271, which corresponds to a branch about 7 x 5 cm in diameter ( Fig. 2A), all the specimens represent pieces of wood with no indication of their origin within the plant. All wood specimens are composed of tracheids and parenchymatous rays. They share a number of characters but differ by ray height and type of radial pitting.

Anatomical characters shared by all samples
Specimen 271 shows several growth rings boundaries (Fig. 2B, C). In all other specimens, the limited extent of non-distorted wood prevents the detection of potential growth rings. The woods typically appear distorted and poorly preserved, often with the rays darker, more conspicuous, than the tracheids both in transverse (Fig. 2E, F) and in tangential ( They have thin, straight unperforated cell walls and often have a dark brown to black content. In radial section, the end wall is perpendicular to sub-perpendicular to the ray (e.g. Fig. 3). There are rare instances of axial parenchyma with no particular organization. When preserved, crossfield pitting is araucarioid, i.e. pits are mostly cupressoid, often in groups of 3 or more, organized in a relatively   However, the important across-specimen variability does not enable the use of this anatomical feature to distinguish between the morphogenera (Fig. 5A).

Fungal remains
One of the possible reasons for the poor preservation of the material is that the woods were already partly decayed when they were silicified. This is supported by the presence of abundant fungal remains in some of them (Fig. 6A-N). Direct evidence of fungal remains occurs in highest concentration in the rays and adjacent cells. Fungi consist of fragmented hyphae that are smooth and narrow (2-5 µm in diameter; Fig. 6A) to wider forms (≥6 µm in diameter; Fig. 6I), both types are connected (Fig. 6I) are sparsely septate (arrow in Fig. 6A; black arrow in Fig. 6E). Hyphae frequently produce perpendicular branching at uneven intervals (white arrow in Fig. 6C). Some hyphae are fractured and disarticulated ( Fig. 6B; black arrow in Fig. 6C), which is likely a result of preservation. Clamp connections are rare but present (Fig. 6D). Small ellipsoidal (6-8 µm long by 3-5 µm high) to spherical propagules occur terminally (Fig. 6E) or intercalary (Fig.  6F) on hyphae, which can be dark in color (Fig. 6E) to opaque (Fig. 6F). Similar structures do co-occur, or are connected, with fungal mycelia, but it is possible that they may represent preservational artifacts (arrowheads in Fig. 6G) or possible tyloses (Fig. 6H). Like the host wood, fungal mycelia are poorly preserved and degraded (Fig. 6G, I, J 1 -J 2 , K), which may have happened before burial or during the taphonomic process. Wide hyphae have a rough, crystalline texture (Fig. 6J 1 -J 2 ); it is possible that the thin, smooth hyphae could have been covered in mineral precipitates during  Fig. 6J 2 ), and 'wide' hyphae represent a biomimetic structure. Mycelia occur in multiple orientations in radial and tangential sections and traverse from tracheid to adjacent tracheid via the pits (Fig. 6K, arrows in 6L). Indirect evidence of fungi includes areas in transverse sections of wood that are highly degraded with numerous, and large, erosional notches and cavities (Fig. 6M), which gives lumina a 'starburst'-like appearance. At higher magnification, erosional troughs are present in all wall layers in adjacent tracheids (Fig. 6N). Although it is impossible to confidently determine the systematic affinity of the fungus (or fungi because there may be multiple species co-occurring within the wood), based on presence of clamp connections, at least one of the fungi is a basidiomycete. The degradational pattern within the wood is similar to decay by some extant soft rot fungi (Schwarze 2017: see figs 52, 54-55, 62-63, 68, 70), which makes the wood soft and spongy, thus, less conducive to preservational processes and making wood anatomical features difficult to discern.

Taxonomic affinities of the woods Philippe and Bamford (2008) published a key to
Mesozoic conifer-like woods that summarizes the diversity known at that time and the characters considered significant to distinguish the different morphogenera. The new specimens from Carapace Nunatak share the following characters from Philippe and Bamford's key: (1) all rays uniseriate, except for some local biseriation,  The specimens from Carapace Nunatak described in this paper indicate that at least three distinct wood morphotypes occurred in Victoria Land during the Early Jurassic: Protocupressinoxylon (= Protobrachyoxylon Holden), Agathoxylon, and Brachioxylon. The trunk from Coombs Hills reported by Garland might represent additional diversity. Very few wood types are reported from the high latitudes of Gondwana during the Jurassic. The cosmopolitan genus Agathoxylon was the only taxa listed by Philippe et al. (2004) in their cold temperate region -which included Antarctica-for the Early Jurassic, based on specimens from Australia. Reports are also extremely scarce for the Middle and Late Jurassic. This low diversity is to be compared with, for example, what is reported in the same region during the Cretaceous, when at least 6 wood morphogenera are present in the high-latitude belt: Agathoxylon, Araucariopitys, Circoporoxylon, Podocarpoxylon, Protocircoporoxylon, and Taxodioxylon (Philippe et al. 2004). The small number of wood morphotaxa in the Jurassic of Antarctica could be caused by a variety of factors, including paleoenvironmental constraints (climate, strong volcanism) and limited sampling. It is indeed important to consider that, while the presence of fossil wood is often mentioned in field reports, it is rarely sampled as extensively as other plant organs for which the presence of various genera can be assessed directly in the field, such as leaves or reproductive structures. In addition, among the sampled specimens, only a few have been the subject of taxonomic studies. The apparent scarcity of Jurassic woods in Antarctica is thus at least in part due to a collecting/study bias, a situation already reported for Triassic woods from this region (Oh et al. 2016).  Hieger et al. 2015). While Agathoxylon is a wood known to occur in a wide range of gymnosperm taxa, it is interesting to note that it is found in Araucariaceae and Voltziales (Philippe, 2011). It is thus possible that the Agathoxylon specimens from Carapace Nunatak are linked to the plant that produced Chimaerostrobus minutus, a pollen cone with a combination of characters reminiscent of these two groups of conifers (Atkinson et al. 2018).