Abstract Patterns of plant distribution by palaeoenvironment were examined across the Pennsylvanian–Permian transition in North–Central Texas. Stratigraphically recurrent packages of distinct lithofacies, representing different habitats, contain qualitatively and quantitatively different macrofloras and microfloras. The species pools demonstrate niche conservatism, remaining closely tied to specific habitats, during both short-term cyclic environmental change and a long-term trend of increasing aridity. The deposits examined principally comprise the terrestrial Markley and its approximate marine equivalent, the Harpersville Formation and parts of lower Archer City Formation. Fossiliferous deposits are lens-like, likely representing fill sequences of channels formed during abandonment phases. Palaeosols, represented by blocky mudstones, comprise a large fraction of the deposits. They suggest progressive climate change from minimally seasonal humid to seasonal subhumid to seasonal dry subhumid. Five lithofacies yielded plants: kaolinite-dominated siltstone, organic shale, mudstone beds within organic shale, coarsening upward mudstone–sandstone interbeds and channel sandstone. Both macro- and microflora were examined. Lithofacies proved compositionally distinct, with different patterns of dominance diversity. Organic shales (swamp deposits), mudstone partings (swamp drainages) and coarsening upward mudstone–sandstone interbeds (floodplains) typically contain Pennsylvanian wetland vegetation. Kaolinite-dominated siltstones and (to the extent known) sandstones contain taxa indicative of seasonally dry substrates. Some kaolinite-dominated siltstones and organic shales/coals yielded palynomorphs. Microfloras are more diverse, with greater wetland–dryland overlap than macrofloras. It appears that these two floras were coexistent at times on the regional landscape.

Late Carboniferous (Pennsylvanian) fossil floras from the United States are well studied as adpression, permineralization, and palynomorph assemblages throughout the stratigraphic column. These data represent an intrabiomic record that can serve as a proxy for climate change in the Carboniferous tropics. The short-term climatic changes that accompanied the alternations between glacial and interglacial intervals did not alter the persistence of the ecological structure of the landscape. Even after floras had been extirpated over large parts of the North American continent in response to marine transgressions, the same plants and plant communities repeatedly returned when the sea receded. However, at the Westphalian-Stephanian boundary (approximately Desmoinesian-Missourian; Moscovian-Kasimovian boundary), major vegetational changes occurred that suggest a significant environmental threshold had been exceeded. Entire clades (most tree lycopsids and medullosans with very large seeds) became extinct, and tree ferns became dominant, changing the aspect of the ecological landscape. This change reflects the overall warming of Earth’s climate, greater seasonality, and shorter periods of wet conditions in the tropics of the late Pennsylvanian.

The fossil record of wetlands documents unique and long-persistent floras and faunas with wetland habitats spawning or at least preserving novel evolutionary characteristics and, at other times, acting as refugia. In addition, there has been an evolution of wetland types since their appearance in the Paleozoic. The first land plants, beginning in the Late Ordovician or Early Silurian, were obligate dwellers of wet substrates. As land plants evolved and diversified, different wetland types began to appear. The first marshes developed in the mid-Devonian, and forest swamps originated in the Late Devonian. Adaptations to low-oxygen, low-nutrient conditions allowed for the evolution of fens (peat marshes) and forest mires (peat forests) in the Late Devonian. The differentiation of wetland habitats created varied niches that influenced the terrestrialization of arthropods in the Silurian and the terrestrialization of tetrapods in the Devonian (and later), and dramatically altered the way sedimentological, hydrological, and various biogeochemical cycles operated globally. Widespread peatlands evolved in the Carboniferous, with the earliest ombrotrophic tropical mires arising by the early Late Carboniferous. Carboniferous wetland-plant communities were complex, and although the taxonomic composition of these wetlands was vastly different from those of the Mesozoic and Cenozoic, these communities were essentially structurally, and probably dynamically, modern. By the Late Permian, the spread of the Glossopteris flora and its adaptations to more temperate or cooler climates allowed the development of mires at higher latitudes, where peats are most common today. Although widespread at the end of the Paleozoic, peat-forming wetlands virtually disappeared following the end-Permian extinction. The initial associations of crocodylomorphs, mammals, and birds with wetlands are well recorded in the Mesozoic. The radiation of Isoetales in the Early Triassic may have included a submerged lifestyle and hence, the expansion of aquatic wetlands. The evolution of heterosporous ferns introduced a floating vascular habit to aquatic wetlands. The evolution of angiosperms in the Cretaceous led to further expansion of aquatic species and the first true mangroves. Increasing diversification of angiosperms in the Tertiary led to increased floral partitioning in wetlands and a wide variety of specialized wetland subcommunities. During the Tertiary, the spread of grasses, rushes, and sedges into wetlands allowed for the evolution of freshwater and salt-water reed marshes. Additionally, the spread of Sphagnum sp. in the Cenozoic allowed bryophytes, an ancient wetland clade, to dominate high-latitude mires, creating some of the most widespread mires of all time. Recognition of the evolution of wetland types and inherent framework positions and niches of both the flora and fauna is critical to understanding both the evolution of wetland functions and food webs and the paleoecology of surrounding ecotones, and is necessary if meaningful analogues are to be made with extant wetland habitats.

Early Silurian (Llandoverian) macrofossils from the lower Massanutten Sandstone at Passage Creek in Virginia represent the oldest known terrestrial wetland communities. Fossils are preserved as compressions in overbank deposits of a braided fluvial system. Specimens with entire margins and specimens forming extensive crusts provide evidence for in situ preservation, whereas pre-burial cracks in the fossils demonstrate subaerial exposure. Developed in river flood plains that provided the wettest available environments on land at the time, these communities occupied settings similar to present-day riverine wetlands. Compared with the latter, which are continuously wet by virtue of the moisture retention capabilities of soils and vegetation, Early Silurian flood-plain wetlands were principally abiotically wet, depending on climate and fluctuations of the rivers for moisture supply. Varying in size from <1 cm to >10 cm, fossils exhibit predominantly thalloid morphologies but some are strap-shaped or form crusts. Their abundance indicates that a well-developed terrestrial groundcover was present by the Early Silurian. Morphological and anatomical diversity of specimens suggests that this groundcover consisted of several types of organisms and organismal associations, some characterized by complex internal organization. Earlier microfossil finds at Passage Creek corroborate an image of systematically diverse but structurally simple communities, consisting only of primary producers and decomposers. Ten to fifteen million years older than the oldest previously known complex terrestrial organisms (e.g., Cooksonia ), they provide a new perspective on the early stages of land colonization by complex organisms, whereby the earliest terrestrial communities were built by a guild of thalloid organisms and associations of organisms comparable to extant biological soil crusts.

The Trout Valley Formation of Emsian–Eifelian age in Baxter State Park, Maine, consists of fluvial and coastal deposits that preserve early land plants (embryophytes). Seven facies are recognized and represent deposits of main river channels (Facies 1, 2), flood basin (Facies 4), storm-influenced nearshore shelf bars (Facies 3), a paleosol (Facies 5), and tidal flats and channels (Facies 6, 7). The majority of plant assemblages are preserved in siltstones and are allochthonous and parautochthonous, with only one autochthonous assemblage identified in the sequence above an apparent paleosol horizon. Taphonomic analysis reveals that plant material within allochthonous assemblages is highly fragmented, poorly preserved, and decayed. Plant material within parautochthonous assemblages shows evidence of minimal transport, is well preserved, and shows signs of biologic response after burial. The one autochthonous assemblage contains small root traces. Trimerophytes ( Psilophyton and Pertica quadrifaria ), rhyniophytes (cf. Taeniocrada ), and lycopods ( Drepanophycus and Kaulangiophyton ) are the most common taxa in estuarine environments. Psilophyton taxa, Pertica , cf. Taeniocrada , and Drepanophycus are found also in fluvial settings. The presence of tidal influence in deposits where parautochthonous and autochthonous assemblages occur shows that these plants occupied coastal-estuarine areas. However, the effects on the growth and colonization of plants of the physical conditions (e.g., salinity) that exist in these settings in the Early to Middle Devonian are unknown.

The Late Devonian Red Hill locality in north-central Pennsylvania contains an Archaeopteris -dominated plant fossil assemblage, a diverse fossil fauna, and an extensive sedimentary sequence ideal for investigating the landscapes and biotic associations of the earliest forest ecosystems. Sedimentological analysis of the main plant-fossil bearing layer at Red Hill indicates that it was a flood-plain pond. A seasonal wet-and-dry climate is indicated by well-developed paleovertisols. The presence of charcoal interspersed with plant fossils indicates that fires occurred in this landscape. Fires appear to have primarily affected the fern Rhacophyton . The specificity of the fires, the distribution profile of the plant remains deposited in the pond, and additional taphonomic evidence all support a model of niche partitioning of the Late Devonian landscape by plants at a high taxonomic level. At Red Hill, Archaeopteris was growing on the well-drained areas; Rhacophyton was growing in widespread monotypic stands; cormose lycopsids grew along the pond edge; and gymnosperms and Gillespiea were possibly opportunists following disturbances. Tetrapod fossils have been described from Red Hill—therefore, this paleoecological analysis is the first systematic interpretation of a specific site that reflects the type of wetland environment within which the earliest tetrapods evolved.

The Horton Group (late Famennian to Tournaisian) of Atlantic Canada provides an unusually complete record of Early Mississippian wetland biota. Best known for tetrapod fossils from “Romer's Gap,” this unit also contains numerous horizons with standing vegetation. The taphonomy and taxonomy of Horton Group fossil forests have remained enigmatic because of poor preservation, curious stump cast morphology, and failure to recognize the unusual sedimentary structures formed around standing plants. Four forested horizons within the Horton Group are preserved as cryptic casts and vegetation-induced sedimentary structures formed by the interaction of detrital sediment with in situ plants. Protostigmaria , the lobed base of the arborescent lycopsid Lepidodendropsis , occur as sandstone-filled casts attached to dense root masses. Mudstone-filled hollows formed when a partially entombed plant decayed, leaving a void that was later infilled by muddy sediment. A scratch semi-circle formed where a current bent a small plant, causing it to inscribe concentric grooves into the adjacent muddy substrate. Obstacle marks developed where flood waters excavated erosional scours into sandy sediment surrounding juvenile Lepidodendropsis . These cryptic lycopsid forests had considerably higher densities than their Pennsylvanian counterparts. Vegetation-induced sedimentary structures are abundant in Horton Group strata and could easily be misidentified as purely hydrodynamic or soft-sediment deformation structures without careful analysis. Recognition of these structures in early Paleozoic strata has great potential to expand our knowledge about the distribution of early land plants.

The Fayetteville Formation of northwestern Arkansas (upper Mississippian/middle Chesterian) contains two compression plant fossil assemblages (one in situ) that represent plant communities, and an allochthonous permineralized assemblage recovered from marine strata that represents the landscape. This preservation of spatial ecological subunits (communities) nested within a larger subunit (landscape) provides a snapshot of vegetation patterns within a Late Mississippian clastic swamp. Fifteen whole plants are recognized. Seed ferns are the most speciose group and lycopsids account for most biomass. Seed fern taxa known only as permineralized specimens include one canopy tree ( Megaloxylon ), two understory trees, and five herbaceous layer plants. Two herbaceous layer seed ferns are observed only as compressions. Lycopsids are represented as two canopy trees that are known from both permineralizations and compressions. Archaeocalamites is also known from both permineralizations and compressions but was an understory tree. Ferns are rare and are preserved only as fragments of permineralized rachises from two species. As revealed by the in situ compression assemblage, the two species of lycopsid canopy trees co-occur and they formed communities that occupied ever-wet bottomlands, with Archaeocalamites occupying the understory, and a single species of seed fern comprising the herbaceous layer. Lycopsids do not co-occur with Megaloxylon . Megaloxylon probably formed a second community type in somewhat water-stressed areas of the swamp with an understory of small arborescent seed ferns, some Archaeocalamites , and an herbaceous-layer seed fern. Ferns probably formed a third type of community in disturbed sites.

An outcrop of the Mississippian Hartselle Sandstone in north-central Alabama preserves in situ, erect cormose lycopsids, assigned to Hartsellea dowensis gen. and sp. nov., in association with a low diversity bivalve assemblage dominated by Edmondia . The isoetalean lycopsids are rooted in a silty claystone in which the bivalve assemblage occurs, representing the transition from tidal flat and tidal channel regime into a poorly developed inceptisol. Two paleosols are preserved in the sequence and each is overlain by a fine-grained quartz arenite, responsible for casting aerial stems and cor-mose bases of the entombed plants. The massive quartz arenites are in sharp contact with interpreted O-horizons of the paleosol, and the lower sandstone displays a lobate geometry. The plant assemblages are interpreted as back-barrier marshes, the first unequivocal marshlands in the stratigraphic record, preserved by overwash processes associated with intense storm surges in a Transgressive Systems Tract. A sample suite curated in the National Museum of Natural History, collected by David White at the turn of the last century in the Greenbrier Limestone of West Virginia, preserves rooting structures, leaves and sporophylls, and sporangia and megaspores of H. downensis in a mixed carbonate mud (micrite). The presence of isoetalean lycopsids in both siliciclastic and carbonate peritidal environments within nearshore shelf settings of the Early Carboniferous indicates that adaptation to periodic brackish water, if not tolerance to infrequent fully marine-water inundation during storm surges, had evolved in these marsh plants by the late Paleozoic.

The earliest tetrapods are known from a handful of Upper Devonian and Lower Carboniferous localities in Europe, North America, and Australia. All Upper Devonian sites and virtually all Early Carboniferous faunas are regarded as predominantly aquatic and most occur within, or are associated with, wetland habitats. A new mid-Carboniferous (Elvirian, Namurian A) fossil locality in Kentucky preserves the first tetrapod fauna from the eastern portion of the Illinois Basin. Four distinct facies at the locality have yielded vertebrate material. Diverse faunas have been found in an abandoned channel/oxbow facies and a floodplain/lake facies. The abandoned channel/oxbow facies contains Colosteidae, Embolomeri, Rhizodontida, Dipnoi, Xenacanthiformes, Palaeonisciformes, and Gyracanthidae remains. This assemblage is similar to known Mississippian freshwater and brackish-water faunas, providing further evidence of a cosmopolitan tetrapod province during the Mississippian. A different fauna, rich in tetrapods but lacking fish, is associated with granular carbonate masses, rooting structures, and a paleosol in the floodplain/ lake facies. Isolated and associated tetrapod elements from this facies exhibit morphological adaptations that may suggest a fauna of more highly terrestrial vertebrates than previously known from the North American Mississippian.

Standing lycopsid trees occur at 60 or more horizons within the 1425-m-thick coal-bearing interval of the classic Carboniferous section at Joggins, with one of the most consistently productive intervals occurring between Coals 29 (Fundy seam) and 32 of Logan (1845) . Erect lepidodendrid trees, invariably rooted within an organic-rich substrate, are best preserved when entombed by heterolithic sandstone/mudstone units on the order of 3–4 m thick, inferred to represent the recurring overtopping of distributary channels of similar thickness. The setting of these forests and associated sediments is interpreted as a disturbance-prone interdistributary wetland system. The heterogeneity and disturbance inherent to this dynamic sedimentary environment are in accord with the floral record of the fossil forests and interpretation of the peat-forming wetlands as topogenous, rheotrophic forest swamps. Candidates for the erect, Stigmaria -bearing trees, which range in diameter (dbh) from 25 to 50 cm, are found in prostrate compressions and represent a broad range of ecological preferences amongst the Lycopsida. This record, which is not significantly time averaged, closely parallels the megaspore record from thin peaty soils in which they are rooted, but differs significantly from the miospore record in studies of other, thicker coals. Dominant megaspores are Tuberculatisporites mamillarus and Cystosporites diabolicus , derived from Sigillaria and Diaphorodendron / Lepidodendron respectively. Intervening beds preserve a record of an extra-mire flora composed in the main of seed-bearing pteridosperms and gymnosperms (and ?progymnosperms). Reproductive adaptation to disturbance appears to have played a key role in ecological partitioning of plant communities within these wetlands. Burial of lycopsid trees by onset of heterolithic deposition resulted in the demise of entire forest stands. Disturbancetolerant Calamites regenerated in the episodically accruing sediment around the dead and dying lycopsid stands, a succession identified here as typical of Euramerican fossil forests. Rapid, ongoing subsidence of the basin accommodated the submergence of the fossil forests, and abiotic disturbance inherent to the seasonal climate facilitated their episodic entombment. Disturbance is inferred to have been mediated by short-term (?seasonal) precipitation flux as suggested by the heterolithic strata and in the record of charred lycopsid trees, recording wildfire most probably ignited by lightning. Within this fossil forest interval is found a glimpse of animal life within the wetland ecosystem beyond the confines of the tree hollows, whence the bulk of the terrestrial faunal record of Joggins historically derives.

The Pittsburgh, Redstone, and Sewickley coal beds all occur in the Late Pennsylvanian Pittsburgh Formation of the Monongahela Group in the northern Appalachian Basin. The goal of this study is to compare and contrast the palynology, petrography, and geochemistry of the three coals, specifically with regard to mire formation, and the resulting impacts on coal composition and occurrence. Comparisons between thick (>1.0 m) and thin (<0.3 m) columns of each coal bed are made as well to document any changes that occur between more central and more peripheral areas of the three paleomires. The Pittsburgh coal bed, which is thick (>1m) and continuous over a very large area (over 17,800 km 2 ), consists of a rider coal zone (several benches of coal intercalated with clastic partings) and a main coal. The main coal contains two widespread bone coal, fusain, and carbonaceous shale partings that divide it into three parts: the breast coal at the top, the brick coal in the middle, and the bottom coal at the base. Thymospora thiessenii , a type of tree fern spore, is exceptionally abundant in the Pittsburgh coal and serves to distinguish it palynologically from the Redstone and Sewickley coal beds. Higher percentages of Crassispora kosankei (produced by Sigillaria , a lycopod tree), gymnosperm pollen, and inertinite are found in association with one of the extensive partings, but not in the other. There is little compositional difference between the thin and thick Pittsburgh columns that were analyzed. The Redstone coal bed is co-dominated by tree fern and calamite spores and contains no Thymospora thiessenii . Rather, Laevigatosporites minimus , Punctatisporites minutus , and Punctatisporites parvipunctatus are the most common tree fern representatives in the Redstone coal. Endosporites globiformis , which does not occur in the Pittsburgh coal, is commonly found near the base of the coal bed, and in and around inorganic partings. In this respect, Endosporites mimics the distribution of Crassispora kosankei in the Pittsburgh coal. Small fern spores are also more abundant in the Redstone coal bed than they are in the Pittsburgh coal. Overall, the Redstone coal bed contains more vitrinite, ash, and sulfur than the Pittsburgh coal. The distribution of the Redstone coal is much more podlike, indicating strong paleotopographic control on its development. Compositionally, there are major differences between the thin and thick Redstone columns, with higher amounts of Endosporites globiformis , gymnosperm pollen, inertinite, ash, and sulfur occurring in the thin column. The Sewickley coal bed is palynologically similar to the Redstone coal in that it is co-dominated by tree fern and calamite spores, with elevated percentages of small fern spores. Tree fern species distribution is different, however, with Thymospora thiessenii and T. pseudothiessenii being more prevalent in the Sewickley. The distribution of Crassispora kosankei in the Sewickley coal bed is similar to that in the Pittsburgh coal, i.e., more abundant at the base of the bed and around inorganic partings. By contrast, Endosporites is only rarely seen in the Sewickley coal. The Sewickley is more laterally continuous than the Redstone coal, but not nearly as thick and continuous as the Pittsburgh coal. Overall, the vitrinite content of the Sewickley coal is between that of the Pittsburgh (lowest) and Redstone (highest). Ash yields and sulfur contents are typically higher than in the Pittsburgh or Redstone. The thin and thick Sewickley columns are palynologically and petrographically very similar; ash and sulfur are both higher in the thin column.

Diverse wetland vegetation flourished at the margins of the Midland Basin in north-central Texas during the Pennsylvanian Period. Extensive coastal swamps and an ever-wet, tropical climate supported lush growth of pteridosperm, marattialean fern, lycopsid, and calamite trees, and a wide array of ground cover and vines. As the Pennsylvanian passed into the Permian, the climate of the area became drier and more seasonal, the great swamps disappeared regionally, and aridity spread. The climatic inferences are based on changes in sedimentary patterns and paleosols as well as the general paleobotanical trends. The lithological patterns include a change from a diverse array of paleosols, including Histosols (ever-wet waterlogged soils), in the late Pennsylvanian to greatly diminished paleosol diversity with poorly developed Vertisols by the Early–Middle Permian transition. In addition, coal seams were present with wide areal distribution in the late Pennsylvanian whereas beds of evaporates were common by the end of the Early Permian. During this climatic transition, wetland plants were confined to shrinking “wet spots” found along permanent streams where the vegetation they constituted remained distinct if increasingly depauperate in terms of species richness. By Leonardian (late Early Permian) time, most of the landscape was dominated by plants adapted to seasonal drought and a deep water table. Wetland elements were reduced to scattered pockets, dominated primarily by weedy forms and riparian specialists tolerant of flooding and burial. By the Middle Permian, even these small wetland pockets had disappeared from the region.

Dedicated to the memory of William T. Holser, colleague and friend. A gap in the fossil record of coals and coral reefs during the Early Triassic follows the greatest of mass extinctions at the Permian-Triassic boundary. Catastrophic methane outbursts during terminal Permian global mass extinction are indicated by organic carbon isotopic (δ 13 C org ) values of less than –37‰, and preferential sequestration of 13 C-depleted carbon at high latitudes and on land, relative to low latitudes and deep ocean. Methane outbursts massive enough to account for observed carbon isotopic anomalies require unusually efficient release from thermal alteration of coal measures or from methane-bearing permafrost or marine methane-hydrate reservoirs due to bolide impact, volcanic eruption, submarine landslides, or global warming. The terminal Permian carbon isotopic anomaly has been regarded as a consequence of mass extinction, but atmospheric injections of methane and its oxidation to carbon dioxide could have been a cause of extinction for animals, plants, coral reefs and peat swamps, killing by hypoxia, hypercapnia, acidosis, and pulmonary edema. Extinction by hydrocarbon pollution of the atmosphere is compatible with many details of the marine and terrestrial fossil records, and with observed marine and nonmarine facies changes. Multiple methane releases explain not only erratic early Triassic carbon isotopic values, but also protracted (∼6 m.y.) global suppression of coral reefs and peat swamps.

The Main Seam in the Greymouth coalfield (Upper Cretaceous Paparoa Coal Measures) is exceptionally thick (>25m) and occurs in three locally thick pods, termed north, middle, and south. These pods are separated by areas of thin or absent (“barren”) coal. The barren zone between the north and middle coal pods is characterized by a sequence that is 60 m thick comprising relatively thin (1–2.5 m thick) but laterally extensive (up to 500 m) sandstone units. The orientation of both the thin and the barren coal zones is approximately east to west. This is coincident with basement fault systems that occur in the region. Therefore, the stacked nature of the sandstones within this narrow zone may be a result of differential subsidence across basement fault blocks. The Main Seam, like the sandstone units in the “barren” zone, is inferred to represent a stacked sequence. Two zones of thin partings (<20 cm in thickness) occur in the coal, and even where these zones do not occur, an interval of abundant vitrain bands is present. As has been suggested for other coal beds, intervals with high vitrain content may represent a demarcation between different paleomire systems, or, as in the case of the Main Seam, periods where the paleomire was rejuvenated with plant nutrients, allowing continued aggradation of the mire. The low ash yield (<5% dry basis) indicates that the Main Seam was rarely affected by flood incursions. This may have been the result of both doming of the peat surface as well as restriction of the dominant sediment flow by syn-sedimentary faulting. Palynological analyses indicate that the Main Seam mire throughout most of its time was dominated by gymnosperms, particularly a relative of the Huron pine ( Lagarostrobus franklinii ). However, a distinct floral change to a Gleichenia -dominated mire occurs in the upper few meters of the Main Seam. This vegetation change may have resulted from basinwide environmental or climatic change. Gleichenia does not produce much biomass, and if it was the dominant mire plant it may not have been able to keep peat accumulation rates higher than subsidence. Whether the cause was a decrease in peat accumulation or a drying of mire, the result would have been lowering of the surface to a degree that flooding and final termination would be likely.

Late Quaternary history and paleoecology of a small oxbow wetland on glaciated terrain were investigated using sediment lithology (cores, bulk samples, backhoedug trenches), ground-penetrating radar, vascular plant and moss macrofossil stratigraphies, and accelerator mass spectrometric radiocarbon dating. A nearly complete mastodon skeleton was recovered from late Pleistocene detrital peat and peaty marl near the top of the sediment sequence. Sedimentation in the basin began with silt and clay over dense cobble outwash transported southward from the nearby Hyde Park Moraine. Overbank sediment deposition occurred between ∼13,000 and 12,220 yr B.P. during a period of tundra vegetation, which ended with a sharp rise in spruce needle abundance and a shift to autochthonous marl and finally peat deposition. Fossils of aquatic and wetland plants began to accumulate before the tundra-spruce transition and increased after it. Rich fen wetland began to infill the pond with peat, while the upland supported open white spruce and later white spruce–balsam fir–tamarack forest. The mastodon, 11,480 ± 40 radiocarbon years old, was contemporaneous with spruce–balsam fir–tamarack forest and rich fen wetland. Many mastodon bones were articulated or nearly so, indicating that the animal died in the basin and that postmortem bone dispersal was slight.

Roof-shale floras have been a major source of data for the understanding of Carboniferous vegetation. Early debate on their origin centered around the question of whether these megafloral assemblages are autochthonous or allochthonous. In these discussions, the sedimentological context in which the preserved fossil assemblage (taphoflora) occurred was largely ignored. W. C. Darrah saw the complexity of these issues, presented helpful starting points for further investigations, and influenced the thinking of the next generation. This chapter characterizes the sedimentological and taphonomic features of a spectrum of roof-shale floras. There are three levels at which the preservation of plant parts can be viewed: (1) early taphonomic processes and earliest diagenesis can destroy or preserve plant parts in a given clastic depositional setting; (2) those plant parts that are preserved can be autochthonous, parautochthonous, or allochthonous in relationship to their original place of growth; (3) with respect to a peat layer (coal bed), the overlying clastic material can be deposited in a continuous transition, after a short temporal break (discontinuity), or after a significant hiatus of time. Characterization of roof-shale floras must take into consideration the sedimentological interpretation of the associated lithologies, the stratigraphic sequence, and the taphonomic processes involved in their formation. Characterization is essential before such floras can be used in higher-level interpretations, such as paleoecological reconstructions.