PALEOBOTANICAL AND PALEOECOLOGICAL CONSTRAINTS ON MODELS OF PEAT FORMATION IN THE LATE CARBONIFEROUS OF EURAMERICA

The dominant plants of the Late Carboniferous lowland tropics were taxonomically and structurally distinct from those of any later time periods. Dominance was distributed among lycopsids, ferns, sphenopsids, pteridosperms and cordaites, and each of these groups had distinctive and different ecological preferences and amplitudes. Peat-forming habitats were dominated by lycopsids throughout the Westphalian, with a significant cordaitean element in the middle Westphalian; during the Stephanian tree ferns were dominant, following major extinctions near the Westphalian-Stephanian transition. Each of the major plant groups had distinctive architectures and tissue composition. Trees contributed up to 95% of the peat biomass and tree forms of lycopsids, Psaronius and Medullosa lack good modern analogues. The cordaites were the only woody plant group to contribute significantly to peat, and then only during the mid-Westphalian. Structurally wood-like lycopsid bark is the major ''woody'' tissue encountered in most Westphalian coals. Tree ferns and pteridosperms were largely parenchymatous in construction; the stigmarian root systems of lycopsids also were largely parenchymatous. The tissue structure of these dominant plants suggests the need for extreme caution in the inference of mire ecological conditions or vegetational structure from coal petrographic data. Peat formed under arborescent ferns or pteridosperms, or peat repeatedly exposed to decay and rerooting by stigmarian root systems of lycopsids, would have a distinctly non-woody signature and yet would have formed in a forested environment. A summary is presented of the autecology and synecology of mire plants, emphazing the structural framework provided by lycopsids during the Westphalian. Certain constraints in the links between peat biomass and miospore palynology are discussed in terms of over-representation, under-representation and non-representation. The formulation of Smith's four-phase hydroseral model is discussed and compared with more recent data available from plant paleoecology. The current debate over an ombrotrophic vs. rheotrophic origin of Late Carboniferous peats relies in large part on paleobotanical data, almost entirely palynological, in combination with petrographic analyses. Ecological studies of miospores and of coal-ball and compression macrofossils, and the linkage of miospores to source plants, permit the re-evaluation of mire successional models. Evidence for tree lycopsids, sphenopsids, pteridosperms and cordaites suggests growth mainly in rheotrophic mires. Tree ferns are likely candidates for growth in domed mires, although evidence is ambiguous and some tree ferns clearly grew under rheotrophic conditions. Densospores, produced by at least Sporangiostrobus lycopsid subtrees, have been considered diagnostic of ombrotrophic conditions; abundant evidence refutes this simplistic interpretation and suggests broad ecological amplitudes for densospore producers, including growth under rheotrophic conditions. Although plant fossils alone can not resolve most of the major debates in modern coal geology, paleobotany does contribute significantly to our understanding of ancient mires. An approach combining paleobotanical data with petrography, sedimentology and geochemistry, on a case by case basis, is most likely to produce a clear picture.