A dwarf conifer tree from the Triassic of Antarctica: the first fossil evidence of suppressed growth in a favourable climate?

Abstract

Background and Aims

The complexity of fossil forest ecosystems is difficult to reconstruct due to the fragmentary nature of the fossil record. However, detailed morpho-anatomical studies of well-preserved individual fossils can provide key information on tree growth and ecology, including in biomes with no modern analogue, such as the lush forests that developed in the polar regions during past greenhouse climatic episodes.

Methods

We describe an unusual-looking stem from Middle Triassic (~240 Ma) deposits of Antarctica with over 100 very narrow growth rings and conspicuous persistent vascular traces through the wood. Sections of the specimen were prepared using the cellulose acetate peel technique to determine its systematic affinities and analyse its growth.

Key Results

The new fossil shows similarities to the form genus Woodworthia and with conifer stems from the Triassic of Antarctica, and is assigned to the conifers. Vascular traces are interpreted as those of small branches retained on the trunk. Growth-ring analyses reveal one of the slowest growth rates reported in the fossil record, with an average of 0.2 mm per season. While the tree was growing within the Triassic polar circle, sedimentological data and growth-ring information from other fossil trees, including from the same locality, support the presence of favourable conditions in the region.

Conclusions

The specimen is interpreted as a dwarf conifer tree that grew under a generally favourable regional climate but whose growth was suppressed due to stressful local site conditions. This is the first time that a tree with suppressed growth is identified as such in the fossil record, providing new insights on the structure of polar forests under greenhouse climates and, more generally, on the complexity of tree communities in deep time.

INTRODUCTION

During the Triassic (~250–200 Ma), the global climate was significantly warmer than today, and diverse forests were growing in the presently barren high-latitude regions (Cantrill and Poole, 2012). Antarctica, then a part of the supercontinent Gondwana, was located at relatively high latitudes, with large parts of it well within the polar circle (e.g. Lawver et al., 1998; Veevers, 2004; Scotese, 2018; Fig. 1). Middle to Late Triassic deposits from the Central Transantarctic Mountains, northern Victoria Land, southern Victoria Land and the Prince Charles Mountains have yielded abundant fossils of bryophytes, lycopsids, sphenopsids, ferns, conifers, cycads and ginkgophytes, and of a variety of extinct seed-plant groups including Umkomasiales, Peltaspermales and Petriellales (Decombeix et al., 2010a; Escapa et al., 2011; Bomfleur et al., 2014,a). In addition to this lush vegetation, the existence of humid and mild climatic conditions in Antarctica during that time is supported by the abundance of peat and coal deposits, by the presence of frost-sensitive plants like cycads (Hermsen et al., 2007), and by high productivity as reflected in growth-ring data (Gabites, 1985; Taylor and Ryberg, 2007; Gulbranson et al., 2020) and in analyses of tree density in in situ fossil forests (Gabites, 1985; Cúneo et al., 2003).

The study of fossil plants from these Triassic polar forests provides information not only on the composition and diversity of plants growing in these environments without modern analogue, but also on their biology and ecology. For example, the structure of growth rings has been analysed in detail for several assemblages of fossil wood (Gabites, 1985; Taylor and Ryberg, 2007; Gulbranson et al., 2020; and references therein). The results support the existence of a mild climate allowing an substantial productivity during the growing season. They also reflect the strong seasonality of the photoperiod characteristic of high latitudes, with an abrupt cessation of tree growth as days shortened (e.g. Taylor and Ryberg, 2007). These studies, however, focused on general trends and not on the growth of individual trees. As a result, we still lack information on tree adaptation to local factors such as competition, edaphic conditions and topography (e.g. Fritts, 1976; Schweingruber, 2007) to fully understand the structure of Triassic polar forests.

Geological exploration in the far north of Victoria Land by international teams in 2015–16 recovered abundant plant fossils in the newly described Middle to Upper Triassic Helliwell Formation in the Helliwell Hills, central Rennick Glacier area (Fig. 1; Bomfleur et al., 2021). The Helliwell Formation contains carbonaceous and rooted overbank deposits, coal and plant compressions, as well as silicified peat and wood, reflecting the presence of highly productive and locally peat-forming vegetation growing under favourable conditions. Pieces of silicified wood collected from the Helliwell Formation are currently under study, and can be assigned to the morphogenera Agathoxylon and Kykloxylon. These two taxa represent the most common types of wood in Triassic deposits of Antarctica and can be respectively assigned to the conifers and to the Umkomasiales (= Corystospermales), an extinct group of seed plants (Decombeix et al., 2010b, 2014, and references therein; Bomfleur et al., 2013). However, the collection of silicified wood from the Helliwell Formation also includes one unusual-looking small gymnospermous axis, with over 120 very narrow growth rings and conspicuous persistent vascular traces through the wood. The objectives of this paper are to describe this peculiar fossil and to discuss (1) its possible systematic affinities and (2) elements of its life history based on the analysis of its growth rings and branch traces. We suggest that this unusual specimen represents the trunk of a dwarf conifer tree that grew under a generally favourable regional climate but whose growth was suppressed due to stressful local factors, such as poor soil conditions or shading from dominant trees. This adds to our understanding of extinct tree biology and the complexity of past forest ecosystems.

MATERIALS AND METHODS

The specimen is a ~2-cm-long fragment of a silicified woody axis about 6 × 6.5 cm in diameter. It was collected in austral summer 2015–16 during the 11th GANOVEX (German Antarctic North Victoria Land Expedition) as a piece of slope debris from a weathered hillside exposure of the Helliwell Formation (Bomfleur et al., 2021) at Helliwell Hills, central Rennick Glacier area, north Victoria Land (Fig.1). The Helliwell Formation is a ~235-m-thick sedimentary succession consisting mainly of dark-grey to blackish carbonaceous mudstone, locally with thin coal seams and intercalations of thin (centimetres to decimetres thick), greenish, ripple-laminated silt- and fine-grained volcanoclastic sandstones (Mörs et al., 2019). Two major cliff-forming intercalations (55 and 27 m thick) of fine- to medium-grained volcanoclastic sandstone occur in the middle and the upper part of the formation (Bomfleur et al., 2021). Its type section is located on the south-eastern slope of an unnamed inselberg in the northern Helliwell Hills (71°44ʹ2.0″S, 161°21ʹ36.1″E; Fig.1), where the present specimen was collected.

The formation is locally fossiliferous, with individual mudstone beds containing compressions of sphenophyte shoots and leaves, of lycopsid sporophylls, or of seed-fern fronds (Bomfleur et al., 2021). One bed near the top of the formation contains a mass accumulation of conifer leaves (Heidiphyllum elongatum), with occasional occurrences of putative conchostracans. Moreover, a slab of fine-grained sandstone in the central part of the section has yielded a vertebrate footprint (Mörs et al., 2019). Fossil wood occurs in the form of coalified logs in channel-fill deposits and in the form of loose blocks of silicified wood on weathered surface exposures of the formation (Bomfleur et al., 2021).

The depositional setting is interpreted as a ‘transitional facies’ (Collinson et al., 1987) between a meandering-stream system with extensive floodplains and a sand-dominated braided stream system (Bomfleur et al., 2021). Volcanoclastic components in the sandstones indicate strong contemporaneous volcanic activity, and the large abundance of plant fossils, silicified peat and coal indicates humid and highly favourable climatic conditions (Bomfleur et al., 2021).

The silicified specimen is permineralized and still contains some organic matter. It could therefore be prepared at UMR AMAP, Montpellier, using the classical peel technique (Joy et al., 1956; Galtier and Phillips, 1999): polished transverse and longitudinal section surfaces were etched for 3–4 min in 49 % hydrofluoric acid to remove the silica. The remaining organic matter was transferred onto sheets of cellulose acetate using acetone. The resulting peels were observed directly under the microscope. Selected portions of peels were also mounted in Geofix (Escil, France) on microscope slides to enhance contrast and allow better observation of finer anatomical details. The specimen, peels and slides were photographed using a Keyence VHX 7000 digital microscope and the associated software. Some general views are composed of several photos automatically stitched and focus-stacked during acquisition. Plates were prepared with Adobe Photoshop v.21.2.2 (Adobe Systems Inc.). Measurements were taken using ImageJ (Rasband, 1997–2018). Growth-ring thickness was measured on high-resolution photographs of peel GXI-HCSE-069 CT5 along the largest preserved radius, in the opposite direction and in the perpendicular direction. The early wood–late wood boundary was determined when 2a ≥ b, where a is the thickness of two adjacent cell walls and b is the diameter of the lumen, which corresponds to Denne’s first formula (Mork, 1928; Denne, 1989). This approach seems to be the best for detecting the presence of early wood–late wood boundaries in Permian and Triassic fossil wood from Antarctica (Taylor and Ryberg, 2007). However, we also provide the results using Denne’s second formula in Supplementary Data File S1. The surface of wood produced in a given ring (SR) was also calculated, using the following formula:

where SR_t is the surface of wood produced in year t, R_t is the radius of the stem at the end of the ring, and R_t−1 is the radius of the stem at the beginning of the ring (Biondi and Qeadan, 2008).

The specimen, peels and slides are part of the collection of the AG Paläobotanik at the Institut für Geologie und Paläontologie, Universität Münster, Germany, and accessible under number GXI-HCSE-069. Additional specimens from the locality used for growth-ring comparison are accessible under accession numbers GXI-HCSE-008, GXI-HCSE-041, GXI-HCSE-042 and GXI-HCSE-057. Average growth-ring values at other Antarctic Triassic localities were directly extracted from publications (Briand et al., 1993; Francis and Hill, 1996; Cúneo et al., 2003; Taylor and Ryberg, 2007; Decombeix et al., 2018) or calculated from their online supplementary data (Gulbranson et al., 2020).

DESCRIPTION

General aspect

The specimen corresponds to a thick slice of a stem 6–6.5 cm in diameter (Fig. 2A, B). Only 2 cm of its length is preserved. The stem is decorticated, with all tissues beyond the wood absent (Figs 2–4). Its external surface shows conspicuous small shoots coming out perpendicularly to the main axis (Fig. 4E). In transverse section, the stem has a small central pith surrounded by wood that contains conspicuous vascular traces and shows a significant number (>100) of thin to very thin growth rings (Figs 2–4).

Stele

The pith has an oval shape (Fig. 2C) that may be in part due to a slight compression of the axis. It is small, about 9 × 27 mm, and poorly preserved. The cells that are preserved are polygonal in cross-section, have relatively thin walls, and are interpreted as parenchyma. Some darker/brown areas might correspond to more sclerotic/secretory cells but no conspicuous sclerotic nests are visible (Fig. 2C, D). The periphery of the pith is poorly preserved and the location of the primary xylem is difficult to distinguish. No large conspicuous strands of primary xylem are present and there is also no evidence of medullary bundles or of centrifugal secondary growth (Fig. 2D).

Wood anatomy

The secondary xylem is 30–35 mm in thickness (Fig. 2A, B). It is exclusively composed of tracheids and parenchymatous rays. Tracheids are rectangular to slightly rounded in shape in transverse section (Fig. 2E). They have an average radial diameter of 22 µm in the innermost part (8.5–33 µm) and 20 µm in the outermost part (7.6–34 µm). In radial section, their walls display one or two rows of radial pits (Fig. 2F). Rays are uniseriate and only a few cells high in tangential section, typically one to five (Fig. 2G). Cross-field areas contain several pits (Fig. 2F).

Growth rings

The best-preserved side of the stem shows over 120 growth rings (Figs 3A and 5A). There is a strong change between the 30 or so inner rings, which are 0.5–0.6 mm in thickness (ranging from 0.3 to 1.7 mm) and the following ones, which are only a few cells in thickness and typically <0.2 mm thick, with an average of 0.15 mm (Figs 3B, C and 5A). In the remainder of the text they will be respectively called ‘inner rings’ and ‘outer rings’. Cumulative growth along three radii of the stem shows a similar trend, indicating that the growth is relatively equal around the stem (Fig. 5B). The small difference observed in the radius perpendicular to the other two may be due either to a slight natural asymmetry of the stem or to compression during diagenesis.

There is no marked difference in tracheid diameter between the two types of rings (Figs 3B, C and 6) and the difference in thickness of the outer rings is due to a reduced number of cells. Growth in the inner part of the stem is quite irregular, with important variations in thickness between two successive rings (Fig. 5A) and some variations in tracheid diameter and wall thickness within some individual rings (Fig. 6A). Individual rings can contain more than 30 cells radially; they are characterized by a few cells of late wood with a reduced radial diameter forming a conspicuous boundary. The outer rings have a different anatomy, with a boundary marked by a single row of tracheids with a reduced diameter (Figs 3C and 6B). The smallest rings are only four cells in thickness.

The surface of wood produced in each ring (annual surface increments) is another way to analyse the growth of the tree. This surface varies from 2 to 56 mm² in the inner rings, with an average of 30 mm² per ring. In the outer rings, the surface of wood produced ranges from 12 to 45 mm², with an average of 22 mm² per ring. There is no distinct trend towards an increase or decrease in the production of wood in terms of surface in the last 50 years of growth (Supplementary Data File S1).

Branch traces

Vascular traces crossing the wood cylinder are abundant in the preserved zones (Fig. 2A, B). Sections show some small traces close to the pith and at least one of these traces shows a connection to the pith (Fig. 4A, B). Traces can branch within the wood and typically have a small amount of secondary xylem (Fig. 4C–F). Although the preserved length of the stem is too short to understand the trace emission pattern, it does not seem random and the traces are likely not adventitious. When seen on the outside surface of the stem, the traces are a few millimetres in diameter and show a conspicuous pith and secondary xylem (Fig. 4E, F). There is no evidence that the organs vascularized by these traces were dead, i.e. no callus formation as seen in some other fossils from the locality. The vascular traces are thus interpreted as those of small woody branches that were still alive on the main axis at the time of its death.

SYSTEMATIC AFFINITIES

Possible assignment to Woodworthia

Some features of the new specimen from the Helliwell Hills are reminiscent of Woodworthia, a fossil morphogenus commonly used for Permian–Triassic stems and trunks from several regions of the world that show conspicuous small branch traces through their wood. Woodworthia was erected with the type species W. arizonica for Triassic fossil wood from the Petrified Forest in Arizona that shows small horizontal vascular traces crossing the entire radial width of the secondary xylem and visible as narrow (1.5–3 mm) pits or projections on the decorticated surface of the trunk (Jeffrey, 1910). Woodworthia is not very common in the Petrified Forest, but stems of up to 18 m length and fragments with a diameter of up to 76 cm have been described (Jeffrey, 1910; Creber and Ash, 2004). Jeffrey, and later Daugherty (1941), interpreted the vascular traces as those of short shoots, with a short-lived subtending leaf. Although the specimens have the wood anatomy of ‘a quite typical Araucarioxylon’, Jeffrey chose to erect a new genus because of what he interpreted as a combination of araucarian wood and abietinean short shoots (Jeffrey, 1910, pp 331–332).

In their detailed study of W. arizonica and Schilderia adamanica – another unusual wood from the Petrified Forest – Creber and Ash (2004) discussed the fact that W. arizonica could possibly be merged with Araucariopitys americana but still kept the two taxa separated. Creber and Collinson (2006) also included an early Permian specimen from the Irati Formation of Brazil in the species. This specimen showed ‘vascular traces, as in W. arizonica’. There is to the best of our knowledge no information regarding other details of the anatomy of that stem. Creber and Ash (2004) compared the traces in the wood of Woodworthia with those of preventitious epicormic buds in extant conifers. Creber and Collinson (2006) further developed this new interpretation and W. arizonica can now be understood as a tree with an Agathoxylon (formerly Araucarioxylon; Rößler et al., 2014) type of wood that could produce abundant preventitious epicormic shoots. This calls into question the systematic value of this genus given that (1) an Agathoxylon type of wood is known to occur in multiple different groups of fossil plants (Rößler et al., 2014 and references therein); (2) preventitious buds are present in a variety of plants today (Meier et al., 2012 and references therein), as they likely were in the geological past (e.g. Decombeix et al., 2010b, 2018; Césari et al., 2012); and (3) in extant plants such buds are produced neither equally in all specimens of a same species nor in equal numbers along the trunk length of a single individual (Meier et al., 2012 etc.).

The question also arises as to whether all the ‘traces’ seen in fossil stems with a Woodworthia morphology are equivalent, or if they represent the vascularization of different structures that would have a similarly small and persistent vascular trace, such as preventitious epicormic buds, short shoots, and/or long-lived leaves and small branches. In addition, information about significant characters of the primary vascular system (e.g. pith and primary xylem anatomy, mode of leaf trace production) and bark anatomy are lacking in the diagnosis of Woodworthia, so comparisons are only based on wood anatomy and the presence of numerous vascular traces in this wood. As a result, it is quite possible that fossil wood specimens assigned to Woodworthia may in fact encompass axes with various systematic affinities. In light of these uncertainties, we chose not to assign the new specimen from Antarctica to Woodworthia, although it would fit within the diagnosis of this genus.

Comparison with Kykloxylon and other Triassic Umkomasiales

Trunks and pieces of wood assignable to Kykloxylon are abundant at the locality. The genus Kykloxylon includes young stems and trunks from the Triassic of Antarctica (Meyer-Berthaud et al., 1993; Decombeix et al., 2014; Oh et al., 2016). Evidence from attached Dicroidium leaf bases (Meyer-Berthaud et al., 1992, 1993) and consistent co-occurrence at multiple localities indicate that the genus belongs to the Umkomasiales, an extinct seed-plant group. Similarities between Kykloxylon axes and the new specimen from Helliwell Hills include the presence of a small pith and of wood with small uniseriate rays. However, the new specimen can be distinguished from Kykloxylon based on the lack of ‘unusual’ secondary growth typical of older axes of this genus, such as unusual rings with some thin-walled cells and an irregular cambial activity producing lobed axes (Decombeix et al., 2014; Oh et al., 2016). In addition, traces in Kykloxylon are typically paired, which is not the case in the new specimen.

To date no other umkomasialean axis has been recovered from the Triassic of Antarctica. Jeffersonioxylon (Del Fueyo et al., 1995; Cúneo et al., 2003) and a portion of axis assigned to Rhexoxylon by Taylor (1992) have both been reinterpreted as belonging to Kykloxylon (Decombeix et al., 2010c, 2014). Other umkomasialean axes reported in South Africa and South America at the time typically also have ‘unusual secondary-growth’ features in their wood (Artabe and Brea, 2003). In addition, most of them have other conspicuous characters, such as centripetal secondary growth and/or medullary bundles. None of these features are present in the new specimen and an affinity with Umkomasiales is unlikely, at least in the current state of our understanding of their trunk anatomy.

Comparison with Agathoxylon and previously reported conifer stems (Telemachus, Notophytum)

Agathoxylon is a genus defined based only on wood anatomy. In the Triassic of Antarctica, such wood occurs, for example, in conifer taxa. One described in particular detail is the Notophytum wood of the putative voltzian conifer Telemachus (Meyer-Berthaud and Taylor, 1991; Bomfleur et al., 2013). Similarities with the new specimens include a small pith, a compact wood cylinder with low rays and no unusual rings, and simple (as opposed to paired) traces to lateral organs. Notophytum has conspicuous groups of sclerotic cells in the pith – a feature that is difficult to assess in the new specimen. All in all, the diagnostic anatomical characters observed in the new specimen are largely compatible with those of Notophytum, and we consider a coniferous affinity of the fossil stem to be most likely.

GROWTH AND HABIT OF THE PLANT

Slow growth and local palaeo-environmental conditions

A striking feature of the stem described in this paper is the very high number of small growth rings, more than 120, for a wood thickness of only ~3 cm. The largest measured ring is 1.7 mm thick but the average size for the inner (i.e. largest) rings is only 0.5–0.6 mm. In the 90 outer rings, the average thickness drops to 0.15 mm. The average growth for the whole stem is thus 0.2 mm per season, with a millimetre of wood added every 2 years in the first years and only every 6–7 years for most of the later life of the plant (Fig. 4). It could be objected that the plant actually formed more than one ring per year, a phenomenon noted today in regions with two favourable seasons (Cherubini et al., 2003). However, this interpretation does not fit with the presence of rings with a ‘normal’ width in the first years of growth of the plant. We thus chose here to consider that each ring of our new specimen does indeed correspond to one growth season, as has been done by previous workers on Triassic wood from Antarctica.

A comparison of the new specimen with previously reported growth ring data for these other woods shows that its growth is indeed unusual (Fig. 7). Middle Triassic in situ trunks from Gordon Valley in the Central Transantarctic Mountains have an average ring thickness of 0.9–2.5 mm (Cúneo et al., 2003). In allochthonous specimens with epicormic shoots from the same locality, the average thickness is 1.27 mm (Decombeix et al., 2018). An assemblage of wood from nearby deposits at Fremouw Peak reported by Taylor and Ryberg (2007) includes specimens with average ring thicknesses ranging between 0.5 and 5.30 mm, showing a 10-fold difference in average growth rate. The average ring width in this assemblage is 1.69 mm. Three specimens for which fewer than 15 rings were measured and that have larger rings than the others may represent central portions of young trees. When they are taken out of the analysis, the average ring width for the Fremouw Peak locality is 1.23 mm (Fig. 7). Woods from the Lashly Formation at Allan Hills, South Victoria Land, reported by Gulbranson et al. (2020) have rings ranging from 0.8 to 7.2 mm (average ring width 0.87 mm; n = 1342), while Kykloxylon woods from Skinner Ridge in northern Victoria Land described by Oh et al. (2016) have rings between 0.9 and 11.2 mm in thickness. Growth rings in four Kykloxylon trunk fragments collected from the same outcrop at about the same stratigraphic position as the new specimen range from 0.84 to 1.33 mm (average 1.07 mm; Fig. 7). This value is comparable to that found at other Triassic localities, and is different from that of the new specimen. The new fossil from the Helliwell Hills thus represents the slowest-growing woody stem reported to date for the Triassic of Antarctica

Growth-ring width reflects the favourability of growing conditions and is commonly used to extrapolate climatic conditions from fossil wood including in deep time and under global conditions very different from the present (e.g. Creber and Chaloner, 1984; Francis, 1986). However, while ring width is indeed linked to climate in extant trees (e.g. Falcon-Lang, 2005), it is clearly modulated by local environmental conditions – and to some degree by genetics (Fritts, 1976; Schweingruber, 2007). Dwarf conifers growing in unfavourable sites such as permafrost sites in Siberia, raised bogs or high-altitude sites also show very reduced growth, with an average ring width <0.1 mm (Schweingruber, 2007 and references therein). Angiosperm trees also show reduced growth in these conditions; for example, bearberry willows (Salix uva-ursi) from subarctic Canada reported by Opała-Owczarek and collaborators (2020) have an annual radial growth of ~0.2 mm. Such slow growth linked to unfavourable climatic conditions is also documented in the fossil record, although examples are extremely rare. Francis and Hill (1996) reported stems of Nothofagus beardmorensis from the Pliocene of Antarctica with a maximum diameter of 1.3 cm but with numerous (>60) growth rings, indicating they were mature specimens. The rings measure 0.04–0.58 mm, with average thickness of 0.07, 0.29 and 0.34 mm in the best-preserved specimens (Francis and Hill, 1996). Based on the contorted aspect of the stems, they were interpreted as representing dwarf trees with a prostrate habit. Unlike for the new specimen from Helliwell Hills, however, there is no evidence of other vascular plants co-occurring there with N. beardmorensis, and sedimentological evidence indeed indicates that these dwarf trees were growing in a tundra-like environment under a cold climate, with an average temperature estimated around −12 °C. Their growth, although comparable to that of the new specimen from Helliwell in terms of slowness, can thus be explained in this case by the general climatic conditions.

In light of the other data available from the Triassic of Antarctica, including from the same locality (Fig. 7), we consider that the slow growth of the new tree reported here was likely due to its individual site-specific growth conditions, and not to a more global factor such as the climate. Because the stems found at the Helliwell Hills locality are allochthonous, they may represent plants from different habitats that were transported and ended up deposited in the same area. Hence, possible factors explaining the slow growth of the new specimen could include a higher altitude, drier environment, poorer soil conditions, or higher stand density. It is also possible that the specimen comes from the same habitat as other trees but was subjected to slightly different biotic or abiotic factors. Local factors that can affect tree-ring width include light availability, which can be affected by neighbouring trees, and site conditions determined by topography and nutrient and/or soil water availability (Fritts, 1976; Schweingruber, 2007; St George, 2014). The new tree could thus have had a different social status in its stand (competition) or have been growing, for example, in a more exposed location or poorer soil conditions. In a study of eastern white cedars (Thuja occidentalis) from the Niagara Escarpment, Ontario, Canada, Briand et al. (1993) showed that trees from the exposed cliff faces (both young and old) had annual rings containing as few as three tracheids and an average ring thickness of 0.15–0.16 mm, a value similar to the outer rings of the new Antarctic specimen (Fig. 7). By comparison, the white cedar trees growing in less exposed places had rings that were on average ten times thicker and had up to 120 cells per ring. American beech (Fagus grandifolia) provides another extant example of slow tree growth in a favourable climate linked to limiting abiotic factors, in this case soil conditions. Busby and Motzkin (2009) describe dwarf beech stands in Massachusetts with trees <8 cm in diameter and characterized by growth rates of <0.1 mm per year for numerous years (>25), while adjacent forests on better soil conditions have large trees up to 70 cm in diameter. ‘Pigmy forests’ growing on podzols are another extant example of soil conditions limiting tree growth under a favourable climate (e.g. Jenny et al., 1969; Westman and Whittaker, 1975). The stunted growth of trees in these forests differs strikingly from that of adjacent communities.

Differences in light conditions can have an impact on tree growth comparable to that of local soil conditions and, within a stand, shaded or overtopped trees typically show significantly reduced growth rates compared with the neighbouring dominant trees (e.g. in Fagus sylvatica; Nicolini et al., 2001). A study of Faber’s firs (Abies fabri) from the Tibetan plateau (Liu et al., 2018) showed that the dominant trees had earlier onset and later cessation of xylem growth and hence a longer duration of xylogenesis than the suppressed trees from the same forests.

In the light of available information on extant tree growth and on growth-ring anatomy of co-occurring wood specimens, we interpret the unusual stem from Helliwell as representing an individual with growth reduced compared with coeval trees from the region. If instead of focusing only on the width of annual increments, one considers the total surface of wood produced each year, the new specimen is still remarkable by its slow growth, with less than 50 mm² (0.5 cm²) of wood produced per year for most of the life of the tree. The nature of the fossil record prevents us from identifying the exact cause of this extremely reduced growth. The fact that the inner rings are larger indicates that growth was less constrained during the first decades of the tree’s life. This could suggest later suppression by neighbouring trees (competition), but also a change in abiotic conditions such as increased soil acidity, nutrient runoff or waterlogging.

Nature of the branch traces

A second striking feature of the specimen is the abundance of small branches on the outer stem surface and their vascular traces visible in the wood cylinder. Even though the short preserved length of the specimen prevents us from following the origin of all the branches, there is evidence that (1) at least some of the traces are connected to the pith, and (2) traces seem to have a relatively regular pattern. These two elements support a preventitious origin of the branches (Meier et al., 2012 and references therein). The facts that they have a non-random organization and that their production is not correlated to the reduction in growth-ring width indicate that they are not epicormic branches produced as a reaction to external factors as seen in some extant taxa (Nicolini et al., 2001; Colin et al., 2012), but rather sequential branches that form part of the normal architecture of the tree. The branch traces are not surrounded by callus or other reaction tissue, which indicates that the branches were alive when the tree died. The fact that the branches are still borne on the trunk after more than 100 growth seasons suggests that they likely correspond to dwarf branches or short shoots that were not self-pruned as usually happens with older branches as the tree grows in height. This situation is in agreement with our interpretation of the Helliwell specimen as a dwarf tree with suppressed growth.

CONCLUSIONS

The excellent preservation of some permineralized fossil plants allows us to access detailed morpho-anatomical characters. By comparing some features with what is seen in extant plants, we can determine whether they are most likely linked to systematic affinities, to regional palaeo-environmental conditions or to the ecology of the studied individual. The new stem from the Helliwell Hills shows enough anatomical similarities with the form genus Woodworthia and with conifer stems previously reported from the Triassic of Antarctica such as Notophytum to be assigned to the conifers. It is interpreted as representing a dwarf tree, characterized by very slow growth and packed small branches retained on the trunk. This provides a new example of tree response to stress in deep-time ecosystems, and one of the slowest growth rates reported to date in the fossil record. The co-existence of trees with contrasting growth due to competition status or to local abiotic conditions is taken into account when analysing growth-ring data from living and subfossil woods in order to keep only the more general, climatic signal (e.g. Fritts, 1971; Yang et al., 2014). However, this phenomenon is not typically acknowledged in the fossil record, and to our knowledge this is the first time that the presence of a tree with suppressed growth in a fossil wood assemblage has been demonstrated. It seems very likely that a careful review of other fossil wood assemblages may reveal the existence of similar examples of suppressed trees, which would be of interest in order to fully understand the complexity of past forest communities.

SUPPLEMENTARY DATA

Supplementary data are available at Annals of Botany online and consist of the following.

File S1: growth-ring and cell measurements used in Figs 5–7, and calculated surface increment per ring.

FUNDING

This work received financial support by the Swedish Research Council (VR grant 2012-5323 to B.B.) and the German Science Foundation DFG (Emmy Noether project 268272651 Latitudinal Patterns in Plant Evolution and research grant 500787157 to B.B.).

ACKNOWLEDGEMENTS

We thank Yves Caraglio and Eric Nicolini (Montpellier, France) for helpful discussions on epicormic shoots and growth-ring anatomy in extant trees. We also thank Jakub Sakala and another, anonymous reviewer for their constructive and detailed comments that helped improve the manuscript. AMAP (botAny and Modelling of Plant Architecture and vegetation) is a joint research unit that associates Montpellier University, CNRS (UMR 5120), CIRAD (UMR51), INRAe (UMR931) and IRD (UR123).

AUTHOR CONTRIBUTIONS

A.L.D. contributed to the conceptualization and performance of the investigation, to its visualization, writing the original draft and editing the article. P.H. contributed to performing the investigation, to its visualization, and to reviewing and editing the article. B.B. contributed to the conceptualization of the study, funding acquisition, data acquisition, and to reviewing and editing the article.

DATA AVAILABILITY

The specimen and corresponding peels and slides are accessible under number GXI-HCSE-069 in the collections of the AG Paläobotanik at the Institut für Geologie und Paläontologie, Universität Münster, Germany. Measurements used for Figs 5–7 and the reconstruction of growth-ring area values through time are provided in Supplementary Data File S1.

REFERENCES