relative species abundances in individual zones (Phillips et al., 1977; Phillips and DiMichele, 1981) is the basic unit for further quantitative studies. Inferences can be drawn regarding the vegetational structure of the forest from which the litter sample was drawn, and the dynamics of that forest can be deduced from analysis of the relative proportions, life histories, and habits of the plants.
Aspects of the physical habitat also can be determined or deduced for individual zones, revealing the conditions under which the parent forest grew. Physical evidence of mineral matter in coal, either through ash determinations (Eble and Grady, 1993) or petrography (Johnson, 1979; Esterle and Ferm, 1986; Grady and Eble, 1990), suggests susceptibility of the swamp to floods, that introduce clastics, or peat decay, which may concentrate mineral matter (Cecil et al., 1979; Ruppert et al., 1985). Relative abundances of mineral charcoal (fusain) indicate the presence of fires in the swamp and can point to particular plant groups or tissues that are most frequently preserved as charcoal, or to patterns of preservation that suggest fire intensity (charred outer rinds on tree bark) (Phillips and DiMichele, 1981; DiMichele and Phillips, 1988). The proximity of a coal-ball zone to the underclay, or to thick mineral partings in the coal bed, indicates significant changes in edaphic conditions to or from those characteristic of the peat substrate. In many cases, the mineral partings represent disturbances by clastic-laden floodwaters. Specific plant groups or assemblages have been found to have a disproportionate association with these indicators.
Taphonomic criteria, that is, the characteristics of the preservation of the coal-ball peat, also provide insights into the physical character of the habitat associated with a particular zonal assemblage (Scott and Rex, 1985; Raymond, 1987; Phillips and DiMichele, 1989). Peat exposure can be gauged from the degree to which the peat is rotted and the plant remains poorly preserved. Exposure is also indicated by detritivore activity, such as frass-filled borings in wood, periderm, or soft tissues, or extensive fecal pellet deposits left by surface feeders. The framework-to-matrix ratio of the peat is an indicator of the degree to which fine, particulate material has been flushed from the peat fabric by water through-flow, which may also have affected peat acidity.
Finally, habitat inferences can be drawn from the biotic character of the assemblage itself, through ecomorphic analysis. For example, the relative proportion of ground cover and free-sporing life histories (the "fern" life cycle) can suggest the frequency or length of surface water cover. Relative proportions of opportunistic/invasive life histories, in combination with high species richness and high ecomorphic diversity of an assemblage, can point to recent disturbance and subsequent colonization. The proportions of monocarpic (one reproduction at the end of the lifetime) versus polycarpic (multiple reproductions during lifetime) lycopsid life histories suggest the relative extent to which the environment was stable and "predictable."
These basic patterns are the fundamental building blocks of our higher-order inferences about coal-swamp structure and temporal dynamics. We extend them through time and space to construct a larger, more complex picture.
Coal-ball occurrences in multiple layers, or profiles, provide access to changes in vegetation, vegetational dynamics, and habitats through time. The time interval may vary from 103 to 105 yr, depending on how thick the coal seam is and how much of it was preserved as coal-ball peat.
Dominance and diversity patterns on a site through time can be quantified through profile analysis. In many cases, patterns of vegetational change are associated with physical markers that can be correlated between coal-ball masses within a single seam, such as clastic partings, fusain layers, or distinctive petrographic composition. The evidence we have accumulated to date suggests little directional succession at a whole-seam level, consistent with a planar-peat origin for most coal-ball bearing coals, and no discernible evolutionary change in the morphology of the component species within the time necessary to accumulate a 1- to 2-m thick coal seam.
Landscape patterns within a single coal swamp can be reconstructed from profile data, using single or multiple profiles, or from random sample data. Within any one profile the succession of zones can be taken to represent an atemporal record or sample of the kinds of habitats and assemblages that existed within the swamp. Ordination (see Digby and Kempton, 1987) of individual zones, or of individual coal balls in the case of random samples, provides a base map of landscape-level variability within the coal, including the relative proportions of habitats, dominance-diversity patterns, and the relative similarity of assemblages from different subenvironments. The patterns at the landscape level, including the relative proportions of habitats, plant assemblages, life histories, and community dynamics, are characteristic of a coal or group of coals. These provide a "fingerprint" of a particular coal swamp, and more broadly of a time and geographic region.
It is on the longest time scales that evolutionary change can be detected, and on which the effects of climatic or other extrinsic factors on swamp community dynamics