set of species-habitat dynamics was established. In combination with replacement patterns during the Westphalian, the extinction suggests that biotic interactions among species assist in creating a "fabric" or multiniche system that helps constrain the ecomorphic nature of species replacements, whether such replacements occur through evolution within or migration into the system. Loss or decline of a species creates a vacant niche, whose limits are partially defined by the remaining biota. Major disruption of this system by extrinsically induced extinction permits the system to reestablish the interaction fabric based on the biologies of the new suite of species.
One of the most important tenets of conventional wisdom currently dominant in plant community ecology is the concept of individualism. This is a highly reductionist principle that attributes all apparent structure in plant communities to the properties of individual organisms; associations of species are a consequence only of their similar tolerances to physical conditions and similar resource requirements. Such a view allows no "emergent" properties, characteristics of ecosystems that result from the interactions of coexisting species populations (avatars in the terminology of Damuth, 1985). Some plant ecologists have turned to the fossil record, almost exclusively that of Quaternary pollen, to test the predictions of this hypothesis when dynamics are viewed over a longer time scale (e.g., Delcourt and Delcourt, 1987; Overpeck et al., 1992) and have found much congruence. Consequently, the pollen record of the past 10,000 yr has come to stand as proxy for the past 420 million years (m.y.) of terrestrial plant history. What about the rest of the record?
Most of the terrestrial fossil record is less complete than that of the Quaternary. This is not to imply that it cannot be sampled at the same scale. It can, as studies of spores and pollen from very ancient sediments indicate (e.g., Smith and Butterworth, 1967; Mahaffy, 1985; Farley, 1989; Eble and Grady, 1993). The major difference lies in our understanding of contemporaneous, potentially causative changes in climate and paleogeography, resolved much more coarsely as one looks deeper into time. Yet the older record preserves patterns of change over a much broader span of time intervals than 10,000 to 100,000 yr.
Where studied in detail, the past does not reveal a pattern of ever-changing community structure and composition. Rather, remarkable periods of persistence of species associations and ecomorphic patterns are found, in which species turnover may occur but structure and dynamics remain largely unchanged for a few to many millions of years. Examples do not abound because little detailed research on community paleoecological patterns has been pursued outside the Late Carboniferous (e.g.; Scott, 1978; Phillips and DiMichele, 1981; Gastaldo, 1987; Raymond, 1988) or the early Tertiary (e.g., Wing, 1988, Farley, 1989). However, these vastly different ecosystems show remarkably similar patterns and suggest that the Quaternary record cannot be used reliably as a blanket model of terrestrial plant community dynamics through time.
In this chapter we focus on changes in the taxonomic composition and structural properties of late Paleozoic plant communities from wetland habitats, in particular those of peat-forming swamps. Peat-forming communities of the Late Carboniferous (Pennsylvanian) provide some of the best opportunities to characterize and analyze an ancient ecosystem. Each of the five major plant groups (lycopsids, ferns, sphenopsids, pteridosperms, and cordaites) is highly distinctive morphologically and anatomically, the generic and species diversities are quite low, and taxa were largely pantropical in distribution within the wetlands. Particularly important in resolving the plants and their intraswamp habitats are coal balls, which are essentially in situ accumulations of plant litter preserved relatively uncompacted in concretions within coal seams. Coal balls occur in many coals, mostly in the upper Carboniferous. Coal balls, miospores, and compression fossils have been studied from across the ancient Euramerican paleocontinent largely as a consequence of their exposure during coal mining and because of their relevance to coal geology as well as paleobotany. Similarly, studies of depositional patterns, and interpretations of regional and global climate have engendered considerable debate (e.g., Wanless et al., 1969; Horne et al., 1978; Phillips and Cecil, 1985; Heckel, 1986; Klein and Willard, 1989; Cecil, 1990). Despite disagreements, the data, and varying conclusions drawn from them, have yielded greater insight than for any pre-Tertiary systems.
The implications of this research may go far beyond the late Paleozoic. Paleoecology provides a baseline against which we can examine, for greater generality, our concepts formulated almost entirely from extant ecosystems. The fossil record is our only access to long-term patterns and their implications for questions of community persistence, stability, and response to major extrinsic perturbations. By examining the response of ancient ecosystems to major disruptions we hope to gain insights into the ways in which plant communities of today may behave—not the