University of Birmingham A high-latitude Gondwanan lagerstätte : the Permian permineralised peat biota of the Prince Charles Mountains, Antarctica

• Users may freely distribute the URL that is used to identify this publication. • Users may download and/or print one copy of the publication from the University of Birmingham research portal for the purpose of private study or non-commercial research. • User may use extracts from the document in line with the concept of ‘fair dealing’ under the Copyright, Designs and Patents Act 1988 (?) • Users may not further distribute the material nor use it for the purposes of commercial gain.


Introduction
The Permian was a crucial period in the history of terrestrial life; the Cisuralian (Early Permian) saw the diachronous demise of the Carboniferous-style wetland floras that had dominated equatorial Euramerica during the Carboniferous and Cathaysia during the Asselian-Kungurian (Knoll, 1984;Hilton et al., 2002;Hilton and Cleal, 2007). In the Southern Hemisphere, the Permian witnessed the flourishing of glossopterid cool-temperate swamp forests, which dominated southern Gondwana until their extinction at the end of the period (e.g. White, 1998;McLoughlin, 2011b). Our knowledge of the diversity, vegetation structure, ecology, biotic interactions, and trophic links within these ecosystems is unfortunately limited by a paucity of konservat lagerstätten (sites of exceptional preservation of organisms) in comparison to other time periods in Earth history (e.g. Briggs and Gall, 1990;Selden and Nudds, 2004;Cascales-Miñana, 2011). This preservational bias has hindered our understanding of the developments in austral terrestrial ecosystems during the Permian. A more detailed picture of the trophic complexity and inter-relationships between plants, insects and soil microorganisms would enhance our understanding of how terrestrial communities evolved in the wake of the Gondwanan glaciations up to the end-Permian biotic crisis. The end-Permian mass extinction, which purportedly eradicated up to 95% of all life (Benton and Twitchett, 2003), marks the most significant reduction of diversity in the Phanerozoic. Unlike the Cretaceous/Palaeogene (K/Pg) extinction event 66 Ma, which was likely precipitated by an instantaneous impact mechanism (Alvarez et al., 1980;Vajda and McLoughlin, 2007), the terrestrial biotic turnover at the end of the Palaeozoic appears to have developed as a multiphase series of extinctions (Racki and Wignall, 2005;Yin et al., 2007;de la Horra et al., 2012) that were not necessarily synchronous between disparate regions (Rees, 2002). Evidence from several sources including brachiopod, bivalve, foraminiferal and plant extinctions, shows that major biotic disruptions began in the Capitanian (although often erroneously referred to as the 'end-Guadalupian extinction') and this was followed by a protracted diachronous decline in Palaeozoic life throughout the rest of the Permian (Yin et al., 2007;Bond et al., 2010). Discussion of the timing and proposed causes of the extinction(s) beginning in the Capitanian can be found in several sources and include diverse triggers and mechanisms for biotic turnover, such as the massive outpourings of flood basalts, which now form the Siberian Traps, climatic warming, increased aridity, loss of coastal habitats, ocean anoxia or a combination of these mechanisms (Racki and Wignall, 2005;Retallack et al., 2006;Yin et al., 2007;Clapham et al., 2009;Isozaki, 2009Isozaki, , 2010Ali, 2010;Bond et al., 2010;de la Horra et al., 2012;Benton and Newell, 2013;Retallack, 2013).
The Roadian-Wordian-aged Toploje Member chert of the Prince Charles Mountains (PCMs) preserves, in exceptional three-dimensional detail, the permineralised remains of a terrestrial mire ecosystem prior to the biotic decline that began in the Capitanian and continued through the Lopingian until the Permo-Triassic transition (Retallack et al., 2006;Yin et al., 2007;Bond et al., 2010;de la Horra et al., 2012;Retallack, 2013). The Toploje Member chert offers a snapshot of the final phases of 'stable' terrestrial life before the crisis that engulfed and overthrew these ecosystems.
Aside from its significance in recording a key episode in terrestrial life, the Toploje Member chert also preserves an important in situ community of macro-and micro-organisms that constituted part of the high-latitude Glossopteris mire flora that typified vast expanses of southern Gondwana during the Permian (e.g. Anderson et al., 1999;Pigg and Nishida, 2006;McLoughlin, 2011b) and contributed to the Southern Hemisphere's major economic coal resources. The structure of Gondwanan coal is relatively well understood in terms of maceral content and distribution (e.g. Navale and Saxena, 1989;Diessel and Smyth, 1995;Osório et al., 2006;Kalkreuth et al., 2010;Van de Wetering et al., 2013), but how this relates to the taxonomic representation of plant constituents and their component parts is less well resolved, since the transition from peat to coal involves significant volumetric and compositional changes due to differential compaction of plant parts and diagenetic loss of volatiles. Quantitative comparison of the constituents of the Toploje Member permineralised peat and coals from the same stratigraphic unit provide a means of evaluating the original composition of the coal-forming biota and the changes in coal composition with diagenesis.
The diversity of species in ancient terrestrial ecosystems is inherently difficult to assess. Although not without taphonomic filtering, marine deposits tend to offer a much richer sampling of the shelly biota in the environment as a consequence of bioclast persistence and sedimentary sorting (see Cleal et al., 2012). Therefore, it falls to the patchy occurrences of terrestrial konservat lagerstätten to provide a more accurate picture of what life was like at any one place in time on land. Exceptional preservation occurs elsewhere in Antarctica during the Late Permian with silicified plant remains known from two main deposits in the central Transantarctic Mountains. The Skaar Ridge and Collinson Ridge silicified peats of the Transantarctic Mountains appear to be derived from small lenses or possibly fluvially rafted mats of peat associated with volcaniclastic sediments (Taylor et al., 1989;McManus et al., 2002), in contrast to the laterally extensive Toploje Member chert representing a large autochthonous mire community preserved in a succession lacking volcanogenic sediments.
We employ a battery of techniques to elucidate the biotic constituents of a typical peat-forming mire community in Gondwanan high latitudes during the Middle Permian. Further, we assess the taphonomy, quantitative representation of components, and evidence of biotic and other physical interactions to elucidate the depositional setting and palaeoecology of the coal-forming mires of the Lambert Graben. We also survey the fossil record of plantarthropod interactions across Gondwana to assess the diversity and importance of disparate herbivory strategies in high-latitude glossopterid-dominated communities prior to the end-Permian biotic crisis. Finally, we contrast these findings with the results of a maceral analysis of associated coals to assess whether the petrography of the coals provides a meaningful representation of the original peat community structure.

Material and methods
Samples of a range of sizes were collected from multiple localities along a low ridgeline exposing the Toploje Member chert (Fig. 1). A variety of palaeobotanical techniques were then used to investigate the contents of the peats; blocks of the chert were sectioned using a Buehler Isomet 5000 linear precision saw. Following sectioning, acetate peels were produced from the blocks using the technique outlined by Galtier and Phillips (1999). The surface of each block was polished and then submerged in a shallow bath of cold 30% hydrofluoric acid solution for approximately 90 s in order to etch away the silica matrix and leave a thin layer of organic matter standing proud of the surface. Each block was then rinsed with distilled water, dried, then covered with acetone before laying a cellulose acetate sheet on the surface to create a peel that was then studied using a transmitted light microscope. Peels were found to be inferior to thin sections for the study of fungi, Peronosporomycetes and coprolites in accordance with the findings of Taylor et al. (2011) and, where possible, thin sections were preferentially produced for the study of these elements. Several samples from a range of localities across the peat outcrop were selected for bulk maceration in a cold 30% hydrofluoric acid solution. Samples were left in the solution for two weeks and then the remaining organic debris was extracted using a 150 micron nylon sieve. Sieved organic remains were then placed into a petri dish of distilled water and studied using an optical stereomicroscope. Plant, arthropod and fungal remains were then picked while hydrated using a fine art brush. Elements of interest were then mounted on aluminium stubs and sputter-coated with gold to enhance conductivity for imaging with a Hitachi S-4300 field emission scanning electron microscope at the Swedish Museum of Natural History (Naturhistoriska riksmuseet). Several elements of the flora extracted via bulk maceration were also analysed using synchrotron Xray tomographic microscopy. X-ray microtomography was conducted at the TOMCAT beamline of the Swiss Light Source, Paul Scherrer Institute, Switzerland  using the techniques described by Donoghue et al. (2006). Illustrated material is registered in the palaeobotany collections of the Swedish Museum of Natural History, Stockholm (prefixed NRM) and Geoscience Australia (prefixed CPC).
Quantitative analysis of the silicified peat was made by point counting across 20 randomly selected thin sections at 200 μm increments for 4000 points. In addition, four thin sections made from charcoal-rich samples were selected for point counting to analyse variation in peat composition between regular and wildfire-affected microfacies within the Toploje Member chert.
The organic petrology of a selected set of Middle to Upper Permian coal samples from the Bainmedart Coal Measures was carried out by a commercial coal analytical contractor (Keiraville Konsultants Pty Ltd, Wollongong, Australia). Results from proximate analyses of these samples were presented by Holdgate et al. (2005); only the data on maceral proportions in the coals are presented here.

Geological setting and palaeogeography
Antarctica occupied a central position within Gondwana through the late Palaeozoic and early Mesozoic (McLoughlin, 2001;Fig. 1). This location endowed Antarctica with a key role in floristic interchange between the various peripheral regions of the supercontinent (McLoughlin, 2001;Ryberg, 2010). Outside the Transantarctic Mountains, the only Permo-Triassic sedimentary succession in East Antarctica is preserved in the Lambert Graben within the Prince Charles Mountains region. The Lambert Graben has been interpreted to represent the southern extension of the Mahanadi Graben in India in pre-breakup palaeogeographic reconstructions of Gondwana (Fedorov et al., 1982;Stagg, 1985;Veevers, 2004;Harrowfield et al., 2005;Boger, 2011;Slater et al., 2011), although alternative alignments with the Godavari Graben have also been mooted . Throughout the Early and Middle Permian, the northern Prince Charles Mountains (Fig. 1) occupied a palaeolatitude of 65-70°S (McLoughlin et al., 1997), similar to its modern position. The studied chert constitutes the silicified uncompressed upper portion of a coal seam, which forms the uppermost bed of the 303 m thick Toploje Member (Fig. 2) of the lower Bainmedart Coal Measures in the northern Prince Charles Mountains, East Antarctica (McLoughlin and Drinnan, 1997a). The chert bed is locally up to 40 cm thick and is exposed over a strike length of 3 km (Fig. 1). The Bainmedart Coal Measures represent the middle unit of the Permian-Triassic Amery Group (see Fielding and Webb, 1996;McLoughlin and Drinnan, 1997a, for detailed stratigraphic sections of this unit). The coal measures rest disconformably or partially unconformably on the Radok Conglomerate of Kungurian to early Roadian age, and are in turn conformably overlain by the Flagstone Bench Formation of Triassic age (McLoughlin and Drinnan, 1997b). The Bainmedart Coal Measures are dominated by thick cross-bedded sandstones, siltstones and coals deposited in a cyclic pattern, which has been attributed to the action of Milankovitchinduced changes to sediment supply (Fielding and Webb, 1996).
The chert bed represents the upper part of a peat profile that was impregnated with silica prior to any significant sedimentary loading or compression, hence the entombed plant remains have avoided coalification . Palynological correlation to the Australian Didecitriletes ericianus Zone gives a Roadian-Wordian age for the Toploje Member chert (Lindström and McLoughlin, 2007). Flooding with mineral-rich lake waters and then permineralisation terminated the peat-forming process. The conformably overlying Dragons Teeth Member consists of a series of sideritic shales and minor sandstones deposited in a lacustrine setting (Fielding and Webb, 1996).
Data and findings of this study of the Toploje Member chert will now be presented and discussed.

Composition of the silicified peat
The macrofloral diversity within the silicified peats is relatively low, being dominated by the constituent dispersed organs of arborescent glossopterid and cordaitalean gymnosperms. Matted leaves of both Glossopteris (Glossopteridales) and Noeggerathiopsis (Cordaitales), together with roots (Vertebraria) and stem wood (Australoxylon) are the most prominent constituents of the peat (McLoughlin and Drinnan, 1996;Holdgate et al., 2005). However, the cryptic micro-and mesofossil components of the peat reveal a much greater biotic diversity in the mire ecosystem than is evident from the macroscopic remains. The fossil micro-organism assemblage includes a broad range of fungal hyphae and reproductive structures together with superficially similar organisms such as Peronosporomycetes (Slater et al., 2013). Although arthropod exoskeleton fragments are sparse, disarticulated and fragmentary, a rich entomofauna is indicated by the wealth and diversity of invertebrate feeding traces and coprolite morphotypes, both dispersed in the peat matrix and preserved within specific plant organs (Weaver et al., 1997;Holdgate et al., 2005;Slater et al., 2012; see Table 1 for a list of the biota found to date in the peats). The preservation of groups such as saprotrophic Peronosporomycetes (Oomycota) adds to the sparse, but growing fossil record of these important elements of terrestrial ecosystems (Schwendemann et al., 2009;Krings and Taylor, 2011;Slater et al., 2013). These, together with a broad diversity of fungal interactions, show that the glossopterid plant was the primary host of a 'component community' of saprotrophs, herbivores and detritivores at high latitudes until the end-Permian extinction of this plant group. The roles of soil-inhabiting microorganisms in modern high-latitude peats are still poorly understood (Tveit et al., 2012), so additional investigation of fossil occurrences will improve our knowledge of the evolution of such ecosystems (Adl et al., 2010). The dispersed palynoflora in the silicified peat has not yet been fully documented but includes a broad range of fern, sphenophyte, and lycophyte spores and cycadophyte and pteridosperm pollen that attest to a much higher floristic diversity in the immediate vicinity of the mire ecosystem. Diverse glossopterid pollen morphotypes are also evident within the peat but significant intraspecific variation of the grains has been recorded within Arberiella (glossopterid) sporangia from the peat layer  suggesting that apparent levels of glossopterid diversity in the ecosystem based on dispersed pollen floras are inflated. Although Paracalamites australis (sphenophyte) axes occur sparsely as impressions in sediments of the Dragons Teeth Member overlying the Toploje Member, macrofossils of this group have not yet been identified in the silicified assemblage. A single undescribed herbaceous lycophyte taxon has been recorded from the permineralised peat .
Analysis of the relative abundance of the various organisms and their constituent organs/tissues contributing to the peat was carried out by Holdgate et al. (2005) and again in this study using a different set of thin sections of the Toploje Member chert (Fig. 3A). The new results largely confirm the gross composition of the peat identified by Holdgate et al. (2005), although some differences were noted. Vertebraria roots form an even larger proportion of the total peat composition than was found in the previous study. Further, charcoal was found to be proportionally more abundant in this study. It is clear that the peats of the Toploje Member chert contain a higher proportion of subterranean components, such as roots, compared to subautochthonous canopy elements, such as leaves and seeds, than is evident in other Permian silicified peats, such as those of the Bowen Basin, Australia (McLoughlin, 1992). These differences attest to subtle variations in microfacies within the peat profiles that will require further investigation to delineate their composition and genesis. Our preliminary assessment of compositional variation in the Toploje Member peats via a series of point counts on thin sections of specimens rich in charcoal ( Fig. 3B) reveals plant assemblages dominated by charcoalified wood. In general, these charcoal-rich microfacies contained much more woody material (both charcoalified and noncharcoalified) than is evident in the typical composition of the peat.

Composition of associated coals
In order to assess whether the Toploje Member's silicified peat bed is representative of organic accumulations throughout the Middle and Late Permian of the Lambert Graben, 13 coal samples from the major coalbearing intervals of the Bainmedart Coal Measures were analysed for their maceral content. One of these samples from outcrop near Soyuz Station on the eastern side of Beaver Lake is not constrained stratigraphically and is only included for maceral ratio comparisons to coals of other regions. The other 12 samples derive from reconnaissance-scale sampling of coals throughout the Bainmedart Coal Measures (generally only 1 sample per 100-150 m of stratigraphic section: see Holdgate et al., 2005, fig. 3). Trace and major element results, vitrinite reflectance values and some petrological data from these samples were presented by Holdgate et al. (2005). Petrological and coal maturity data were published for a few additional samples from the lower part of the Bainmedart Coal Measures by Bennett and Taylor (1972). Here we elaborate on the organic petrology of the coal and compare the compositions of the coals and silicified peat. The~40 cm of silicified peat is chosen for comparison with the coals since it represents a snapshot of the peat before compaction, diagenesis and coalification.

Vegetation structure
Vegetation stratification is evident in the assemblage of plants preserved in the Toploje Member chert. Autochthonous upright stumps with a height of 30 cm and diameter of up to 20 cm are locally preserved in the chert together with more abundant parautochthonous axes generally b20 cm in diameter preserved parallel to bedding . These stumps and stems attest to the presence of abundant arborescent gymnosperms. Two stem wood morphospecies, Australoxylon bainii and Australoxylon mondii, were recognised by Weaver et al. (1997) and clearly correspond to two similar gymnosperms; the most likely affiliations being with glossopterids or cordaitaleans based on the co-preserved leaf genera (Glossopteris and Noeggerathiopsis). Australoxylon mondii stem wood cannot be distinguished anatomically from Vertebraria (glossopterid) root wood, apart from the absence of  schizogenous cavities, suggesting that they likely originated from the same glossopterid plant (Weaver et al., 1997;McLoughlin, 2011b; see Bateman and Hilton, 2009). Two morphotypes of Vertebraria roots were recognised and described from the peats by Neish et al. (1993), categorised as 'polyarch' and 'solid cylinder' based on their ontogeny and architecture (Fig. 5A). The polyarch and solid cylinder roots may correspond to two distinct plant species or may be functionally different roots belonging to the same parent glossopterid plant (Neish et al.,  Baxendale (1979), plate 65, figure 1) and akin to Baxendale's (1979), plate 65, figure 8] 'type B' coprolites from Pennsylvanian coal balls, scale = 1 mm. E. NRM S089553, matted Noeggerathiopsis leaves with prominent abaxial trichome-bearing furrows, scale = 500 μm. F. CPC34952, transverse section through the axis of a herbaceous lycophyte, scale = 500 μm. G. NRM S087932-01-03, pyrite crystals amongst plant debris embedded in silica matrix, scale = 500 μm. 1993). A broad range of dispersed bisaccate, monosaccate, monosulcate and polyplicate pollen within the peat denotes an additional range of shrub-to tree-sized gymnosperms in the vicinity of the mire.
Stems and microphylls of diminutive herbaceous lycophytes are moderately common within the peat and are co-preserved with three genera and species of megaspore . These appear to have been centimetre-scale Paurodendron-like heterosporous lycophytes that grew on the consistently moist peat surface (Fig. 5F). Roots of these plants occur sporadically within the leaf debris microfacies of the peat.
Fern sporangia with intact contents are relatively common in both thin sections and bulk macerations of the Toploje Member chert, though no leaves or stems of these plants have yet been discovered. The abundance of sporangia and paucity of fern vegetative remains may be the result of: (1) an extremely delicate nature of the plants resulting in their low preservational potential; (2) an epiphytic habit, with ferns residing high on the stems of Glossopteris trees and retaining their foliage there until decay; or (3) transport of sporangia and spores into the mire via wind or water from hinterland plant communities. Small epiphytic ferns of Tubicaulis sp. are known from Early Permian deposits from Chemnitz in Saxony, Germany (Rößler, 2000). These epiphytes occur among the assemblage of plants that grew in close association with a mantle of adventitious roots of the well-studied tree-fern Psaronius (Rößler, 2000). No direct evidence of epiphytes has yet been found associated with glossopterid remains, though they are a common feature of most modern moist forest communities (Bartels and Chen, 2012).
Adaptations for the high-palaeolatitude environment include a deciduous habit for glossopterids based on the numerous thick mats of leaves that occur as compression fossils and in the silicified peat representing seasonal leaf-shedding events (Retallack et al., 1995;Krull, 1999;Retallack, 1999;Holdgate et al., 2005;McLoughlin, 2011b). Such mats of apparently monospecific glossopterid leaves are common features of the Toploje Member chert (Fig. 5D). Australoxylon bainii and A. mondii woods found in the peats also have distinct growth rings that terminate abruptly, indicating a swift transition to winter dormancy (Weaver et al., 1997;Gulbranson et al., 2012), seasonal water regimes (Francis, 1986), or both.
Special anatomical adaptations to a waterlogged environment are expressed in the architecture of the preserved gymnosperm axes. Vertebraria roots contain large schizogenous chambers (Figs. 5A, B, 6B, 7C-E) that possibly functioned to aid respiration in the anoxic or dysoxic peat environment (Retallack and Dilcher, 1988;Neish et al., 1993;Decombeix et al., 2009). Vertebraria is very characteristic of pallid to dark palaeosols immediately underlying coal seams throughout Gondwana. These soils developed in dysoxic waterlogged environments and commonly host great densities of horizontal and low-angle root (Vertebraria) systems (Schopf, 1982;McLoughlin, 1993;Slater et al., 2012). Australoxylon mondii also has notable gaps between the ray cells that may have functioned as an aeration system in the lower stem wood of the glossopterid plant (Weaver et al., 1997). Dense horizontally extensive Vertebraria root mats may also have helped to stabilise the glossopterid trees in poorly cohesive waterlogged soils by intermeshing with the roots of neighbouring trees, as occurs in extant plants that inhabit boggy soils, e.g. Kahikatea (Dacrycarpus dacrydioides) of New Zealand (Wardle, 1974;Wardle, 1991).
Other adaptations to moist or semi-aquatic environments amongst the PCM plants include the elaborately ornamented surface of the most abundant megaspore species, Singhisporites hystrix (Figs. 6I, 7H, I). Its furcate spines may have functioned to aid dispersal through hydrochory in saturated environments and to facilitate entrapment of conspecific microspores in water (Tewari et al., 2009;Slater et al., 2011). The weakly spinose microspores associated with S. hystrix can be seen to interlock with the complex ornamentation of the megaspore in X-ray synchrotron tomographic images . Lycopsids were typical of wetland habitats or riparian areas from the Devonian to present (Falcon-Lang, 2003).
Modern high-latitude forests occupying the Boreal biome are limited in their distribution primarily by temperature and the proximity of permafrost to the surface, which hampers water and nutrient uptake (Sayre, 1994). The Glossopteris forests of the mid-Permian would have been subject to similar light regimes, but ambient temperatures were probably much higher than at their equivalent modern latitudes (Angiolini et al., 2009). Hence, permafrost development was unlikely to have limited water supply for plant growth.

Fungi and fungi-like organisms
In contrast to the low-diversity macroflora, there is a notably high diversity of fungi and fungi-like organisms in the Toploje Member silicified peat (Figs. 5A, C, 6D-F). This is consistent with models of modern high-latitude forests dominated by deciduous trees (Wu et al., 2011), where the soil can be rich in leaf litter and other plant detritus exploitable by fungi. Examples of fungi in the peat matrix and within dispersed plant organs within the Toploje Member chert include septate and nonseptate hyphae, a range of fungal spores, saprotrophic chytrid fungi, fungal sclerotia, and possible mycorrhizal associations (Figs. 5A, 6D, E, F). Regularly (seasonally) distributed pocket rot (Weaver et al., 1997) occurs within the Australoxylon wood ( Fig. 5C). Appositions are evident in secondary xylem cells of subaerial axes, presumably representing a plant response to fungal invasion (Bhuiyan et al., 2009). Other fungilike microorganisms include two species of Peronosporomycetes or 'water moulds' represented by distinctive furcate spinose oogonia, Combresomyces caespitosus and Combresomyces rarus (Slater et al., 2013). These were likely saprotrophic rather than parasitic forms, and such organisms are very common in modern moist terrestrial habitats (Jobard et al., 2010).

Plant-animal-fungal interactions
The Toploje Member chert is one of only a few Palaeozoic-Mesozoic deposits from which a detailed survey of the arthropod-plant interactions has been undertaken Slater et al., 2012) and one of only a small number of assemblages across Gondwana for which multiple forms of herbivory on Permian plants has been documented ( Fig. 1; Appendix 1). The majority of past records are examples of folivory traces on the margins of Glossopteris leaves but a few records reveal attack on other parts of the Glossopteris plant and on other gymnosperms and pteridophytes in the flora (Appendix 1). Most of these plant-animal interactions have been documented from compressionimpression floras (Appendix 1). Other Palaeozoic and Mesozoic assemblages from which invertebrate-plant relationships and coprolite suites have been extensively documented include those from the Silurian and Lower Devonian deposits in the Welsh Borderland (Edwards, 1996;Edwards et al., 2012), the Rhynie Chert (Habgood et al., 2004), the Pennsylvanian coal balls of the United States (Baxendale, 1979), the 'component community' of invertebrate feeders on the Palaeozoic tree fern Psaronius (Rößler, 2000;D'Rozario et al., 2011aD'Rozario et al., , 2011b, and the Upper Triassic permineralised peat of Hopen in the Svalbard archipelago (Strullu-Derrien et al., 2012). The identification of feeding traces in the roots (Vertebraria), stems (Australoxylon) and leaves (Glossopteris) of the arborescent Antarctic glossopterid whole-plant, integrated with records of feeding traces from other assemblages (see Appendix 1 and references therein), illustrates that all major organs of the glossopterid plant were attacked by a community of invertebrates . Arthropod coprolites inside fern sporangia show that exploitation of plant food sources was not limited to glossopterids. The herbaceous lycophytes of the understorey  may have also provided an important source of food or egg-hosting sites for invertebrates based on the extra-Gondwanan occurrence of Triassic Isoetites bearing oviposition scars on their leaves (Moisan et al., 2012).
Small faecal pellets attributed to oribatid mites are the most abundant coprolites in the Toploje Member silicified peat (Fig. 6C). These small coprolites commonly occur in dense clusters and are dimensionally and morphologically identical to those of extant oribatid mites (Rusek, 1975;Vegter, 1983;Slater et al., 2012). Molecular dating has placed the origin of the oribatid mites at 571 ± 37 million years ago (Schaefer et al., 2010). This predates their first occurrence in the fossil record by a large 'mite gap' of~130-90 ±~40 million years (Schaefer et al., 2010). There must have been significant pre-Devonian radiation and adaptation of oribatid mites since the group was already moderately diverse by the Devonian (Norton et al., 1988) and complex land based food webs already existed in the Silurian (Edwards, 1996), including some mite-like microarthropod coprolites (Hagström and Mehlqvist, 2012). Hence, microcoprolite studies offer the potential to not only track the early history of some arthropod clades in the absence of body fossils, but also to assess patterns of food consumption and guild/trophic complexity.
Alongside being potential early colonisers of the terrestrial environment, oribatid mites and other soil-dwelling microarthropods were likely important drivers in the development of the soil-based ecosystems in which Palaeozoic vascular plants thrived. A diverse and abundant invertebrate soil fauna including oribatid mites is known to be important for macro-vegetation in the recycling of nutrients from decaying plant matter, and for enhancing soil aeration and water retention through increased porosity (Bardgett, 2005). Recent studies of extant oribatid mites have revealed that they play a role in dispersing the spores of ectomycorrhizal fungi within the soil via their faeces and through entrapment of spores in the hairs of their exoskeleton (Lilleskov and Bruns, 2005). The presence of oribatid mites likely also supported a community of soil micro-predators such as pseudoscorpions, which prey upon extant mites and also have a fossil record extending to the Devonian (Shear et al., 1989b). The abundance of oribatid mite coprolites in the Toploje Member chert attests to their key role in recycling organic matter and opening up pathways for fungal decomposers in the plant litter and dead wood of high-latitude Gondwanan coal-forming mires. Modern boreal forests that include a significant proportion of deciduous trees have been shown to house a high diversity of soil organisms comparable even to that of tropical rainforests (Wu et al., 2011). The exceptional preservation in the Toploje Member chert provides an opportunity to assess the invertebrate diversity and trophic guilds in these mires prior to the onset of ecosystem decline that began in the Capitanian, and could aid our understanding of the impact of the end-Palaeozoic floral turnover on the associated terrestrial invertebrate biota (Anderson et al., 1999;Labandeira, 2005;Slater et al., 2012).
The remains of animals are probably more common in silicified plant-bearing deposits than has been generally documented (Smoot and Taylor, 1985), since it is inherently difficult to identify an arthropod body fossil in thin section amongst densely packed plant remains, particularly since plant and arthropod cuticles look similar in thin section (Bartram et al., 1987). Further, distinctive waxy clitellate annelid egg cases have not yet been recorded from any Palaeozoic peats, which is surprising given that these remains are readily recognisable and common in Mesozoic to modern floodbasin deposits (Manum et al., 1991;Jansson et al., 2008;Tosolini and Pole, 2010;Bomfleur et al., 2012). More extensive use of the bulk maceration technique on silicified peats and other organic-rich sediments offers the potential to extract more taxonomically informative components of arthropod exoskeletons (e.g., wings) and annelid egg capsules.
Sparse trace fossils occur in the Permian-Triassic Amery Group fluvial sediments of the Prince Charles Mountains including Planolites trails and short vertical burrows ) that attest to the presence of vermiform invertebrates. Permian-Triassic strata elsewhere in Antarctica host a variety of non-marine arthropod ichnogenera giving clues to the stream and soil entomofauna; e.g., Diplopodichnus and Diplichnites myriapod traces were reported from the Permian-Triassic sediments of the central Transantarctic Mountains, together with an interpreted resting trace of a jumping insect (Briggs et al., 2010).
Tetrapods are currently unknown from Permian strata of the Prince Charles Mountains as either body fossils or ichnofossils . The fossil record of Gondwanan Permian terrestrial vertebrates is poor beyond the confines of the Karoo Basin in South Africa (Rubidge et al., 1995;Lucas, 2004). It is possible that macro-herbivores were relatively sparse in southern high latitudes during the Permian, particularly in forest mire communities. The deciduous habit of the dominant woody plants, coupled with strongly seasonal climatic fluctuations at high latitudes, probably made conditions unfavourable for large populations of herbivores without special physiological adaptations or the capacity to migrate long distances (Collinson and Hammer, 2007).

Energy pathways
Based on the composition of, and interactions between, the fossil biota, we reconstruct the energy pathways in this Middle Permian mire ecosystem (Fig. 8). In the absence of tetrapods, we interpret the highlatitude mire community to have been dominated by a low diversity of woody gymnosperms and sparse understory ferns and lycophytes that supported a broad range of invertebrate herbivores and detritivores. The primary producers were dominantly tree-sized glossopterid gymnosperms together with a significant proportion of Noeggerathiopsis (Cordaitales) of similar stature. Less abundant understorey elements of the flora included ferns and herbaceous lycophytes (Fig. 8). Detritivores were dominantly oribatid mites, but other larger forms of generalistand specialist-feeding arthropods are evidenced by the diverse range of coprolites. Fungi and fungi-like organisms appear to have played key roles in nutrient cycling through saprotrophy, and possibly via mycorrhizal interactions with the plants. The Glossopteris trees themselves appear to have been the keystone species within the ecosystem, since many groups of arthropods and fungal saprotrophs fed directly or indirectly on both the living and decaying tissues of the Glossopteris plant. Based on the high-latitude setting and evidence of pronounced annual growth increments in the gymnosperms, the energy flux and consumer activity in this ecosystem likely varied greatly on a seasonal basis. Abiotic factors that influenced this ecosystem included consistently saturated substrates (either high rainfall or groundwater supply) and regular fire events as evidenced by significant quantities of charcoal commonly occurring in distinct bands within the peats (Fig. 8).

Accumulation model
The plant organs found in the peat lack signs of regular abrasion and are preserved relatively intact. The organic components of the peat also lack either imbrication or systematic sorting with the exception of a few discontinuous centimetre-scale bands that are enriched in macroscopic charcoal. The delicate structures preserved on some organs, such as fine trichomes in the stomatal grooves of Noeggerathiopsis leaves (Fig. 5E), hirsute leaf margins (Fig. 6H), stomatal guard cells (Fig. 7F), and spines on lycophyte megaspores and on the oogonia of Peronosporomycetes (McLoughlin and Drinnan, 1996;Slater et al., 2013) suggest that the bulk of the detached plant remains had not been transported any significant distance before incorporation into the peat. Some upright stumps and intertwined Vertebraria roots in the peat profile denote that at least some glossopterid remains were preserved in growth position. This contrasts with some Late Permian silicified peats from the central Transantarctic Mountains that may represent rafted peat mats preserved out of context of the original depositional environment (Taylor et al., 1989). Rafting of peat loads by rivers and in lakes and by the action of ice is well known in modern environments (Argow et al., 2011), some organic mats, tussocks and floating islands of vegetation even being rafted across large marine water bodies (Houle, 1998). Floatation of peats has even been proposed as a mechanism for the accumulation of organic matter forming the lower Maastrichtian coals of the eastern Pyrenees of Catalonia, Spain (Villalba-Breva et al., 2012). The peats preserved in the Toploje Member chert are the silicified remains of an in situ mire, since they are laterally extensive (persisting over 3 km of outcrop) and preserve few siliciclastic grains Slater et al., 2012).

Silicification and compaction
The silicification process in the uppermost Toploje Member appears to have occurred before any significant compression of the peats, since delicate and even hollow structures, such as small seeds and megaspores, are preserved in their original three-dimensional form . Soft tissues such as phloem are locally preserved, which suggests a rapid silicification process and inhibition of bacterial degradation. The strongly acidic nature of the original waterlogged peat profile probably suppressed decay prior to silicification. Pyrite is common through the peat matrix and occurs primarily as small crystals that appear to have developed contemporaneously with silicification (Fig. 5G). For the most part, these crystals do not impact on the quality of preservation of the organic matter and do not show preferential nucleation on any particular plant tissues. Some larger cavities within the peat (e.g., voids within Vertebraria roots) show weakly defined concentric infilling by cryptocrystalline silica but little textural or compositional differences between these chalcedony/moganite layers is apparent. Hand specimens of the chert show very few cross-cutting mineral veins. These factors suggest that the bulk of the chert's silica was introduced and precipitated in a short interval rather than in multiple phases over a longer time period, during which a greater range of textural and compositional variation would be expected in the precipitates.
The primary source of silica in the chert remains unresolved. No volcanic ash beds or other volcanogenic sediments are preserved in the Bainmedart Coal Measures, in contrast to silicified peat occurrences in other Permian strata of eastern Australia and the Transantarctic Mountains (Gould and Delevoryas, 1977;Taylor et al., 1989;McLoughlin, 1992;Pigg and McLoughlin, 1997). Further, no evidence of strong lateral or vertical textural or vegetational gradients within the silicified ecosystem is evident in contrast to typical sinter deposits (Trewin, 1994(Trewin, , 1996Walter et al., 1998;Trewin et al., 2003;Channing and Edwards, 2004;Guido et al., 2010). The stratigraphic position of the chert bed immediately below the lacustrine sideritic shales of the Dragons Teeth Member, suggests that drowning of the peat surface by mineral-charged lake waters was key to the entombment process. Mineral-charged springs emanating from basin-margin faults offer one potential source of silica, with precipitation around the organic matter of the drowned peat potentially being facilitated by strongly alkaline conditions in the highpalaeolatitude lake. The precursor to chert formation, silica gels, have been reported from some modern lacustrine settings, particularly in brine rich/saline playa lakes and also lagoonal environments (Peterson and Von Der Borch, 1965;Colinvaux and Goodman, 1971;Wheeler and Textoris, 1978;Wells, 1983). Such silica gels form in the sediments of modern lake beds when silica, derived from quartz or other sources, is dissolved into the lake water due to highly alkaline conditions and then precipitates out of solution when the pH drops sharply due to increased acidity because of the decay of plant matter in the lake (Hesse, 1989).

Wildfire
Abundant micro-and macro-charcoal is evident in some hand specimens and bulk-macerate residues of the Toploje Member chert (Fig. 3). Fire plays a significant role in shaping the structure, diversity and ecological succession of many modern plant communities (Crutzen and Goldammer, 1993). Wildfire has been a major factor in the environment since land vegetation produced a highly oxygenated atmosphere in the Carboniferous (Scott, 1989(Scott, , 2000. Even earlier evidence of fire in the form of sparse but dispersed fusain and a few identifiable charcoalified plant organs extends back to the dawn of terrestrial vascular plant communities (Scott and Glasspool, 2006). Jasper et al. (2013) argued that fire was a consistent feature of the Gondwanan Permian landscape and may have played an important role in vegetation overturn across the supercontinent.
Besides preserving a record of the distribution of wildfires through time, charcoalification can aid plant preservation by transforming the cell walls of the plant material into almost pure carbon, which hinders the decay process and retains anatomical detail. Such charcoalification of plant tissues occurs in wildfires with a specific temperature range (240-370°C), providing insights into the style of wildfire propagation in palaeoecosystems (Scott and Jones, 1991;. The common presence of charcoal in the silicified peats of the PCMs (measured at 1.2% organic volume by Holdgate et al., 2005, but locally in much higher proportions in this study: Fig. 3B) means that wildfires were a frequent feature of the Middle Permian high-latitude mire environment.
The charcoalification of the PCMs plant material is also of importance taphonomically (Scott, 2000). Organic remains that have been charcoalified in the Toploje Member chert include macroscopic fragments of Australoxlyon wood, gymnosperm seeds (Fig. 6A), Glossopteris leaf midribs, Vertebraria axes ( Fig. 7C-E), lycophyte axis fragments and microphylls, coprolites (Fig. 5D) and arthropod cuticle ( Fig. 6G; Weaver et al., 1997;Holdgate et al., 2005). Some of these organs (e.g. seeds and lycophyte microphylls) are otherwise ill-preserved in the peats suggesting a taphonomic bias against their preservation in the absence of wildfire charring. The greater cohesion and particle strength endowed by the charcoalification process aids the recovery of anatomical information from delicate plant tissues. Remains of charcoalified Australoxlyon wood locally exceed 5 cm in diameter . The concentration of charcoal in bands within some chert samples (Figs. 3B, 7G) may be related to aqueous sorting of particles or accumulation of charcoal in 'fire-pockets' on the peat surface in the aftermath of burning events (Staub and Cohen, 1979).

Depositional setting
Glossopterid gymnosperms have been documented from a wide range of terrestrial settings (Cúneo et al., 1993;McLoughlin, 1993). They ranged from periglacial to warm and intermittently dry temperate climates (Chandra, 1992), and are preserved in a diverse  array of deltaic, alluvial valley, lake-and lagoon-margin, and potentially even upland settings (Cúneo et al., 1993;McLoughlin, 1993;Rigby, 1993;Guerra-Sommer et al., 2008). Indeed the ubiquitous occurrence of Glossopteris leaves in Permian continental deposits of Gondwana suggests that they occupied most parts of the fluvial and deltaic landscapes from levees to floodbasin mires and welldrained valley fringes, but detailed quantitative studies of the partitioning of plants within alluvial systems of Gondwana have yet to be undertaken.
In the context of the Permian deposits of the Prince Charles Mountains, the Toploje Member was deposited in a narrow but very long (N 1000 km) graben complex (the Lambert Graben) that developed as part of an incipient rifting phase in central Gondwana (Harrowfield et al., 2005). Deposition of basin-wide peats occurred in alternation with pluvial pulses depositing extensive blankets of channel sandstones within braided river systems (Fielding and Webb, 1996). We envisage glossopterids occupying predominantly fluvial levees and raised floobasin mires, but also extending onto wetter parts of the valley flanks (Fielding and Webb, 1996;Holdgate et al., 2005).

Rheotrophic vs ombrotrophic mire
The peats preserved in the Toploje Member chert are interpreted to represent the product of an ombrotrophic mire as opposed to a rheotrophic mire . Ombrotrophic mires obtain their water and nutrient supply through high levels of precipitation, whereas rheotrophic (or minerotrophic) mires are supplied with water by streams or springs (Diessel, 1992). Rheotrophic mires receive a higher nutrient supply and are characteristically richer in inorganic content than ombrotrophic mires (Diessel, 1992). The paucity of siliciclastic grains coupled with the abundance of fungi and charcoal (and especially of charcoalified roots) suggests the peat layer was the product of an ombrotrophic mire , since such deposits are raised above the regional water table and are subject to greater aerobic decay and sporadic burning than rheotrophic systems. Ombrotrophic mires generally also produce thicker organic accumulations than rheotrophic systems and this is consistent with the presence of some seams reaching 11 m thick within the Bainmedart Coal Measures (McLoughlin and Drinnan, 1997a). Alternatively, there is the possibility that some of the thicker coal seams may be the product of stacked mires containing several episodes of peat formation. Stacked mires could potentially be identified through the recognition of vertical changes in palynological content and coal petrography of individual seams (Jerrett et al., 2011).
Euramerican late Palaeozoic coals that developed under palaeotropical climates tend to contain high percentages of vitrinite macerals (Mackowsky, 1975 ; Fig. 9)a consequence of enhanced gelification of organic constituents. High-palaeolatitude Gondwanan coals, especially those of Permian age, are renowned for their high inertinite content (Mackowsky, 1975), although this is not always the case (Diessel and Smyth, 1995). High inertinite concentrations are considered characteristic of ombrogenous peats, in which the upper part of the organic profile is subject to oxidation and regular burning due to a fluctuating water table (Moore, 1989;Teichmüller, 1989). The high inertinite levels in most coal samples from the Bainmedart Coal Measures are consistent with proportions represented in other Gondwanan Permian (glossopterid-derived) coals (Fig. 9). Further, the low sulphur content (0.38 to 0.81 wt.%: Holdgate et al., 2005) and dominance of inertodetrinite amongst the inertinite maceral component is consistent with the interpretation of the Lambert Graben Permian peat-forming environments as raised forest mires of alluvial valley settings subject to strongly fluctuating water tables (Tie and Esterle, 1991) and experiencing fine degradation and oxidation of plant components.

Structure (and reconstruction) of a glossopterid mire
Reconstructions of the food web, trophic levels and soil ecology of ancient forest ecosystems have been attempted in several studies of exceptionally preserved fossil assemblages (Habgood et al., 2004;Adl et al., 2010). Our reconstruction of the Glossopteris-dominated forest mire within the Lambert Graben valley system (Fig. 10) features the dominance of hygrophilous arborescent glossopterid gymnosperms and subsidiary cordaitaleans. Understorey vegetation within the glossopterid forests is represented by herbaceous lycophytes and ferns (Fig. 7A). Sphenophytes, more typical of the lacustrine deposits of the Dragons Teeth Member, are envisaged to have occupied more open, lakemargin or disturbed settings similar to the habitats occupied by modern Equisetum. Abundant deadwood, hosting a range of fungal saprotrophs, and widespread waterlogged soil habitats would have characterised the landscape.
The topology of arborescent plants at higher latitudes tends towards tall tapered forms in order to maximise the collection of light for photosynthesis from low-angle solar radiation (Creber and Chaloner, 1985;McLoughlin, 2011b); thus we reconstruct the dominant trees to have a conical canopy shape (Fig. 10). Relatively open woodlands also characterise modern high-latitude woody vegetation, especially those areas close to the temperature-controlled tree-line (Doležal et al., 2006). Hence, we interpret these glossopterid communities to represent open deciduous forests and woodlands sensu many Carboniferous T 0 assemblages (fossil assemblages representing a geological instant), which reveal an open canopy distribution of large plants (Gastaldo et al., 2004;DiMichele et al., 2009;DiMichele and Falcon-Lang, 2011) rather than the closed canopy system seen in many angiosperm forests.
The extensive portrayal of wetlands in the reconstruction (Fig. 10) highlights year-round high moisture levels and is consistent with the broad representation of hygrophilous elements in the flora and the abundance of moisture-loving micro-organisms (e.g., Peronosporomycetes). The distal part of the reconstruction (Fig. 10) portrays features representative of the sedimentology of the lower Bainmedart Coal Measures. Sedimentary facies associations of the Toploje Member are represented by thick sandstone packages deposited in braided river channels, alternating with extensive floobasin mudrocks and coals (Fielding and Webb, 1996;McLoughlin and Drinnan, 1997a) within an alluvial valley setting flanked by fault-delineated valley margins of considerable relief that were developed in Precambrian crystalline rocks (Boger, 2011).

Implications for coal geology
There is a notable contrast between some of the coal petrographic results (Fig. 4) and the composition of the silicified peat obtained via point counts of transects through various permineralised blocks (Fig. 3A). In general, the peat appears to contain much lower proportions of obviously oxidised components (charcoal) and waxy material (spores and cuticle) and higher proportions of non-oxidised woody and leaf mesophyll components than is registered in the coal.
Charcoalified (oxidised) material typically has greater compactional resistance than other plant remains, hence its relative volumetric representation probably increases through the early stages of coalification. However, the very high levels of inertinite (up to 50%) in some coal samples are greater than would be expected from differential compaction alone. It is possible that the small selection of coal samples examined were preferentially enriched in charcoal through local fire events. Such bands in the permineralized peat show comparable levels of N50% charcoal (Fig. 3B). Further, the preservation of the silicified peat differs from other organic accumulations in the Bainmedart Coal Measures in that the peat surface was drowned rapidly by lake waters that saw deposition of the overlying Dragons Teeth Member. Hence, the upper part of this peat mound may not have developed to typical maturity or have been subjected to the prolonged oxidation and degradation by fungi, microbes and fires that was experienced by other raised peat (coal-forming) deposits in this formation. We consider that this hypothesis best explains the differential representation of oxidised components.
The organic matter in the silicified peat has undergone relatively little compression before mineral entombment, whereas the associated coals of sub-bituminous to high-volatile bituminous rank  have probably experienced N 80% loss of volume through compaction, gelification and diagenesis (Teichmüller and Teichmüller, 1982). Apart from the loss of pore space, much of this volume reduction was probably accommodated by compaction, loss of volatiles and mobilisation of organic gels from unoxidised wood and leaf mesophyll cellulosic cell walls. We suspect that this accounts for the high proportion of woody and leafy tissues in the silicified peat compared to the relatively modest levels of vitrinite in the coals.
The high proportion of liptinite macerals in the Bainmedart Coal Measure coals also appears anomalous with respect to the low volumetric proportion of spores, pollen and sporangia (b2%) in the permineralised peat . Low degrees of compaction for these dense waxy materials may account for somewhat higher percentages in the associated coals but it is unlikely to account for the very high levels of liptinite (up to 40%) in some coal samples. Holdgate et al. (2005) suggested several other factors that might account for the anomalously high spore content of the Lambert Graben coals in contrast to the silicified peat. These included potential misidentification of small cutinite fragments in the coal as sporinite, and the possibility that a considerable proportion of the finely dispersed unidentifiable organic matter in the peat represents unrecognised sporopollenin or degraded cuticle material. Cameron et al. (1989) noted that the proportion of liptinite commonly increases upwards in domed (ombrogenous) peats together with oxidised detrital matter. As per the explanation for depleted oxidised components, the abrupt termination of peat development at the top of the Toploje Member via flooding may indicate that the silicified peat capping this unit did not reach the optimal domed stage of other coals in the host formation. In general, high liptinite contents, in the absence of alginite, might also be diagnostic of coals derived from high-latitude alluvial valley forest mires dominated by low-diversity deciduous gymnosperm vegetation with high pollen production. In this respect, it is notable that Permian coals from the continentalinterior, alluvial valley deposits of the Godavari-Son-Mahanadi Graben system in India also have relatively high liptinite contents and have the most similar proportions of maceral groups to the Lambert Graben coals (Navale and Saxena, 1989;Mishra, 1996;Fig. 9).
Quantitative analyses of the permineralised and coalified organic accumulations in the Bainmedart Coal Measures indicate that although silicified peats provide excellent details of the botanical constituents of coal-forming deposits, they do not always reflect the ultimate volumetric representation of macerals in the coals. In addition to differences in the oxidation state at the time of permineralization of silicified peats versus non-mineralised peats, significant losses of pore space, volatile components and mobilisation of organic gels during the peatification and coalification processes may markedly modify the volumetric proportions of some phytoclasts. Furthermore, permineralised peats, coal balls, and coal itself may not be fully representative of the standing biomass of the peat-forming community, since peats tend to be enriched in root material compared to subaerial parts of the vegetation (Raymond, 1987;DiMichele and Phillips, 1994).

Integrated discussion
The permineralised community represented in the Toploje Member chert shares many similarities with the Permian and Triassic peats of the Transantarctic Mountains that occur at marginally higher palaeolatitudes than those of the Prince Charles Mountains. The Permian palaeocommunity in the Transantarctic Mountains is also dominated by glossopterids, evidenced by a range of plant vegetative organs and megasporophyll and microsporophyll genera (Ryberg et al., 2012a), but differs in the absence of Noeggerathiopsis, which is common in the Toploje Member peats. Other notable shared features with the permineralized floras of the Transantarctic Mountains include the presence of herbaceous lycophytes in the Permian (Schwendemann et al., 2010;Ryberg et al., 2012b) and diverse fungal elements in both the Permian and Triassic peats (Stubblefield and Taylor, 1986;Krings et al., 2012).
Vertebraria roots were found to be the dominant component of the Toploje Member peats alongside significant quantities of matted Glossopteris and Noeggerathiopsis leaves (Fig. 3A), whilst charcoalified wood is locally dominant in particular peat microfacies (Fig. 3B). The paucity of arboreal components of the plants such as stem wood and fruiting bodies in comparison to the dominance of root tissues leads us to interpret that there was a taphonomic bias against incorporation of at least some subaerial organs into the peat profile. Similar proportions of floral elements were found in quantitative analysis of coal ball vegetation from Late Carboniferous wetland floras of Pennsylvania, USA, where root tissue was also found to be the dominant component of the preserved assemblage (Feng, 1989).
Lower-latitude Gondwanan permineralised assemblages, such as those from the Late Permian Fort Cooper Coal Measures of the Bowen Basin, Australia (Gould and Delevoryas, 1977;Nishida et al., 2007) are characterised by slightly higher macrofloral diversity, although those floras have yet to be fully described. Outside the Gondwanan phytogeographic province, silicified Permian terrestrial communities differ in their macrofloral composition; however many of the key microorganism groups are represented and invertebrate-plant interactions appear to have been established in a similar ecological structure throughout the late Palaeozoic world (Baxendale, 1979;Labandeira, 1998Labandeira, , 2013Rößler, 2000).
Beyond the Permian, Triassic silicified peat from Hopen Island in the Svalbard Archipelago and deposited in northern high middle latitudes contains many of the soil microbe groups preserved in the Toploje Member chert (Strullu-Derrien et al., 2012); these include a diverse range of coprolites (Figs. 5D, 6C, 7B) indicating a rich soil entomofauna and a range of fungal morphotypes and interactions (Figs. 5A, 6D-F). Equivalent microbial communities, though associated with a more diverse macroflora, are evident in Late Triassic permineralised peats from the Transantarctic Mountains (Schwendemann et al., 2009;Krings et al., 2012).
Similar communities of microorganisms are also present in the Upper Pennsylvanian Grand-Croix cherts of France, including a diverse fungal inventory and examples of Peronosporomycetes . The Glossopteris component community of the Toploje Member chert also closely resembles the structure and resource partitioning evident in the Psaronius component community preserved in the early Late Pennsylvanian coal balls of the Illinois Basin (Labandeira, 1998), the earliest Permian Chemnitz fossil forest (Rößler, 2000), and the Permian Cathaysian coal measures of southwest China (He et al., 2008;D'Rozario et al., 2011a) with exploitation of every part of the dominant host plant by invertebrate herbivores and detritivores .
The relatively low diversity of the Prince Charles Mountains macroflora compared to palaeotropical Euramerican late Palaeozoic peat assemblages shows that, as is the case today, there was a latitudinal diversity gradient in the Permian. Even within Gondwana, greater floristic diversity at generic level is evident in adpression assemblages from lower palaeolatitudes of South America and southern Africa (Archangelsky and Arrondo, 1969;Anderson and Anderson, 1985) compared to higher palaeolatitude regions of eastern Australia and Antarctica (McLoughlin, 1992(McLoughlin, , 1994a(McLoughlin, , 1994bCúneo et al., 1993). The more pronounced seasonal light regime and presumably colder winters may have inhibited many (especially Euramerican) taxa from colonising the highest latitudes of Gondwana during the Permian.
The absence of complete body fossils of aquatic crustaceans and other arthropods may be due to the speed of silicification and the environment the assemblages accumulated in. Unlike the Rhynie Chert or the ex situ Yorkshire cobble, the mode of entombment in the Toploje Member chert does not appear to be related to very rapid hydrothermal precipitation of silica that might have killed and preserved free-moving arthropods in their life positions (Guidry and Chafetz, 2003;Anderson et al., 2004;Womack et al., 2012). Instead, the Toploje Member peat represents a long-lived acidic mire environment, into which silica was subsequently introduced in high concentrations associated with flooding of the mire surface. Any arthropod remains would likely be represented by dissociated exoskeletons and exuviae from the moulting of ecdysozoans that were incorporated into the peat or trapped within excavations inside woody tissues. Fragments of such arthropod cuticle with attached setae were recovered by bulk maceration of the Toploje Member chert Slater et al., 2012) and are also known from overlying Triassic sediments . These fragments likely represent arthropod cuticle because the setae are hollow and are collapsed or flattened in places, whereas similar setae of annelid origin are solid (Orrhage, 1971). The setae may have served a mechanosensory function in life (Crouau, 1997;Keil, 2012). The cuticles are unlikely to be of collembolan affinity since the exoskeletons of that group are covered in distinctive granule structures of antiadhesive function (Nickerl et al., 2013). Beyond this, the arthropod fragments cannot be identified.
Relatively little attention has been directed towards arthropod cuticles extracted from coals, silicified peats and coal balls (Bartram et al., 1987), perhaps in part because these sediments are generally studied for their palaeobotanical content. The mode of preservation of such arthropod cuticles in silicified deposits and coals has been the subject of some debate (e.g. Stankiewicz et al., 1998;Appendix 2). Examples of scorpion exoskeleton have been described from Pennsylvanian (Westphalian) coals of Yorkshire, UK (e.g. Bartram et al., 1987) and the Late Triassic Lower Keuper Sandstone of Bromsgrove, Worcestershire, UK (Dunlop et al., 2007). Cuticles of possible eurypterid affinity are also known from the Lower Devonian of Podolia, Ukraine (Filipiak et al., 2012) and the body of a trigonotarbid was also recovered via HF maceration from Přídolí shales of Shropshire, UK (Dunlop, 1996). Arthropod cuticles have also been recovered through HF maceration of early Silurian (Llandovery) terrestrial strata from Pennsylvania, providing some of the earliest direct evidence for land or freshwater animals (Gray and Boucot, 1994). Records of such arthropod cuticle and other invertebrate remains have been recovered from acid maceration of various sediments, primarily coals, clays, siltstones and cherts (Appendix 2). Exceptionally preserved Palaeozoic arthropods are also known from siderite concretions, particularly from Carboniferous sites such as Coseley, UK (Garwood and Sutton, 2010), Mazon Creek, Illinois (Carpenter, 1997) and the Montceau Lagerstätte, France (Garwood et al., 2012). Arthropods preserved in siderite concretions have increasingly been studied using X-ray synchrotron microtomography, which unveils exceptional detail in the fossil (Garwood et al., 2012). However, the maceration of silicified peats and scanning electron microscopy of residues could potentially yield comparable anatomical information and also be used to more accurately assess diversity through the comparison of different cuticle types in coals and other organic-rich lithologies. Several studies have also used the acetate peel technique to study the cuticle of arthropods such as eurypterids and scorpions in detail (Braun, 1999;Tetlie et al., 2008). The potential of such fragmentary arthropod cuticle remains has recently been highlighted by employing a long-used palaeobotanical technique to extract animal fossils from Cambrian marine sediments (Butterfield and Harvey, 2012;Harvey et al., 2012aHarvey et al., , 2012b and this strategy might offer comparable results in unveiling cryptic arthropod diversity in late Palaeozoic terrestrial ecosystems.

Conclusions
The Toploje Member chert in the Bainmedart Coal Measures of the Lambert Graben, East Antarctica, contains a low-diversity Middle Permian macroflora dominated by arborescent glossopterids and subordinate cordaitaleans. A substantially greater floristic diversity is evident in meso-and palyno-fossil assemblages from the same bed. The chert preserves part of the profile of an ombrotrophic mire that was entombed with moderate rapidity by silica precipitated after flooding of the peat surface by mineral-rich lake waters. Diverse fossil micro-organisms and coprolites, together with feeding traces and sparse exoskeleton fragments attest to a complex biota of primary producers, arthropod herbivores and saprotrophs inhabiting the peat-forming ecosystem. The peat deposit represents primarily autochthonous to parautochthonous accumulation of plant remains, which due to early silicification, have endured little compression or diagenesis. Fire was a common factor in the Middle Permian high-latitude mire ecosystem, and is locally evidenced by discrete charcoal bands within the peat profile. The glossopteriddominated mire ecosystem of the Prince Charles Mountains is reconstructed to occupy valley flanks and floodbasin settings within a braided river complex in a fault-bound alluvial valley.
The exceptional ultrastructural preservation of the macroflora, the diverse microorganisms and the complex interactions between the invertebrates, plants and fungi make the Prince Charles Mountains silicified peat an important Permian Lagerstätte and a valuable source of information on terrestrial ecosystems in southern high latitudes during the late Palaeozoic. The Toploje Member chert provides a snapshot of a high-latitude Middle Permian terrestrial ecosystem and, thus, elucidates the composition and ecology of Gondwanan climax mire forests prior to the two major extinctions (Capitanian and end-Permian) at the close of the Palaeozoic (Retallack et al., 2006;Bond et al., 2010). Since the plants, microorganisms and fauna preserved in the Toploje Member chert inhabited an ombrotrophic mire rather than a sinter-pool habitat or other mineral-charged spring setting, they provide the opportunity to study the ecology of the more widespread peat-forming biotas of the late Palaeozoic in contrast to the specialised thermophilic biotas preserved three-dimensionally in hot spring habitats. In light of the discovery of invertebrate-plant-fungal interactions in the Toploje Member chert, there are significant opportunities for advancing knowledge of ancient terrestrial arthropod and micro-organism occurrences and trophic relationships through studies of other silicified organic deposits and via bulk maceration of organic sediments.