Climate change, plant extinctions and vegetational recovery during the Middle-Late Pennsylvanian Transition: the Case of tropical peat- forming environments in North America WILLIAM A. DIMICHELE* & TOM L. PHILLIPS^ ^Department of Paleobiology, National Museum of Natural History, Smithsonian Institution, Washington, DC 20560, USA '^Department of Plant Biology, University of Illinois, 505 So. Goodwin Ave., Urbana, IL 61801, USA Abstract: A major extinction of terrestrial plants occurred at the end of the Westphalian (Middle Pennsylvanian) in the lowland tropics of North America. Approximately 67% of the species in peat-forming mires, and at least half the species in clastic wetlands were eliminated by changing climatic conditions, probably protracted moisture deficits or exaggeration of seasonal dryness. Independent studies suggest that the end of the Westphalian was marked by deglaciation in Gondwana and increase in global temperature. In peat-forming habitats major floristic changes followed the extinction. Earliest Stephanian (Late Pennsylvanian) dominance-diversity patterns were highly variable temporally; several different plant groups that were of minor importance in Westphalian mires became major framework dominants, with high coal to coal variability. Ultimately opportunistic tree ferns, previously subdominant, became the dominant elements. Westphalian tree ferns were mostly small, cheaply constructed colonists with massive reproductive outputs and broad ecological amplitudes. Stephanian tree ferns were much larger and appear to have occupied sites for extended periods; they retained the cheap construction of earlier forms but added massive root mantles, which permitted greater height and girth. The 'marsh'-forming lycopsid Chaloneria also became common in Stephanian mires. These elements formed mire landscapes that had few analogues in the older Westphalian mire forests, dominated by tree lycopsids. Generic dominance patterns in the Stephanian became fairly consistent after the initial period of irregularity; however, the species composition of Stephanian mires was highly variable vnthin generic themes. Ecological assemblages were persistent during most of the Westphalian and Stephanian. Patterns at the end-WestphaUan suggest that high levels of species extinction disrupt self- regulatory properties of groups of species and create intervals of-l6ttery-like ecological dynamics. Opportunists may have an advantage during these periods, but ultimately they must give rise to species capable of site occupation. Thus the size of the extinction, and the proximity of existing species that can recolonize vacated resource space, will dictate whether speciation or colonization will rebuild the new landscapes. A major consequence of mass extinction is the tence to invasion, as long as extinction levels are vast change in ecological conditions. Disappear- low. Such patterns suggest that ecosystems can ances of species or extreme reduction in their evolve properties of self-regulation and hier- numbers or areal extent release previously archical structure (Miller 1990; Allen & Hoek- sequestered resources, change the ecomorphic stra 1992), which may have been important character of multispecies assemblages, and alter modes of organization on land and sea during the numbers, kinds and strengths of interspecific much of the Phanerozoic. Separating periods of interactions (Pimm 1991). The result is altera- persistent dynamics are briefer intervals with tion of the ecological background through which elevated extinction levels and somewhat more evolutionary changes are filtered. Evidence from chaotic dynamics. It is within these time periods both the marine (Boucot 1983; Brett et al. 1990; that major evolutionary changes may be con- Miller, 1993; Brett & Baird 1995) and terrestrial centrated (Vermeij 1987; Brett & Baird 1995). (Wing 1984; DiMichele & Phillips 1995) fossil Plant communities in the Pennsylvanian-age records indicates that ecological assemblages lowland tropics were characterized by long may be long-Uved with persistent (millions of periods of vegetational persistence. Assemblages years) dominance-diversity structure and resis- with similar species composition and quantita- From Hart, M. B. (ed.), 1996, Biotic Recovery from Mass Extinction Events, 201 Geological Society Special Publication No. 102, pp. 201-221 202 W. A. DIMICHELE & T. L. PHILLIPS tive dominance-diversity structure characterized particular habitats for intervals of millions of years before shifting abruptly to new equilibria. In both peat-forming mires (Phillips et al. 1985; DiMichele SL Phillips 1995) and mineral-soil wetlands (Pfefferkom & Thomson 1982) ele- vated levels of extinction and or origination accompanied major vegetational change. Regio- nal climatic changes, particularly in rainfall, are the most likely factors causing vegetational change (Phillips & Peppers 1984; Cecil 1990; Winston 1990); like the Recent, the Pennsyl- vanian was a time of major polar glaciations (Frakes et al. 1992), and experienced extensive climatic and sea-level fluctuations (Ross & Ross 1987, 1988; Heckel, 1986, 1989; Cecil 1990). The most profound vegetational change of the Pennsylvanian occurred during the Middle to Late Pennsylvanian (Westphalian-Stephanian) transition. At that time peat-forming mires were transformed from lycopsid dominance, which had persisted for more than nine million years, to tree-fern dominance. In addition, the hier- archical organization that characterized West- phalian mire landscapes (DiMichele & Phillips 1995) was replaced by vegetation with less spatio-temporal complexity (Willard & Phillips 1993). WestphaHan landscapes had stereotypic patterns of resource partitioning largely defined by the tree lycopsids (Scott 1978; CoUinson & Scott 1987a, b; DiMichele & PhilUps 1985, 1994). This partitioning created a fabric that endowed ecosystems with a substantial degree of self- regulation of species turnover dynamics over geologically significant spans of time (DiMichele & Phillips 1994&). In contrast, the Early Stephanian wet tropics were dominated by opportunist tree-fern lineages or by seed plants that had survived the end-Westphalian extinc- tions (Pfefferkom & Thomson 1982; Phillips et al. 1985). These extinctions, which affected both mires and clastic wetlands (Phillips et al. 1974; Gillespie & Pfefferkom 1985; Peppers 1985), were at the root of the ecological differences. Such an extensive and temporally abrupt reor- ganization of plant communities suggests a threshold-like response of the system to extinc- tion of the component species. As a conse- quence, the vegetation of the wet lowlands was taxonomically and structurally simplified, and the marked taxonomic differences between mires and clastic wetlands, which had characterized the WestphaHan, were muted (DiMichele & Aronson 1992). The study system: peat-forming mires Peat formation was a major distinguishing feature of the Pennsylvanian tropics. Vast tracts of swampy lowlands were covered with mires, a landscape indicative of high rainfall and high groundwater tables distributed throughout most of the year (Clymo 1987). Mire floras were quantitatively and, to a large extent qualitatively distinct from those of surrounding clastic wet- lands during the Westphalian (Middle Pennsyl- vanian). Lycopsids dominated peat substrates, on average, for the entire Westphalian with subdominant cordaitean gymnosperms from the Westphalian B through early D. In contrast, floodplains, levees, and other clastic lowland environments were dominated by pteridosperms, sphenopsids, and tree ferns (compare Phillips et al. 1985 and Pfefiferkorn & Thomson 1982; Peppers & Pfefferkom 1970). This distinction was dramatically reduced in the Late Pennsyl- vanian when tree ferns and pteridosperms became dominant elements in aU lowland-wet- land habitats (Pfefferkom & Thomson 1982; Phillips e< al. 1985; DiMichele & Aronson 1992), although species differences may have remained. Peat substrates present major physiological challenges to plants (Schlesinger 1978). Gener- ally low pH, flooding during substantial parts of the year, chelation of mineral nutrients, and, especially in the case of domed peats, highly oHgotrophic nutrient status, strongly select against most species. As such, mires can be vis- ualized as semi-closed, edaphic islands. During the Pennsylvanian certain evolutionary hneages, most notably stigmarian lycopsids, were ecolo- gically and evolutionarily centred in swamps and mires and had strongly partitioned these envir- onments (Phillips & Peppers 1984; DiMichele & Phillips 1985, 1994a). Most groups, and the pteridosperms can serve as the best understood example, were centred in the broader lowland wetlands but had few or no species that grew uniquely or predominantly on peat substrates; generaUy only a small subset of species from these lineages could grow on peats (DiMichele et al. 1985; Beeler 1983; Schabihon & Reihman 1985). Sphenopsids, marattialean ferns, and cordaites had patterns similar to pteridosperms, although the cordaites, and to a lesser extent the marattialean ferns, may have evolved some mire-centered lineages (Costanza 1985; Trivett & Rothwell 1985; Lesnikowska 1989; Trivett 1992). Sources of data Data on mire vegetation come from two principal sources, coal-ball macrofossils and pollen-spore microfossils. Coal balls are carbo- nate concretions that contain structurally pre- EXTINCTION AND RECOVERY IN PENNSYLVANIAN COAL SWAMPS 203 microfossils can be linked to parent plants (e.g. Mahaffy 1985; Willard 1993). Most of our analysis and inference is based on coal-ball data of two types, profile and random sample. Profiles of coal balls are collected in situ; relative position of coal-ball layers within the coal bed is noted, which permits recovery of the original zonation of the plant litter (Fig. 1). Random samples of coal balls are collected from one locality without respect to the position of coal balls in the coal bed. Consequently, they represent the average composition of the per- mineraUzed peat at the collection site. A coal- bed summary can be obtained from a profile by averaging the composition of individual coal- ball zones. Palynological and compression fossil data are presented and discussed where appro- priate, and in order to ampUfy coal-ball pat- terns. There are critical gaps in the coal-ball record, particularly in the early Stephanian, immediately following the extinctions at the end of the WestphaUan, and palynological data are of particular importance in this interval. The coal-ball data base used in this analysis is described in Phillips et al. (1985), with some additions of coals from the WestphaUan A and B of the United States (Winston & Phillips 1991; unpublished). Palynological data are mostly from Peppers (1985). Compression-impression species ranges are from Gillespie & Pfefferkom (1979). Fig. 1. Coal-balls i? situ; Herrin No. 6 Coal Member, Carbondale Formation, late WestphaUan D age; Old Ben Coal Company, No. 24 Mine, near Benton, Illinois. Coal balls are light-coloured material within the darker coal. Coal ball layers are extracted and analysed quantitatively, providing a partial record of vegetational change on the site through the time of peat accumulation. served peat stages of the coal and that occur within coal seams (Philhps et al. 1976; DeMaris et al. 1983; Scott & Rex 1985). Coal balls preserve part of the original Utter of the mire forest, often in exceptional anatomical detail. They occur in layers or aggregates (Fig. 1) and can replace much of the thickness of a coal seam, providing the basis for quantitative analysis of vegetational change during the history of peat accumulation (PhilUps et al. 1977). Coals usually contain abundant pollen and spore microfossils, produced by plants of the mire vegetation. Incremental sampling of a coal seam can reveal fine-scale details of the history of peat accumu- lation, particularly when the more abundant The pre-extinction system: late WestphaUan mire landscapes The late WestphaUan was the zenith of coal-age tropical plant diversity and landscape complex- ity. The primordial WestphaUan forests were a complex mixture of habitats each with distinct types of vegetation. Landscapes were strongly partitioned along taxonomic lines; each of several major clades had diversity and dom- inance peaks in particular types of physical settings (DiMichele & PhilUps 1994). A vegetational change occurred during the WestphaUan-Stephanian transition that entailed not only extinction, but a fundamental altera- tion in the dynamics of wetland ecosystems. The Stephanian was a period of much greater vegetational uniformity among habitats. It is possible that the spectrum of environments colonized by plants was smaller than in the WestphaUan, and that, overall, wetland vegeta- tion encompassed significantly shorter physical gradients in substrate wetness, nutrient avail- abiUty, and, perhaps, levels of physical distur- bance. 204 W. A. DIMICHELE & T. L. PHILLIPS 50 100 150 200 250 DCAl 350 Fig. 2. Detrended correspondence analysis of coal-baU zones from late Westphalian D Herrin, Springfield and Iron Post coals. Lines delimit groups of coal-ball zones with similar dominance patterns; letters refer to groups in Table 1 where dominance and life-history patterns are summarized. Species assemblages in late Westphalian D mires Peat-forming mires of the Westphalian D typically were composed of 40 to 50 species. Of the 50 species in our sample 31 were trees or subtrees, 19 were ground cover. As expected, this diversity was not randomly distributed among physical habitats, and several subenvir- onments can be recognized, each with a char- acteristic species assemblage (community-type) and dominance-diversity structure. The basic pattern is summarized in Fig. 2, an ordination of 96 zones from 7 profiles from three late Westphalian D coals, the Iron Post coal of Oklahoma, the Springfield coal of Indiana, and the Herrin coal of Illinois. This type of multi- community landscape is typical of all late Westphalian D coals we have studied; more detailed analyses of the Springfield and Herrin coals are pubhshed (Phillips & DiMichele 1981; Eggert et al. 1983; DiMichele & Phillips 1988; Willard 1993). The zones (vegetational stands) in Fig. 2 form three major groups: assemblages dominated by the monocarpic lycopsid tree Lepidophloios hallii, assemblages dominated by the polycarpic lycopsids Diaphorodendron scleroticum or Sigil- laria sp., and assemblages dominated by Medul- losa spp. and the polycarpic lycopsid Paralyco- podites brevifolius. Average species richness among these assemblage-types ranges from 7.2-8.2; major differences reside in the kinds of life histories that dominate, the average percen- tages of ground cover, and the amounts of associated fusain and mineral matter. Psaronius tree ferns are components (5-15% biomass) of nearly all assemblages except those dominated by Lepidophloios hallii lycopsids. Calamites are minor elements and occur in greatest abundance in association with Medullosa-Paralycopodites and to a lesser extent with Diaphorodendron and Sigillaria. Lepidophloios-dominated assemblages appear to have been best developed in habitats with extended periods of standing water. Species richness averages 7.2 in the 19 zones of the ordination, and includes low biomass and diversity of ground cover, vines, and understory trees and shrubs (Table 1). Quantitatively, lower vascular plants, particularly Psaronius tree ferns, are minor components and drop sharply in abundance as Lepidophloios proportions exceed 70% (DiMichele & Phillips 1988; Eble & Grady 1993). Ground cover and homosporous lower vascular plants both require exposed substrates, and their low abundances support the inference of flooded peat surfaces. In addition, the dominant Lepidophloios species produced seed- hke 'aquacarps' (Phillips & DiMichele 1992), which had physical properties consistent with water dispersal (Phillips 1979). Assemblages dominated by the polycarpic lycopsids, Diaphorodendron and/or Sigillaria, have variable composition. Species richness in these assemblages averages 7.7. Ground cover and vine richness averages 1.8 species per zone. However, quantitatively, ground cover and understory trees and shrubs were significant vegetational components, even if richness was low (Table 1). Levels of mineral matter and EXTINCTION AND RECOVERY IN PENNSYLVANIAN COAL SWAMPS 205 Table 1. Dominance and average abundance of ground cover and life history groups in ordinations of late Westphalian D coals and early Stephanian Calhoun coal. Group Dominant n GC LVP SP HP AQ Late Westphalian D A Lepidophloios > 70% 7.2 1.7 2.7 1.7 0.3 2.5 B Lepidophloios > 50% 8.7 2.3 3.3 2.1 0.2 3.1 C Lepidophloios > 40% 7.1 1.7 3.2 1.6 0.1 2.2 D Lepidodendron > 40% 9.3 1.5 2.8 1.5 0.2 5.0 E DiaphorodendronlSigillaria > 40% 7.7 1.8 2.6 1.9 0.8 2.5 F Medullosa/Paralycopodites/Psaronius 8.2 2.2 3.0 2.0 0.7 2.2 G Mixed Dominance 9.9 3.4 4.2 2.6 0.6 2.6 Average 8.3 2.1 3.1 1.9 0.5 2.8 Calhoun Coal-Early Stephanian A PsaroniusjMedullosajSigillaria 8.0 3.5 3.5 2.5 2.0 000 B Medullosaj Psaroniusj Sigillaria 10.0 5.5 5.5 3.2 1.0 000 C MedullosajPsaronius 10.7 7.0 7.0 2.3 1.3 000 D Psaronius 10.0 7.0 7.5 2.5 000 000 Average 9.7 5.8 5.9 2.6 1.1 000 Group letters refer to coal-ball zones within areas marked by the same letters on the respective ordination (Figs 2 & 3). Numbers refer to average number of species within each category, n, species richness; GC, ground cover; LVP, homosporous lower vascular plants; SP, seed plants; HP, free-sporing heterosporous plants; AQ, aquacarpic plants (sensu Phillips and DiMichele 1992). Ground cover will also be listed in one of the other categories. charcoal are not notably elevated in these assemblages. We suggest that they were char- acteristic of wet peats with occasional flooding and minor levels of physical disturbance. Assemblages enriched in or dominated by meduUosan pteridosperms and Paralycopodites brevifolius are associated with elevated levels of mineral matter and charcoal (Calder 1993; DiMichele & Phillips 1988; Johnson 1979). Average species richness is 8.2, with 2.2 species of ground cover or vines (Table 1). Subdominant species include a variety of life histories and growth forms, many of which are preferentially associated with mineral matter in coal (see literature summary in DiMichele & Phillips 1994) or that also occur in clastic swamp deposits. This vegetation can be characterized broadly as ecotonal, flourishing in habitats disturbed by flooding and possibly fire in areas often near the margins of peat bodies (Eble 1990; Eble et al. 1994). Temporal patterns and hierarchical organization in the Westphalian The type of landscape organization described above persisted within mires of western Eur- america for approximately two to three million years, based on the timescale of Hess & Lippolt (1986; Klein 1990). Landscapes partitioned among lycopsids, tree-ferns, and meduUosans have been identified in all Late Westphalian D coals from which coal balls have been recovered: the Secor coal of Oklahoma (DiMichele et al. 1992), the Iron Post coal of Oklahoma (Phillips et al. 1985), the Springfield coal of Indiana (Eggert et al. 1983; Mahaffy 1988; Willard 1993), the Middle Kittaning coal of Pennsylvania (Feng 1989), the Herrin coal of Illinois (PhiUips & DiMichele 1981; Mahaffy 1985; DiMichele & Phillips 1988), the Upper Freeport coal of Ohio (Phillips et al. 1985), the Baker coal of Kentucky (Phillips et al. 1985), and the Danville coal of Indiana. Palynological studies of other coals within this stratigraphic interval confirm a unity of composition and ecological structure (Peppers 1985; Kosanke 1988; Pierce et al. 1991; Eble et al. 1994). When examined through the entire Westpha- lian, mire landscapes appear to be hierarchically organized. 1. Several levels of spatial organization can be identified, and each has certain temporal dynamics that operate only at that scale. In par- ticular we recognize ecomorphs (guilds) within species assemblages (communities), and species assemblages within landscapes (DiMichele 1994). 2. The basic species assemblages (commu- nities) identified in late Westphahan D mires occurred throughout the entire Westphalian. Distinct dominance-diversity structure, domi- nant species groups, and associated physical 206 W. A. DIMICHELE & T. L. PHILLIPS attributes were conserved for > 9 Ma (Di- Michele & Phillips 1995). Late Westphalian D mires were only one of several mire landscape types that existed sequentially during the West- phalian (PhiUips & Peppers 1984; Phillips et ah 1985). The average species composition of ear- her landscapes varied according to the propor- tions of the major species assemblages each contained. But the biotic characteristics of any particular species assemblage and the physical conditions with which it was associated re- mained distinct throughout the Westphalian. Thus, changes in the relative abundances of dominant species on a coal-bed average, and thus in the landscape composition of mires, was due mostly to changes in the proportions of different physical habitats. 3. Although there was substantial species turnover during the Westphalian, this turnover was largely on ecomorphic themes within habitats; species generally replaced others of the same clade, with similar growth architecture and life history, and remained within the habitat limits typical of those clades. Thus, in the broadest sense of ecomorphic patterns, intraha- bitat organization persisted despite species turn- over. Within habitat species turnover suggests an element of biotic control on species replace- ment, an emergent assemblage property beyond a strictly individuaUstic organization. We have attempted to document and detail these patterns elsewhere (DiMichele & PhilUps 1994a). Westphalian marattialean ferns The marattialean ferns are a single, and most noteworthy, exception to the general pattern of Westphalian habitat restriction. Although broadly present in mires beginning in the Westphahan C (Peppers 1985), the first real abundances recorded by coal-ball macrofossils are not until near the Westphalian C-D transi- tion in Iowa (Phillips et al. 1985). Marattialean ferns do not become generally abundant until the middle of the Westphalian D, and do so in both mires (Phillips & Peppers 1984) and clastic wetlands (Pfefferkorn & Thomson 1982). Most Psaronius (Pecopteris in compression) species in the Westphalian were small trees, possibly a few were sprawling ground cover (Lesnikowska 1989). Large Psaronius trees began to appear in compression preservation during the Westphahan D (Pfefferkorn 1976), but truly large trees are not known from mires until the Stephanian. UnUke most other groups Psaronius species appear to have had very broad ecological amphtudes, permitting them to be- come interstitial opportunists within a variety of other vegetation types. Their free-sporing hfe histories apparently limited the colonizing abihty of Psaronius species to environments with exposed substrates. Extinction patterns and the Westphalian- Stephanian transition The transition from the Westphalian (Middle Pennsylvanian) to the Stephanian (Late Penn- sylvanian) was marked by a major extinction of tropical lowland plants, particularly in North America (western Euramerica). White & Thies- sen (1913) first recognized the major floristic change, based on compression foUage. Later Kosanke (1947) detected the change in Illinois Basin coals using palynology. Recognition that this was a major extinction of lycopsid genera and included a major vegetational change became evident when coal-ball and palynologi- cal data were combined (Phillips et al. 1974). Monographic revisions of plant groups pre- served in coal balls indicate a much broader extinction that includes most of the species of Westphalian tree ferns (Lesnikowska 1989) and pteridosperms (Taylor 1965). Palynological work in the U.S. (Peppers 1985; Kosanke 1988), and Europe (Stechesgolev 1975) demon- strates that the extinction was time transgressive from west to east across the Euramerican tropics, occurring in western Russia in the early part of the Stephanian. Lycospora-hearing ly- copsids persisted in Europe during the Stepha- nian but at markedly reduced abundance (see Lorenzo 1979; Phillips et al. 1985). Stephanian lowland, wetland habitats were both more uni- form and lower in species richness than those of the Westphalian, except perhaps in Cathaysia where floras typical of Euramerican wetlands persisted into the Late Permian (Phillips et al. 1985; Ziegler 1990; Guo 1992). Parts of the Cathaysian tropics apparently remained con- tinuously wet throughout the Carboniferous, providing a refugium for the most moisture- sensitive floras. Causation Extinctions during the Westphalian-Stephanian transition appear to have been caused by changing climatic conditions in the tropics. Reduction in the amount of rainfall and increased seasonality in its distribution are the most hkely factors to have affected the wetland biome. Peat-forming environments would have been the most sensitive landscapes within that biome. Because mires contained many edaphic- EXTINCTION AND RECOVERY IN PENNSYLVANIAN COAL SWAMPS 207 specialist taxa, climatic changes could have greatly affected migration routes between refu- gial safe sites. Lower levels of extinction are expected for clastic compression floras, which represent wetland plants with broader edaphic tolerances and thus access to a greater variety of migration routes. Several studies have inferred regional drying in western Euramerica during the Westphalian- Stephanian transition. Phillips & Peppers (1984) used coal resource abundances from the Appa- lachians and Midcontinent coal basins as indicators of general environmental moisture availabiUty. Their data suggest a marked period of drying coinciding with the transition and extinction. Cecil (1990) and Cecil et al. (1985) drew similar conclusions from the analysis of coal geochemistry, coal-body geometry, and analysis of clastic rocks associated with coals. Winston (1986, 1989) identified plant tissues in polished bocks of coal, and, based on current ecological understanding of the plants, inferred a chmate signature like that of the other studies. Recently Frakes et al. (1992) have re-evalu- ated the evidence for late Palaeozoic glaciations. They find that the Westphalian was a time of glacial maximum. This maximum ended abruptly at the Westphalian-Stephanian bound- ary, and the early Stephanian saw greatly diminished glaciation and possibly an increase in global temperature. Changes in the extent of polar glaciations directly affect the width and position of the intertropical convergence zone (see Parrish 1982; Ziegler et al. 1987), which in turn affect the amount and periodicity of annual rainfall. This joins with the other fines of evidence to suggest a global, allogenic cause for plant extinctions in the Westphafian tropical wetlands. Turnover Species origination and extinction were calcu- lated on the basis of stratigraphic range data. Two types of calculation were performed using coal-ball data. The first considered the actual origination and extinction between coals through the later WestphaUan D and early Stephanian. This method does not account for the Signor-Lipps effect, in which an extinction is 'smeared' backwards stratigraphically due to sampling effects. To account for this we also examined summary turnover between the latest Westphalian mires, a group uniformly domi- nated by lycopsids, tree ferns and pteridosperms, and early Stephanian mires, dominated by tree ferns and pteridosperms. In effect the compar- ison is restricted to two diff"erent types of ecosystem, each internally uniform. The sum- mary turnover also was broken down into two ecomorphic groups, trees and ground cover/ vines/shrubs. Summary extinction and origina- tion were calculated for compression-impression data reported by Gillespie & Pfeff'erkom (1985), and for coal palynological data reported by Peppers (1985). Table 2. Species diversity, origination, and extinction during the late Westphalian and early Stephanian based on coal-ball data Coal First Last Species occurrences occurrences richness Stephanian Calhoun 3 ? 31 Duquesne 5 7 36 Friendsville 5 0 30 Bristol Hill 5 2 27 Westphalian Baker 0 17 39 Lower Freeport 0 7 46 Herrin 7 2 48 Middle Kittaning 0 2 43 Springfield 3 4 47 Table 3. Extinction, survivorship and origination based on spore-pollen data (from Peppers 1985) 43 total species in late Westphalian 19 species terminate in late Westphalian D (44.2% extinction) 24 species range through W/S boundary (55.8% persistence) 26 total species in early Stephanian 2 species originate in early Stephanian (7.7% origination) Species richness, origination, and extinction patterns are presented in Table 2 for nine stratigraphically successive coals from the late Westphafian and early Stephanian. A 'species' in this compilation is either a reconstructed whole plant for which isolated organs have been reassembled, or it is a diagnostic organ or group of organs. Seeds of pteridosperms, for example, are used as proxy for parental plant species, even though they may underrepresent true species diversity; at present there are too few recon- structions of most pteridosperm groups to use whole-plant data. Psaronius species are mostly presented as whole plants, based on the work of Lesnikowska (1989); reproductive organs were used as proxy for whole plants where their 208 W. A. DIMICHELE & T. L. PHILLIPS Table 4. Extinction, survivorship and origination based on clastic compression data (from Gillespie & Pfefferkorn 1979) 18 total species in late Westphalian 9 species extinct in late Westphalian (50% extinction) Lepidodendron aculeatum, Asolanus camptotaenia, Neuropteris helerophylla, N. rarinervis, Mariopteris nervosa, Alethopteris serlii, Annularia radiala, Annularia sphenophylloides, Sphenophyllum majus 9 species range through W/S boundary (50% persistence) Neuropteris scheuchzeri, N. ovata, Pecopteris miltonii, P. unita, P. hemitelioides, Asterophyllites equisetiformis, Annularia mucronata, A. stellata, Sphenophyllum emarginatum 10 total species in early Stephanian 1 species originates in early Stephanian (10% origination) Pseudomariopteris ribeyronii Table 5. Stratigraphic ranges of coal-ball pi 'ants in the late ? Westphalian and early Stephanian Stephanian Westphalian Ca Du Fr Br Da Ba LF He Ki Sp Bot. pseudoantiqua X = = = = = = X = X B. forensis X X X X X X = X = X B. cratis X = X Anachoropteris involuta (lateral) X X X X A. clavata X = X A. giUotii X = = = = = = X = = A. involuta (adaxial) X X X X X X A. sp. X Apotropteris minuta X Sermaya biseriata X X = = = = = X = = Z. berryvillensis (Biscalitheca) X X X X Z. illinoiensis (Corynepteris) X X = = X X = X = = Rhabdoxylon americanum X Ankyropteris brongniartii X X = = = = = X = X Ankyropteris n. sp. X X X X Phillipopteris globoformis X Doneggia compleura X Calamite (Pendulostachys) X Arthropitys (Calamocarpon) X X X = X X = X = X Arthropitys (Calamostachys a.) X = X Arthropitys (Palaeostachya d.) X = ? = = = Calamodendron X = = = X Sphenophyllum (Bowmanites sp.) X = = X = X Sphenophyllum (Peltastrobus r.) X = = X = X Callistophyton poroxyloides X X X X Callistophyton boysetii X X X X X X Schopfiastrum decussatum X = = = = X = X = = Heterangium (Conostoma quadrat.) X X X Heterangium (C. villosum) X X Heterangium (C. platyspermum) X X = = = = X X X X Heterangium (C. kestospermum) X = = = X = X Physostoma calcaratum X = = = Cyathotheca ventilaria X Stellastelara parvula X = = = = = = X X S. baxteri X Cordaixylon (Cardiocarpus ovi.) X X X = X X = X = X Mesoxylon priapi (Mitro. vine.) X X Psaronius (Scolecopteris majopsis) X Psaronius (S. parkerensis) X Psaronius (Ariangium pygmaem) X Psaronius (S. calicifolia) X Psaronius (S. valumii) X = X = X Psaronius (S. mamayi) X = X = X Psaronius (S. gnoma) X X = = = X EXTINCTION AND RECOVERY IN PENNSYLVANIAN COAL SWAMPS 209 Psaronius (S. latifolia A) x x = x x x Psaronius (S. minor) x = = x = x Psaronius (S. altissima) x x = = x P. chasei (S. illinoensis) x x x x P. blicklei (S. monothrix) x = x P. magnificus (S. latifolia B) x x x x MeduUosa (Pachytesta hexang.) x x x MeduUosa (P. berryvillensis) x x x MeduUosa (P. incrassata) x MeduUosa (P. iUinoense) x = x = MeduUosa (P. composita) MeduUosa (P. stewartii) x? MeduUosa (P. saharasperma) MeduUosa (P. noei) MeduUosa (P. gigantea) x MeduUosa (P. vera) MeduUosa (P. hoskinsii) MeduUosa (Stephanospermum el.) x = = = MeduUosa (Stephanospermum sp.) x = MeduUosa (Hexapterospermum sp.) x MeduUosa (H. delevoryii) MeduUosa (Albertlongia incostata) MeduUosa endocentrica x x = = Sutcliffia insignis Coronostoma quadrivasatum x Lepidodendron hickii Lepidophloios halUi Lepidophloios johnsonii Paralycopodites brevif. Hizemodendron serratum Diaphorodendron scleroticum Synchysidendron resinosum Sublepidophloios sp. Chaloneria periodica Chaloneria cormosa x x x SigiUaria sp. x x = x = x S. approximata (Mazocarpon oed.) x = x x SigiUaria (M. vUlosum) x SigiUaria (M. bensonii) x Paurodendron fraipontii xxx== x = x= x x, present in a coal; =, presence in a coal inferred by ocurrences in earlier and later coals. Ca, Calhoun coal (IL); Du, Duquesne coal (OH); Fr, Friendsville coal (IL); Br, Bristol HiU coal (IL); Da, DanviUe coal (IN); Ba, Baker coal (KY); LF, Lower Freeport coal (OH); He, Herrin coal (IL); Ki, Middle Kittaning coal (PA); Sp, Springfield coal (IL). Table 6. Tree extinction, survivorship and origination based on coal-ball data. Psaronius taxonomy from Lesnikowska (1989). Taxa of medullosans based on ovules, o/Arthropitys on cones 30 total species of trees in late Westphalian 26 species extinct in late Westphalian (86.7% extinction) Calamodendron sp., Palaeostachya decacnema, Lepidodendron hickii, Lepidophloios hallii, L. johnsonii, Diaphorodendron scleroticum, Synchysidendron resinosum, Hizemodendron serratum, Paralycopodites brevifolius, SigiUaria sp., Psaronius {Scolecopteris) majopsis, P. calicifolia, P. valumii, P. mamayi, P. gnoma, P. latifolia (type A of Lesnikowska, 1979), P. minor, P. altissima, Hexapterospermum delevoryii, Albertlongia incostata, Pachytesta composita, P. stewartii, P. saharasperma, P. noei, P. vera, P. hoskinsii 4 species range through W/S boundary (13.3% persistence) Stephanospermum elongatum, Pachytesta iUinoense, Pachytesta gigantea, Calamocdrpon insignis 17 total species of trees in early Stephanian 13 species originate in early Stephanian (76.5% origination) Psaronius chasei, P. blicklei, P. magnificus, P.parkerensis, P. (Araiangium)pygmaem, Hexapterospermum sp., Pachytesta hexangulata, P. berryvillensis, P. incrassata, SigiUaria approximata, SigiUaria (Mazocarpon villosa), SigiUaria {Mazocarpon bensonii), Pendylostachys cingulatum - = = X = X X X = = X X = X = X X = = X X X = = X = = = = = X = = = = = X X X = X X = X X X X X X X X XX X X = X X = X = X X X X X X = = = X XX XXX X XX XXX X X X XX = X = X 210 W. A. DIMICHELE & T. L. PHILLIPS Table 7. Ground cover, vine and shrub extinction based on coal-ball data. Reproductive organs used as proxy for natural species where whole-plant reconstructions have not been completed 18 total species of ground cover, vines and shrubs in late Westphalian 6 species extinct in late Westphalian (33.3% extinction) Physostoma calcaratum, Callistophyton boysetii, Chaloneria periodica, Botryopteris cratis, Anachoropteris involuta (adaxial shoots), A. cadyii 12 species range through W/S boundary (66.6% persistence) Conostoma kestospermum, C. platyspermum, Schopfiastrum decussatum, Botryopteris 'pseudoantiqua', B. forensis, Anachoropteris gillotii, Sermaya biseriata, Zygopteris illinoiensis, Ankyropteris brongniartii, Medullosa endocentrica, Cordaixylon dumusum, Paurodendron fraipontii 25 total species of ground cover, vines and shrubs in early Stephanian 13 species originate in early Stephanian (52.0% origination) Callistophyton poroxyloides, Conostoma quadrivasatum, C. quadratum, Coronostoma villosum, Anachoropteris involuta (lateral shoots), A. clavata, Zygopteris berryviltensis, Rhabdoxylon americanum, Apotropteris minuta, Ankyropteris n. sp., Phillipopteris globoformis, Norwoodia angustum, Chaloneria cormosa Table 8. Summary extinction and origination patterns at Westphalian-Stephanian boundary based on coal-ball data Extinctions Originations Trees Ground cover Total 86.7% 33.3% 66.7% 76.5% 52.0% 61.9% associations with stems and leaves are unknown. Range-through data were used. This means that a species was assumed to be present in a given coal if it was found in both younger and older coals; range-through data compensate for differ- ences in sampling intensities among coals. Coal balls have not been found in the last few coals of the Westphalian. However, palynological data indicate persistence of the same basic flora up to the end of the Middle Pennsylvanian (Peppers 1985; Kosanke 1988; Kosanke & Cecil 1989). Tables 3 to 8 summarize the species extinction and origination patterns between the late West- phalian and early Stephanian; data in each table are from the same stratigraphic interval. Our ecological analyses indicate that late Westpha- lian and early Stephanian coals represent two distinct types of ecosystems, with ecological homogeneity characteristic of each interval. Coal palynological data (Table 3, from Peppers 1985) sample mire ecosystems, the same environments as coal balls. Pollen and spore extinctions although elevated at the West- phalian-Stephanian boundary are lower than those found using coal balls. Many pollen and spore 'species' have been documented to come from more than one whole-plant species (e.g. Good 1975; Lesnikowska 1989; Willard 1989; DiMichele & Bateman 1992) and thus may represent multi-species clades, which would reduce apparent extinction levels. Table 4 summarizes turnover in compression- impression species listed in Gillespie & Pfeffer- kom (1979); they used only species that occurred widely and abundantly in Euramerica, so the sample is not comprehensive. Summary extinc- tion levels are near 50% for this sample and indicate that losses of dominant species were not simply confined to mires, but affected the entire tropical lowlands. Ecomorphic extinction bias Origination and extinction patterns for coal-ball plants were calculated from data in Table 5. Coal-ball data are divided into ecomorphic groups (Tables 6 and 7) and demonstrate that trees suffered proportionately nearly three times greater extinction than shrubs, ground cover, and vines. Despite such higher losses, tree origination in Stephanian mires was only 25% greater than that for the other groups. Thus the extinction was composed selectively of trees in all major phylogenetic groups. Nearly all the compression-impression taxa reported were trees. Coal-ball data are summarized in Table 8. The post-extinction system: early Stephanian mire communities Stephanian mires typically were composed of 25-35 species. Proportionally more of these species were ground-cover and vines than in Westphalian assemblages; 25 of 42 species in our Stephanian samples compared with 19 of 50 species in our WestphaHan samples (Table 1). Trees continued to be the major biomass components of the mires, although tree dom- EXTINCTION AND RECOVERY IN PENNSYLVANIAN COAL SWAMPS 211 inance-diversity patterns were distinct from those of the WestphaHan. In general, early Stephanian mires had stereotypic quantitative composition at the generic level; the most abundant peat producers were Psaronius tree ferns, meduUosan pteridosperms, and sigillarian lycopsids. Data from both macrofossils (Lesni- kowska 1989; Willard & PhiUips 1993) and microfossils (Peppers 1985; Helfrich & Hower 1989; Willard & Phillips 1993) suggest, however, that species dominance patterns varied quite significantly through time from one coal seam to the next, but not in any directional fashion (Peppers 1985) or in association with clearly identifiable changes in the physical environment of the tropics. Ordination of 11 zones from two coal-ball profiles from the Calhoun coal of Illinois summarizes a pattern typical of early Stephanian (Missourian) mires (Fig. 3). Coal-ball and palynological analyses have revealed similar patterns in the Friendsville and Bristol Hill coals of lUinois (Willard & Phillips 1993), the Otter Creek coals of Kentucky (Helfrich & Hower 1989), the Redstone coal of West Virginia (Grady & Eble 1990), and the Duquesne coal of Ohio (Pryor 1993). Overall, these analyses reveal significantly less taxonomic and dom- inance-diversity distinction among assemblages, and less clear ties to recurrent habitat factors than found in Westphalian mires. In a broad sense, Stephanian mire communities seem to be divisible into three or four principal assem- blages. Most are heavily dominated by or enriched in tree-ferns. Of secondary importance are medullosans, and less commonly still are sigillarians, with variable amounts of tree ferns in either of these assemblages. Sigillaria brardii was a polycarpic lycopsid that typically occurred in clastic substrate habitats; in Stephanian mires sigillarian stands were most common during early phases of peat accumulation and thus enrich lower zones of some coal-ball and palynological profiles (Grady & Eble 1990; Grady et al. 1992; Willard & Phillips 1993; Helfrich & Hower 1989) in closest association with the clastic seat earth. Calamites can occur in association with any of the other assemblages, although they tend to be most abundant in the lower parts of the coal beds or in association with clastic partings (Grady & Eble 1990; Pryor 1993), a distribution similar to that found in the Westphalian. There are additional elements of many Ste- phanian mires not found in the Calhoun coal samples from Berryville. Most importantly are assemblages enriched in the lycopsid Chaloneria, the source of Endosporites microspores. Chalo- neria was abundant in many coals, beginning in the earliest Stephanian, immediately after the extinctions of the latest WestphaHan (Phillips et al. 1974; Peppers 1985). Megafossils or spores of Chaloneria are most common in the lower parts of coal beds or in association with clastic partings (Helfrich & Hower 1989; Grady & Eble 1990; Willard & Phillips 1993; Pryor 1993). Vegetation dominated by Chaloneria probably was physiognomically distinctive; the plant appears to have been short, maybe a metre or so in height, with a pole-like habit (Pigg & Rothwell 1983) and formed the Carboniferous equivalent of 'marshes' (DiMichele el al. 1979). Cordaites also have been found to be important components of some Stephanian mires (e.g., the Duquesne coal of Ohio, Rothwell 1988; Pryor 1993; the Friendsville coal of Illinois, Willard & Phillips 1993), where they occur most commonly in lower zones, or in proximity to clastic partings (Trivett & Rothwell 1985; Pryor 1993). Using the Duquesne coal as an example, Pryor (1993) described recurrent assemblages as a series of 'successional' stages, beginning with cordaites and sigillarians, proceeding through marsh-like assemblages dominated by Chaloneria, and ending with progressively great- er dominance of medullosans and ultimately Psaronius tree ferns. The progression appears to have been driven largely by temporal changes in edaphic conditions as an increasingly thick, nutrient-poor peat substrate developed. Similar 'successional' patterns have been described by Grady & Eble (1990) from palynological studies of the Redstone coal, where they correlate with directional changes in ash and inertinite content of the coal. The successional patterns retain the characteristic Stephanian pattern of generic rather than specific identity. The intermediate and ultimate tree-fern and meduUosan species composition seems to be relatively unpredict- able. Short gradient lengths are typical of most Stephanian mires. The abundance of tree ferns imparts a similar ecomorphic aspect to most of the assemblages, and the generally uniformly high biomass of ground cover and vines suggests that substrates remained exposed for extended periods of time. The vagaries of density inde- pendent disturbance and subsequent coloniza- tion patterns appear to be major controls on local taxonomic composition. When compared at the generic level, Stepha- nian coal-ball floras are similar to contempora- neous compression-impression floras from clastic, floodbasin habitats (compare data in Pfefferkorn & Thomson 1982; Phillips et al. 1985; DiMichele & Aronson 1992). This presents 212 W. A. DIMICHELE & T. L. PHILLIPS Fig. 3. Detrended correspondence analysis of coal-ball zones from Berryville VS 1 and VS 2, early Stephanian Calhoun coal. Lines delimit groups of coal-ball zones with similar dominance patterns; letters refer to groups in Table 1 where dominance and life-history patterns are summarized. a sharp contrast to Westphalian patterns, and is indicative of a general homogenization of the tropical wetland vegetation in the early Stepha- Comparison of pre-extinction and postextinction landscapes Life histories The patterns of WestphaUan and Stephanian mire communities cannot be compared in a one to one manner. Foremost, the biologies and life histories of the respective dominant taxa, lycopsids and tree ferns, were markedly differ- ent. The lycopsids were structurally diverse and had ecological strategies spanning the spectrum from weedy opportunists to site-occupying K- strategists (DiMichele & PhiUips 1985; Phillips & DiMichele 1992), although direct comparisons to any extant forms are difiicult. Tree ferns as a group, on the other hand, had broad ecological amphtudes, reflecting their massive reproduction and cheap construction; trees were non-woody, and were supported by adventitious root man- tles composed mostly of air spaces (Ehret & Phillips 1977; Lesnikowska 1989). There is no evidence that any species of Psaronius were primary colonizers of the deep, standing-water habitats occupied by some species of Westpha- han lycopsids. Palynological patterns from WestphaUan coals indicate differential distribu- tions of tree-fern species relative to contempora- neous channels, although the factors controlling these distributions are open to speculation at present (Willard 1993; Mahaffy 1988). Tree ferns of Stephanian mires retained mass reproduction of small, highly dispersible spores, reflecting the opportunistic Ufe histories of their ancestors. However, they also attained large sizes, estimated from organ diameters (stems, roots, petioles) and in situ trunk diameters of > 1 m; maximum organ dimensions are larger than any reported from WestphaHan mires (Lesnikowska 1989). Clastic deposits of the late WestphaUan do preserve large tree fern trunks (Pfefferkorn 1976; Lesnikowska 1989) suggest- ing that many Stephanian mire species may have descended from Westphalian clastic-swamp taxa. The dominance of tree ferns in mires mirrored their dominance in the clastic wetlands of the Stephanian, reducing the dominance- diversity disparity between mires and clastic wetlands that typified the WestphaUan. Willard and PhilUps (1993) report large Psaronius trunks that apparently persisted through most of the time of accumulation of a Stephanian coal-baU mass over 0.6 m thick that had replaced nearly the entire seam thickness. Mickle (1984) reports large diameter trunks in which the stem had rotted out at the base, implying that the stem could die back at the base as it grew from the top, suppUed with water and nutrients by the living root mantle. These observations imply that large, site occupying Psaronius trees were components of Stephanian mires. Size and evidence of site occupation suggest that the more r-selected, opportunistic species of the WestphaUan had been replaced by, or given rise to, more typically K-selected, site occupying forms. EXTINCTION AND RECOVERY IN PENNSYLVANIAN COAL SWAMPS 213 Table 9. Richness and abundance patterns of ground cover and Herrin coal and Stephanian Calhoun coal vines by coal-ball zone in profiles of Westphalian Coal Age Coal-ball Profiles n/P S/Z %B/Z Calhoun Herrin Herrin Stephanian Westphalian WestphaUan Berryville VS 1 & 2 Old Ben 3 & 5 Sahara 4 & 5 10 9 13 5.3 1.8 3.3 4.2 3.0 4.3 n/P = total species richness of profiles, S/Z = average species richness per profile zone, %B/Z = average percent of total biomass per zone Ground cover and vines only. Ecological gradients Stephanian mires had markedly shorter ecologi- cal gradients than late Westphalian peat-form- ing ecosystems (Fig. 4). This is suggested by detrended correspondence analyses (Fig. 3) in which axis length reflects directly the extent of species turnover, four units representing approximately complete turnover between end- points (Gauch 1982). Short Stephanian gradi- ents are in part an artifact of lower species-level taxonomic resolution of tree fern and pterido- sperm litter, a problem with coal-ball samples throughout the Pennsylvanian. This is mitigated, however, by the generally low species diversity of the dominant tree ferns and pteridosperms in any one deposit, determined by the relative abundances of diagnostic organs such as Scolecopteris, Psaronius stems, or pteridosperm ovules. Overall, the Stephanian gradient, the hfe histories of the plants (many requiring exposed substrates for hfe-cycle completion), and dom- inance-diversity patterns suggest that many of the physical habitats colonized and stabilized by Westphalian plants, particularly lycopsids, either did not occur or were unoccupied in Stephanian mires. Ground cover and vines Comparisons of ground cover and vine abun- dances are summarized in Table 9 for four coal- ball profiles from the Westphalian D Herrin coal and two profiles from the Stephanian Calhoun coal. Average biomass of ground cover is not significantly different between the Westphalian and Stephanian samples, nor is the total number of ground-cover species found within any of the profiles. However, the average number of species per coal-ball zone was markedly higher in the Stephanian samples. This difference reflects greater variabihty among WestphaHan assem- blages, where ground cover is lacking in many cases. This is consistent with longer ecological gradients in the WestphaUan, and is reflected in greater spread of the data in ordinations of coal- ball profile zones. During the Stephanian the same species recur from one zone to the next in profiles, leading to a more uniform taxonomic signature in the groundcover across assem- blages. Disturbance Disturbance levels within Pennsylvanian mires are difficult to quantify. There appear to have been several major disturbance agents: fire, indicated by fusain (mineral charcoal), severe floods, indicated by clastic partings or elevated levels of mineral matter in the coal, generally unfavorable climatic conditions, indicated indir- ectly by plant size, average thickness of coal seams. Disturbance levels may have differed between late Westphalian and early Stephanian mires, although evidence is indirect. Fusain (mineral charcoal) abundances are lower in the Stephanian based on coal-ball analyses (Phillips et al. 1985). Tree sizes are generally larger in the Stephanian, determined from maximum dimen- sions of isolated organs. 'Marshes' composed of small stature vegetation, such as Chaloneria, are proportionally more abundant in the Stepha- nian. Finally, Stephanian coals generally are considerably thinner than those of the late Westphalian. These factors present an ambig- uous picture of disturbance. In general, how- ever, they suggest that individual Stephanian mires may have had lower levels of serious disturbance than prevailed during the Westpha- Uan, permitting the development of larger plants in virtually all Uneages. However, the general thinness of coal seams, and the frequent occurrence of marsh-Uke vegetation suggests that climates and depositional settings conducive to peat-formation were shorter Uved during the Stephanian Stephanian gigantism Plants of thick peat substrates, particularly those minimally influenced by the influx of elastics. 214 W. A, DIMICHELE & T. L. PHILLIPS 1 -Lepidophloios Diaphorodendron Synchysidendron Anabalhra SigiUaria Pteridosperms Tree ferns 8 - Sphenopsids WESTPHALIAN Fig. 4. Hypothetical intramire gradients in Westphalian and Stephanian. Wettest, standing water environments on left, intermittently flooded peat substrates in centre, clastic-peat transitional environments grade to 'levee' environments on left. Plants from similar lineages are shown to be larger in the Stephanian. appear to have been smaller than their relatives on mineral substrates (Pfefferkom 1976; Di- Michele & DeMaris 1987; Gastaldo 1986), which is consistent with patterns in modern tropical mires (Anderson 1964, 1983). In the Westphahan, stem diameters of the dominant lycopsids recorded from coal balls are estimated to be 25-35 cm in Diaphorodendron and Lepi- dophloios respectively (DiMichele 1979, 1981), whereas trunks > 1 m diameter are encountered commonly in shales above or associated with coals. The only other Westphalian plants with moderate to large stem diameters were cordaites (Pennsyhanioxylon and Mesoxylon, up to 20 cm diameter) and calamites {Arthropitys, up to 12 cm diameter) from Westphahan B coals of the Appalachians. During the Westphahan, therefore, lycopsids may have been the only tall trees in what appear to have been stunted mire forests. Excluding the Westphalian B exceptions, maximum organ diameters (stems, roots, pe- tioles) of subdominant taxa in mires were less than 10 cm, mostly less than 5 cm (e.g. Andrews & Agashe 1965; Costanza 1985; Lesnikowska 1989). Stephanian coal balls reveal increases in the maximum sizes of tree ferns, pteridosperms, and sphenopsids, based on measurements of isolated organs. Large plant stature in Stephanian mires has been alluded to (Andrews & Agashe 1965; Galtier & PhilUps 1985) but has not been examined systematically. Large size also has been documented for some Stephanian clastic compression ^taxa, such as Calamites gigas (Barthel & Kerp 1992). Size increases within broad clades cannot be constrained more narrowly by phylogenetic analysis at the present time due to the incompar- ability of anatomically preserved and compres- sion-impression based species. Lacking this framework, we suggest several possible causative mechanisms, perhaps interrelated. The simplest explanation is that trees of larger stature migrated into and occupied mire habitats following the extinction event, in the absence of competitors. Large size has been documented in several Westphahan groups dominant in clastic substrate habitats, including tree ferns (Pfefferkorn 1976) and pteridosperms (Wnuk & Pfefferkom 1984; Pfefferkom et at. 1984). In this EXTINCTION AND RECOVERY IN PENNSYLVANIAN COAL SWAMPS 215 case large stature would be a happenstance of the ancestry of the colonizing species. Sugges- tions that Stephanian mires were largely topo- geneous (Grady & Eble 1990; Donaldson & Eble 1991) may have permitted the survival of species from, or derived from ancestors in, clastic substrate areas. Harvey & Dillon (1985) found higher levels of inertinites in Stephanian coals, suggesting higher nutrient levels either permit- ting attainment of larger size or the migration into mires of species requiring higher levels. Because of the edaphic qualities of peat substrates, and the resultant 'island' nature of mires within the greater lowland wetlands, simple migration of species from clastic to peat substrates may have been inhibited by physical factors alone. In such a landscape mires may have presented opportunities much hke modem islands, on which small to medium stature plants often undergo phyletic size increases. Lineages of ancestrally small plants show consistent trends to larger size in island environments where competition for resources is reduced (Carlquist 1974). In a highly disrupted post- extinction tropics, the edaphic-island nature of mires, in combination with greatly reduced species diversities, may have permitted the rapid evolution of larger size in many surviving clades. Given the extensive extinction of trees, the opportunity for island-type, low-competition, opportunistic dynamics should have been en- hanced in the post-extinction period. A final consideration is a general reduction in the amount of severe, intramire disturbance during the Stephanian. Under such conditions natural selection may have permitted trees to achieve larger sizes, ultimately driven by intrin- sic developmental factors and evolutionary size increase. Discussion Extinction thresholds and ecological disequilibrium The major vegetational change following the end-Westphahan extinctions suggests that high levels of species extinction, especially when concentrated in framework trees, leads to loss of ecosystem self-regulatory properties. Exam- ination of the dynamics of the Westphalian indicates that persistence, both at the level of assemblages (communities) and multi-assem- blage landscapes (DiMichele & Phillips 1995) was the dominant theme. Species turnover occurred between every coal bed, based on palynological (Peppers 1985) or coal-ball (Phil- lips et al. 1985) data. Yet, the ecomorphic aspects of the major assemblages remained recognizably the same throughout the entire WestphaHan. Species tended to be replaced ecologically by congeners with similar mor- phology and ecological preferences, although whether from migration into mires from an external source or by in situ speciation we cannot discern. These data indicate that as long as species turnover was low, the remaining species created a selective filter that only permitted certain forms to colonize vacated resource space. High levels of extinction, and associated major vegetational change suggest thresholds beyond which the remaining species cannot regulate subsequent colonization of available resources. The largest extinction within the Westphalian is considerably smaller than that at the end. It occurred near the A-B boundary and is marked by the disappearance of Lyginopteris and Neuralethopleris, as well as other minor plants. However, the lycopsid trees, which were the major framework plants of mires, survived. Species turnover based on coal balls is approxi- mately 30%; the coal-ball record of this interval is sparse, but there do appear to have been some changes in the vegetational organization of mires after the event. In particular, cordaites appeared as elements of 'ecotonal' assemblages, occupying heavily decayed peat surfaces, often in association with medullosans (e.g.. Rock Springs coal of Tennessee, Phillips et al. 1985). Early cordaites may have extended the length of the Westphalian mire gradient by colonizing severely decayed peat surfaces, enlarging the scope of the 'ecotonal' parts of mires. Overall, though, the cordaites became a part of mires without disrupting the basic landscape structure. The Westphalian A-B extinction and its ecolo- gical consequences indicate that significant, but smaller extinctions, where the major tree forms survive, do not have the same catastrophic eifect on ecosystem structure and composition as extinctions that remove the basic framework trees. Coal-ball macrofossils do not occur in coals immediately above the Westphalian-Stephanian boundary. Consequently palynology provides the most complete record of the ecological events within mires that follow immediately after the major extinctions. With the loss of numerous species the environments of earliest Stephanian mires appear to have been open to colonization by new species or by expansion of surviving, previously subdominant lineages. Most notably, 'marshes' dominated by Chalo- neria became major parts of some thin, boney coals. Through seven coals in the upper half of 216 W. A. DIMICHELE & T. L. PHILLIPS the Modesto Formation of Illinois coal-to-coal variability in spore-species dominance is high; included as occasional dominants on a whole- seam basis are Chaloneria, Sigillaria, and tree ferns producing Punctatisporites obliquus and P. minutus spores. The dominance patterns stabilize in the overlying Bond Formation, where Sigil- laria, several tree fern species, and calamites begin to occur with greater consistency (Peppers 1985). The coal-ball record of the later Stepha- nian A consistently records tree-fern dominance with subdominant pteridosperms (Phillips et at. 1985; WiUard & Phillips 1993: Pryor 1993). It would appear from the palynological patterns during the early post-extinction interval that either chmate or mire edaphic conditions were highly variable, and generally not suitable for the development of major peat bodies. Each successive mire was dominated by a different suite of species, including many that were subtrees, such as Chaloneria, which formed marsh-like vegetation. The mutualistic relation- ships that characterized the Westphahan had evolved over millions of years, beginning in the latest Devonian, permitting persistent vegeta- tional patterns to appear. Disruption of this system apparently eliminated these kinds of interactions, and thus long-persistent vegeta- tional patterns. As a result of low species diversity and Uttle ecological structure (particu- larly evidence of strong resource partitioning), early Stephanian mire assemblages were highly invasible by species that could tolerate the rigours of low-nutrient, physically stressful, peat substrates. They offered opportunities for newly evolved species to establish under reduced resource competition and exclusion. Rise to dominance of opportunists The ultimate dominance of marattialen tree ferns reflects the rise of cheaply constructed plants with massive reproductive output of highly dispersible isospores (capable of estab- hshing a new population from one spore under some circumstances). Westphalian marattialeans had what can be described in general terms as an opportunistic, fugitive life history strategy. Trees were generally of small stature; tree habit was made possible by a mantle of adventitious roots composed mostly of airspaces. The foliage bore large numbers of sporangia, in which were produced collectively billions of spores. Ferns also recovered and dominated many landscapes after the Cretaceous-Tertiary extinc- tions, the now-famous 'fern-spike' (Tschudy et al. 1984). This brief period of dominance was not followed, however, by development of a fern-dominated vegetation. The K-T pattern may differ from the Westphalian-Stephanian pattern for several reasons. First, it is likely that the palynological fern-spike was produced by small, ground cover ferns rather than tree ferns. The potential of these plants to give rise to trees was limited. Second, it is clear that many arboreous hneages of seed plants survived the extinction as local populations. These popula- tions ultimately fuelled the re-estabhshment of vertically stratified woodlands and forests (Wing & Tiflfney 1987; Wolfe & Upchurch 1986, 1987; Hickey 1984; Johnson 1992; Johnson & Hickey 1990). Thus, sources of seed were available from which new ecosystems could begin to develop. In contrast, Stephanian lowlands appear to have been regionally devastated by climate changes that, even if brief geologically, would have been long-term on an ecological timescale. Additionally, whole Westphalian clades were very narrowly distributed ecologically, particu- larly those centered in the lowland wetlands (rhizomorphic lycopsids, meduUosan pterido- sperms, and calamites) restricting the abihty of many to survive as isolated populations. The heterosporous reproductive adaptations of many of the more advanced lycopsids strongly limited them to aquatic and semi-aquatic habi- tats (Phillips 1979), which would have con- strained their ability to recolonize patchy open habitat spaces not connected by continuous wetlands. Finally, marattialians were trees, even if relatively small trees, which gave them the ability to spawn forms capable of canopy formation, and dominance of a stratified eco- system. As disaster forms, marattaUan tree ferns were thus capable of locating available resources rapidly and then dominating local landscapes in the face of other tree taxa through rapid growth, tolerance of low nutrient conditions, and abun- dant reproduction on site. Implications for ecosystem dynamics Vegetational dynamics of the Pennsylvanian tropics, and, in fact, of most times in the Earth's history that have been studied in detail (e.g. the Tertiary, Wing 1982; Burnham 1983), demon- strate a greater degree of congruence with the predictions of 'Clementsian' ecosystem theories than with 'Gleasonian' (individualistic) models. As a generahty, neo-plant ecologists have embraced Gleasonian principles and rejected Clementsian concepts, particularly as originally formulated (Clements 1916). Pointing to Pleis- tocene palynological data (e.g. Webb 1988; Delcourt & Delcourt 1987; Overpeck et al. 1992), the 'fossil record' has been found to be EXTINCTION AND RECOVERY IN PENNSYLVANIAN COAL SWAMPS 217 in full support of the Gleasonian view. The differences between the Pleistocene patterns and those of deeper time do not reside in sampling scale; it is possible to resolve ecological time from palynological sampling nearly as finely in the Pennsylvanian as in the Pleistocene (e.g. Eble & Grady 1990; Willard 1993; Mahafify 1985; Pierce et al. 1993). Such incompatibilities as exist may reflect real dynamical differences between the Pleistocene ice ages and the vast bulk of Earth history, differences between temperate and tropical dynamics, different scales of analysis, or different philosophies about what constitutes a vegetational unit of interest. Data from the pre- Pleistocene fossil record suggest a need for more serious consideration of modern attempts at Clementsian-type syntheses. The view of ecosys- tems as systems, with hierarchical organization and emergent properties above and beyond those of the component populations, has re- mained an integral part of ecology (Patton & Odom 1981; Salthe 1985; O'Neill 1989; O'Neill et al. 1986; Allen & Hoekstra 1992). However, due to the inherent complexity of ecosystems the theoretical elegance of Gleasonian individuaUsm has yet to emerge. A complete theory of ecosystem organization must explain patterns on all scales of spatio-temporal observation; even if a general hierarchical model with the simplicity of the Gleasonian view has not come to fruition, the fossil record suggests a need for it, and for the rejection of individualistic models as all-powerful explanatory defaults. The data from the Pennsylvanian represent an entirely different ecosystem from any that existed afterward. Dominance of substantial parts of the landscape by lower vascular plants, the development of ecological partitioning in concert with the modernization of the vascular plant flora, and the high degree to which the plants were structurally adapted to particular, narrow parts of the landscape, present ecologists with a natural test system for broader theories. To the degree that patterns of temporal beha- viour in Pennsylvanian systems are congruent with those predicted for and documented in modern systems, our level of confidence in general principles is enlarged. We thank Dan S. Chaney for assistance in preparation of the technical figures, and Mary Parrish for the artwork. Numerous people have contributed to this paper through discussions over the past 20 years; we particularly wish to thank Richard Bateman, Alicia Lesnikowska, James Mahaffy, Debra Willard and Richard Winston. TLP acknowledges the support of NSF grant EAR 8313094. This is contribution No. 24 from the Evolution of Terrestrial Ecosystems Con- sortium, which provided partial support for the research. References ALLEN, T. F. H. & HOEKSTRA, T. W. 1992. Toward a Unified Ecology. Columbia University Press, New York. ANDERSON, J. A. R. 1964. The structure and develop- ment of the peat swamps of Sarawak and Brunei. Journal of Tropical Geography, 18, 7-16. 1983. The tropical peat swamps of western Malesia. In: GORE, A. J. P. (ed.) Ecosystems of the World, 4B; Mires: Swamp, Bog, Fen and Moor. Elsevier, Amsterdam, 181-199. ANDREWS, H. N. & AGASHE, S. N. 1965. Some exceptionally large calamite stems. Phyto- morphology, 15, 103-108. BARTHEL, M. & KERP, H. 1992. Catamites gigas - an alternative development within Permian articu- lates. Organisation Internationale de Paleobotani- que, IVeme Conference Abstracts, OFP Informa- tions No. Special 16-B: 18. BEELER, H. E. 1983. Anatomy and frond architecture of Neuropteris ovata and Neuropteris scheuchzeri from the Upper Pennsylvanian of the Appala- chian Basin. Canadian Journal of Botany, 61, 2352-2368. BoucoT, A. J. 1983. Does evolution take place in an ecological vacuum? Journal of Paleontology, 57, 1-30. 1990. Community evolution: its evolutionary and biostratigraphic significance. In: MILLER, W. Ill (ed.) Paleocommunity Temporal Dynamics: the Long-term Development of Multispecies Assem- blages. Paleontological Society Special Publica- tion, 5, 48-70. BRETT, C. E. & BAIRD, G. C. 1995. Co-ordinated stasis and evolutionary ecology of Silurian to Middle Devonian faunas in the Appalachian Basin. In: ERWIN, D. H. & ANSTEY, R. L. (eds) New approaches to speciation in the fossil record. Columbia University Press, New York, 283-315. , MILLER, K. B. & BAIRD, G. C. 1990. A temporal hierarchy of paleoecologic processes within a Middle Devonian epeiric sea. In: MILLER, W. Ill (ed.) Paleocommunity Temporal Dynamics: the Long-term Development of Multispecies As- semblages. Paleontological Society Special Pub- lication, 5, 178-209. BURNHAM, R. J. 1983. Diversification and stable distribution with respect to temperature in early Tertiary Ulmoideae of western North America. Abstracts, Botanical Society of America Annual Meeting, 68. CALDER, J. H. 1993. The evolution of a groundwater influenced (WestphaUan B) peat-forming eco- system in a piedmont setting: the No. 3 seam, Springhill coalfield, Cumberland Basin, Nova Scotia. Geological Society of America, Special Paper, 286, 153-180. CARLQUIST, S. 1974. Island Biology. Columbia Uni- versity Press, New York. CECIL, C. B. 1990. Paleoclimate controls on strati- 218 W. A. DIMICHELE & T. L. PHILLIPS graphic repetition of chemical and clastic rocks. Geology, 18, 533-536. , STANTON, R. W., NEUZIL, S. G., DULONG, F. T., RuppERT, L. F. & PIERCE, B. S. 1985. Paleoclimatic controls on late Paleozoic sedimen- tation and peat formation in the central Appa- lachian Basin (USA). International Journal of Coal Geology, 5, 195-230. CHANEY, R. W. 1947. Tertiary centres and migration routes. Ecological Monographs, 17. CLEMENTS, F. E. 1916. Plant Succession, an Analysis of the Development of Vegetation. Publication No. 242, Carnegie Institution, Washington, DC. CLYMO, R. S. 1987. Rainwater-fed peat as a precursor of coal. In: SCOTT, A. C. (ed.) Coal and Coal- Bearing Strata: Recent Advances. Geological Society, London, Special Publication, 32, 17-23. CoLLiNSON, M. B. & SCOTT, A. C. 1987a. Factors controlling the organization and evolution of ancient plant communities. In: GEE, J. H. R. & GiLLER, P. S. (eds) Organization of Communities Past and Present. Blackwell, Oxford, 399^20. & 1987A. Implications of vegetational change through the geological record on models for coal-forming environments. In: SCOTT, A. C. (ed.) Coal and Coal-Bearing Strata: Recent Advances. Geological Society, London, Special Publication, 32, 67-85. COSTANZA, S. H., 1985. Pennsylvanioxylon of Middle and Upper Pennsylvanian coals from the Illinois Basin and its comparison with Mesoxylon. Pa- laeontographica, Abt. B, 197, 81-121. DELCOURT, P. A. & DELCOURT, H. R. 1987. Long-term Dynamics of the Temperate Zone. Springer, New York. DEMARIS, P. J., BAUER, R. A. CAHILL, R. A. & DAMBERGER, H. H. 1983. Prediction of Coal Balls in the Herrin Coal. Illinois State Geological Survey, Contract-Grant Report 1982-3. DIMICHELE, W. A. 1979. Arborescent lycopods of Pennsylvanian age coals: Lepidophloios. Palaeon- tographica, Abt. B, 171, 57-77. 1981. Arborescent lycopods of Pennsylvanian age coals: Lepidodendron, with description of a new species. Palaeontographica, Abt. B, 175, 85- 125. 1994. Ecological patterns in time and space. Paleobiology, 20, 89-92. & ARONSON, R. B. 1992. The Pennsylvanian- Permian vegetational transition: a terrestrial analogue to the onshore-offshore hypothesis. Evolution, 46, 807-824. & BATEMAN, R. M. 1992. Diaphorodendraceae, fam. nov. (Lycopsida: Carboniferous): systema- tics and evolutionary relationships of Diaphoro- dendron and Synchysidendron, gen. nov. American Journal of Botany, 79, 605-617. & DCMARIS, P. J. 1987. Structure and dynamics of a Pennsylvanian-age Lepidodendron forest: colonizers of a disturbed swamp habitat in the Herrin (No. 6) coal of Illinois. Palaios, 2, 146- 157. ?, MAHAFFY, J. F. & PHILLIPS, T. L. 1979. Lycopods of Pennsylvanian age coals: Polysporia. Canadian Journal of Botany, 57, 1740-1753. & PHILLIPS, T. L. 1985. Arborescent lycopod reproduction and paleoecology in a coal-swamp environment of late Middle Pennsylvanian age (Herrin Coal, Illinois, USA). Review of Palaeo- botany and Palynology, 44, 1-26. & 1988. Paleoecology of the Middle Pennsylvanian-age Herrin coal swamp (Illinois) near a contemporaneous river system, the Walsh- ville paleochannel. Review of Palaeobotany and Palynology, 56, 151-176. & 1994. Paleobotanical and paleoeco- logical constraints on models of peat formation in the Late Carboniferous of Euramerica. Palaeo- geography, Palaeoclimatology, Palaeoecologv 106, 39-90. & 1995. The response of hierarchically structured ecosystems to long-term climate change: a case study using tropical peat swamps of Pennsylvanian age. In: STANLEY, S. M., KNOLL, A. J. & KENNETI, J. P. (eds) Effects of past global change on life. National Research Council, Studies in Geophysics. 134-155. ??. & MCBRINN, G. E. 1991. Quantita- tive analysis and paleoecology of the Secor coal and roof-shale floras (Middle Pennsylvanian, Oklahoma). Palaios, 6, 390-409. - & PEPPERS, R. A. 1985. The influence of climate and depositional environment on the distribution and evolution of Pennsylvanian coal- swamp plants. In: TIFFNEY, B. H. (ed.) Geological Factors and the Evolution of Plants. Yale Uni- versity Press, New Haven, 223-256. DONALDSON, A. C. & EBLE, C. F. 1991. Pennsylvanian coals of central and eastern United States. In: GLUSKOTER, H. J.. RICE, D. D. & TAYLOR, R. B. (eds) Economic Geology, U.S. Geological Society of America, The Geology of North America P-2 523-546. EBLE, C. F. 1990. A palynological transect, swamp interior to swamp margin, in the Mary Lee coal bed. Warrior Basin, Alabama. In: GASTALDO, R. A., DEMKO, T. M. & LIU YUEJIN (eds) Carbonifer- ous Coastal Environments and Paleocommunities of the Mary Lee Coal Zone, Marion and Walker Counties, Alabama. Guidebook for Fieldtrip 6, 39th Annual Meeting, Southeastern Section, Geological Society of America. Alabama Geolo- gical Survey, Tuscaloosa, 65-81. & GRADY, W. C. 1990. Paleoecological interpretation of a Middle Pennsylvanian coal bed from the Central Appalachian Basin, U.S.A. International Journal of Coal Geology, 16 255- 286. , HowER, J. C. & ANDREWS, W. M. 1994. Paleoecology of the Fire Clay coal bed in a por- tion of the Eastern Kentucky Coal Field. Palaeo- geography, Palaeoclimatology, Palaeoecology, 106, 287-307. EGGERT, D. L., CHOU, C.-L., MAPLES, C. G., PEPPERS, R. A., PHILLIPS, T. L. & REXROAD, C. B. 1983! Origin and economic geology of the Springfield Coal Member in the Illinois Basin. Guidebook to Fieldtrip No. 9, Geological Society of America EXTINCTION AND RECOVERY IN PENNSYLVANIAN COAL SWAMPS 219 Annual Meeting, 121-146. EHRET, D. L. & PHILLIPS, T. L. 1977. Psaronius root systems - morphology and development. Palae- ontographica, Abt. B., 161, 147-164. FENG, B.-C. 1989. Paleoecology of an upper Middle Pennsylvanian coal swamp from western Penn- sylvania, U.S.A. Review of Palaeobotany and Palynology, 57, 299-312. FRAKES, L. E., FRANCIS, J. E. & SYKTUS, J. I. 1992. Climate Modes of the Phanerozoic. Cambridge University Press. GASTALDO, R. A. 1986. An explanation for lycopod configuration, 'Fossil Grove', Victoria Park, Glasgow. Scottish Journal of Geology, 11, 11-83. GALIIER, J. & PHILLIPS, T. L. 1985. Swamp vegetation from Grand 'Croix (Stephanian) and Autun (Autunian), France and comparisons with coal ball peats of the Ilhnois Basin. Compte Rendu, IX International Congress Carboniferous Stratigraphy and Geology, 4, 13-24. GAUCH, H. G. 1982. Multivariate Analysis in Commu- nity Ecology. Cambridge University Press. GILLESPIE, W. M. & PFEFFERKORN, H. W. 1979. Distribution of commonly occurring plant mega- fossils in the proposed Pennsylvanian System Stratotype. In: ENGLAND, K. J. ET AL. (eds) Proposed Pennsylvanian System Stratotype, Vir- ginia and West Virginia. Ninth International Congress of Carboniferous Stratigraphy and Geology, Guidebook Field Trip No. 1, 87-96. GOOD, C. W. 1975. Pennsylvanian-age calamitean cones, elater-bearing spores, and associated vege- tative organs. Palaeontographica, Abt. B, 153, 28- 99. GRADY, W. C. & EBLE, C. F. 1990. Relationships among macerals, minerals, miospores and paleo- ecology in a column of Redstone Coal (Upper Pennsylvanian) from north-central West Virginia (U.S.A.). International Journal of Coal Geology, 15, 1-26. , FEDORKO, N. & EBLE, C. F. 1992. Paleoecol- ogy and paleoenvironments of peat formation in the Pittsburgh coal bed, northern Appalachian Basin, U.S.A. Abstracts, Geological Association of Canada Annual Meeting. Geoscience Canada, 17 (supplement), 42. Guo, YiNGTiNG. 1990. Palacoecology of flora from coal measures of Upper Permian in western Guizhou. Journal of China Coal Society, 15, 48- 54. HARVEY, R. D. & DILLON, J. W. 1985. Maceral distributions in Illinois coals and their paleoenvir- onmental implications. International Journal of Coal Geology, 5, 141-165. HECKEL, P. H. 1986. Sea-level curve for Pennsylvanian eustatic marine transgressive-regressive deposi- tional cycles along midcontinent outcrop belt, North America. Geology, 14, 330-334. 1989. Updated Middle-Upper Pennsylvanian eustatic sea level curve for midcontinent North America and preliminary biostratigraphic characterization. Uth International Congress of Carboniferous Stratigraphy and Geology, C.R., 4, 160-185. HELFRICH, C. T. & HOWER, J. C. 1989. Palynologic and petrographic variation in the Otter Creek coal beds (Stephanian, Upper Carboniferous), western Kentucky. Review of Palaeobotany and Palyn- ology, 60, 179-189. HESS, J. C. & LIPPOLT, H. J. 1986. ""Ar/'^Ar ages of tonstein and tuffanidines: New cahbration points for improvement of the upper Carboniferous time scale. Isotope Geoscience, 59, 143-154. HICKEY, L. J. 1984. Changes in the angiosperm flora across the Cretaceous/Tertiary boundary. In: BERGGREN, W. A. & VAN COUVERING, J. A. (eds) Catastrophies in Earth History. Princeton Uni- versity Press, 279-313. JOHNSON, K. R. 1992. Leaf-fossil evidence for extensive extinction at the Cretaceous-Tertiary boundary. North Dakota, USA. Cretaceous Research, 13, 91-117. & HICKEY, L. J. 1990. Megafloral change across the Cretaceous/Tertiary boundary in the northern Great Plains and Rocky Mountains, U.S.A. Geological Society of America, Special Paper, 247, A^'i-'Wi. JOHNSON, P. R. 1979. Petrology and environments of deposition of the Herrin (No. 6) Coal Member, Carbondale Formation, at Old Ben Coal Company Mine No. 24, Franklin County, Illinois. M.S. Thesis, University of Illinois, Urbana-Cham- paign. KLEIN, G. DE V. 1990. Pennsylvanian time scales and cycle periods. Geology, 18, 455-457. KosANKE, R. M. 1947. Plant microfossils in correlation of coal beds. Journal of Geology, 55, 280-284. 1988. Palynological studies of Middle Penn- sylvanian coal beds of the proposed Pennsylva- nian System Stratotype in West Virginia. United States Geological Survey Professional Paper, 1455, 1-73. cfe CECIL, C. B. 1989. Late Pennsylvanian climate changes and palynomorph extinctions. Abstracts, 28th International Geological Congress, 1, 214-215. LESNIKOWSKA, A. D. 1989. Anatomically Preserved Marattiales from Coal Swamps of the Desmoine- sian and Missourian of the Midcontinent United States: Systematics, Ecology, and Evolution. Ph.D. Thesis, University of lUinois, Urbana-Cham- paign. LORENZO, P. 1979. Les sporophylles de Lepidodendron dissitum Sauver, 1848. Geobios, 12, 137-143. MAHAFFY, J. F. 1985. Profile patterns of coal and peat palynology in the Herrin (No. 6) Coal Member, Carbondale Formation, Middle Pennsylvanian of southern Illinois. Proceedings 9th International Congress of Carboniferous Stratigraphy and Geology, 5, 155-159. 1988. Vegetational history of the Springfield coal (Middle Pennsylvanian of Illinois) and the distribution of the tree fern miospore, Thymo- spora pseudothiessenii, based on miospore profiles. International Journal of Coal Geology, 10, 239- 260. MILLER, W. III. 1990. Hierarchy, individuality and paleoecosystems. In: MILLER, W. Ill (ed.) Paleo- 220 W. A. DIMICHELE & T. L. PHILLIPS community Temporal Dynamics: the Long-term Development of Multispecies Assemblages. Paleon- tological Society Special Publication, 5, 31-47. 1993. Models of recurrent fossil assemblages. Lethaia, 26, 182-183. MicKLE, J. E. 1984. Taxonomy of Specimens of the Pennsylvanian-age Marattialean Fern Psaronius From Ohio and Illinois. Illinois State Museum, Scientific Paper, 19. O'NEILL, R. V. 1989. Perspectives in hierarchy and scale. In: ROUGHGARDEN, J. A., MAY, R. M. & LEVIN, S. A. (eds) Perspectives in Ecological Theory. Princeton, Princeton University Press, 140-156. DeANGELis, D. L., WAIDE, J. B. & ALLEN, T. F. H. 1986. A Hierarchical Concept of Ecosystems. Princeton University Press, Monographs in Po- pulation Biology, 23. OVERPECK, J. T., WEBB, R. S. & WEBB, T. III. 1992. Mapping eastern North American vegetation change of the past 18 ka: no analogues and the future. Geology, 20, 1071-1074. PARRISH, J. T. 1982. Upwelling and petroleum source beds, with reference to the Paleozoic. AAPG Bulletin, 66, 750-774. PATTON, B. C. & ODOM, E. P. 1981. The cybernetic nature of ecosystems. American Naturalist, 118, 886-895. PEPPERS, R. A. 1985. Comparison of miospore assemblages in the Pennsylvanian System of the Illinois Basin with those in the Upper Carbonifer- ous of Western Europe. 9th International Congress of Carboniferous Stratigraphy and Geology, C.R., 2, 483-502. ^ & PFEFFERKORN, H. W. 1970. A comparison of floras of the Colchester (No. 2) Coal and Francis Creek Shale. In: SMITH, W. H., NANCE, R. B., HOPKINS, M. E., JOHNSON, R. G. & SHABICA, C. W. (eds) Depositional Environments in Parts of the Carbondale Formation - Western and Northern Illinois. Illinois State Geological Survey Guide- book Series 8, 61-74. PFEFFERKORN, H. W. 1976. Pennsylvanian tree fern compressions Caulopteris, Megaphyton, and Arti- sophyton gen. nov. in lUinois. Illinois State Geological Survey Circular 492. , GILLESPIE, W. H., RESNICK, D. A. & SCHEIH- iNG, M. H. 1984. Reconstruction and architecture of medullosan pteridosperms (Pennsylvanian). The Mosasaur, 2, 1-8. & THOMSON, M. 1982. Changes in dominance patterns in Upper Carboniferous plant-fossil assemblages. Geology, 10, 641-644. PHILLIPS, T. L. 1979. Reproduction of heterosporous arborescent lycopods in the Mississippian-Penn- sylvanian of Euramerica. Review of Palaeobotany and Palynology, 11, 239-289. , AVCIN, M. J. & BERGGREN, D. 1976. Fossil peat of the Illinois Basin. Illinois State Geological Survey, Educational Series, 11. & DiMiCHELE, W. A. 1981. Paleoecology of Middle Pennsylvanian age coal swamps in south- em Illinois - Herrin Coal Member at Sahara Mine NO. 6. In: NIKLAS, K. J. (ed.) Paleobotany, Paleoecology, and Evolution. Praeger, New York 1, 231-284. & 1992. Comparative ecology and life- history biology of arborescent lycopsids in Late Carboniferous swamps of Euramerica. Annals of the Missouri Botanical Garden, 19, 560-588. , KuNZ, A. B & MiCKiSH, D. J. 1977. Paleobotany of permineralized peat (coal balls) from the Herrin (No. 6) Coal Member of the Illinois Basin. In: GIVEN, P. N. & COHEN, A. D. (eds) Interdisciplinary Studies of Peat and Coal Origins. Geological Society of America, Micro- form Publication 7, 18^9. & PEPPERS, R. A. 1984. Changing patterns of Pennsylvanian coal-swamp vegetation and imph- cations of climatic control on coal occurrence. InternationalJournal of Coal Geology, 3, 205-255. , PEPPERS, R. A., AVCIN, M. J. & LAUGHNAN, P. F. 1974. Fossil plants and coal: patterns of change in Pennsylvanian coal swamps of the Illinois Basin. Science, 184, 1367-1369. > ~ & DIMICHELE, W. A. 1985. Strati- graphic and interregional changes in Pennsylva- nian coal-swamp vegetation: environmental inferences. International Journal of Coal Geoloev 5, 43-109. PIERCE, B. S., STANTON, R. W. & EBLE, C. F. 1991. Facies development in the Lower Freeport coal bed, west-central Pennsylvania, U.S.A. Interna- tional Journal of Coal Geology, 18, 17-43. > & 1993. Comparison of the petrography, palynology, and paleobotany of the Stockton coal bed. West Virginia and impKcations for paleoenvironmental interpretations. Organic Geochemistry, 20, 149-166. PIGG, K. B. & ROTHWELL, G. W. 1983. Chaloneria gen. nov.: heterosporous lycophytes from the Pennsyl- vanian of North America. Botanical Gazette 144 132-147. PIMM, S. L. 1991. The Balance ofNaturel University of Chicago Press. PRYOR, J. S. 1993. Patterns of ecological succession within the Upper Pennsylvanian Duquesne coal of Ohio (USA). Evolutionary Trends in Plants, 1, 57- 66. Ross, C. A. & Ross, J. R. P. 1987. Late Paleozoic Sea Levels and Depositional Sequences. Cushman Foundation for Foraminiferal Research, Special Publication, 24, 137-149. & 1988. Late Paleozoic transgressive- regressive deposition. Society of Economic Paleon- tologists and Mineralogists Special Publication 42, 227-247. ROTHWELL, G. W. 1988. Upper Pennsylvanian Steu- benville coal-ball flora. Ohio Journal of Science 88, 61-65. SALTHE, S. N. 1985. Evolving Hierarchical Systems. Columbia University Press, New York. SCHABILION, J. T. & REIHMAN, M. A. 1985. Anatomy of petrified Neuropteris scheuchzeri pinnules from the Middle Pennsylvanian of Iowa: a paleoecolo- gical interpretation. Proceedings 9th International Congress of Carboniferous Stratigraphy and Geology, 5, 3-12. EXTINCTION AND RECOVERY IN PENNSYLVANIAN COAL SWAMPS 221 ScHLESiNGER, W. H. 1978. Community structure, dynamics and nutrient cycling in the Okefenokee cypress swamp forest. Ecological Monographs, 48, 43-65. SCHUBERT, J. K. & BOTTJER, D. J. 1993. Recovery from the end-Permian mass extinction event: paleo- ecology of Lower Triassic marine invertebrate assemblages in the Great Basin. Geological Society of America, Abstracts with Programs, A- 156. SCOTT, A. C. 1978. Sedimentologjcal and ecological control of Westphalian B plant assemblages from west Yorkshire. Proceedings of the Yorkshire Geological Society, 41, 461-508. & REX, G. 1985. The formation and signifi- cance of Carboniferous coal balls. Philosophical Transactions of the Royal Society of London, 311B, 123-137. STECHESGOLEV, A. K. 1975. Die entwicklung der Pflanzenbedeckung im Suden des Europaischen Teils der UdSSR, vom Ende des Mittlekarbons bis zum Perm. Umfang und Gliederung des oberen Karbons (Stefan). C.R. 7th International Congress Carboniferous Stratigraphy and Geology (Krefeld), 4, 275-280. TAYLOR, T. N. 1965. Paleozoic seed studies: a monograph of the genus Pachytesta. Palaeonto- graphica, Abt. B. 117, 1^6. TRJVETT, M. L. 1992. Grovrth architecture, structure, and relationships of Cordaixylon iowensis nov. comb. (Cordaitales). International Journal of Plant Sciences, 153, 272-287. & ROTHWELL, G. W. 1985. Morphology, systematics, and paleoecology of Paleozoic fossil plants: Mesoxylon priapi sp. nov. (Cordaitales). Systematic Botany, 10, 205-223. TSCHUDY, R. H., PiLLMORE, C. L., ORTH, C. J., GiLMORE, J. S. & KNIGHT, J. D. 1984. Disruption of the terrestrial plant ecosystem at the Cretac- eous-Tertiary boundary, western interior. Science, 225, 1030-1032. VERMEU, G. J. 1987. Evolution and Escalation. Prince- ton University Press, Princeton. WEBB, T. III. 1988. Eastern North America. In: HUNTLEY, B. & WEBB, T. Ill, (eds) Vegetation History. Kluwer, Dordrecht, 385^14. WnrrE, D. & THIESSEN, R. 1913. The origin of coal. United States Bureau of Mines Bulletin, 38. WILLARD, D. A. 1989. Source plants for Carboniferous microspores: Lycospora from permineralized Lepidostrobus. American Journal of Botany, 76, 820-827. 1993. Vegetational patterns in the Springfield coal (Middle Pennsylvanian, Illinois Basin): com- parison of miospore and coal-ball records. Geo- logical Society of America, Special Paper 286, 139-152. & PHILLIPS, T. L. 1993. Paleobotany and palynology of the Bristol Hill Coal Member (Bond Formation) and Friendsville Coal Member (Mattoon Formation) of the Illinois Basin (Upper Pennsylvanian). Palaios, 8, 574-586. WING, S. L. 1984. Relation of paleovegetation to geometry and cyclicity of some fluvial carbonac- eous deposits. Journal of Sedimentary Petrology, 54, 52-66. cfe TIFFNEY, B. H. 1987. The reciprocal interaction of angiosperm evolution and tetrapod herbivory. Review of Palaeobotany and Palynology 50, 179-210. WINSTON, R. B. 1986. Characteristic features and compaction of plant tissues traced from permi- nerahzed peat to coal in Pennsylvanian coals (Desmoinesian) from the Illinois Basin. Interna- tional Journal of Coal Geology, 6, 21-41. 1989. Identification of plant megafossils in Pennsylvanian-age coal. Review of Palaeobotany and Palynology, 57, 265-276. 1990. Implications of paleobotany of Pennsyl- vanian-age coal of the central Appalachian basin for climate and coal-bed development. Geological Society of America Bulletin, 102, 1720-1726. & PHILLIPS, T. L. 1991. A Structurally Preserved, Lower Pennsylvanian Flora from the New Castle Coal Bed of Alabama. Geological Survey of Alabama, Circular 157. WNUK, C. & PFEFFERKORN, H. W. 1984. Life habits and paleoecology of Middle Pennsylvanian me- dullosan pteridosperms based on an in situ assemblage from the Bemice Basin (Sullivan County, Pennsylvania, U.S.A.). Review of Palaeo- botany and Palynology, 41, 329-351. WOLFE, J. A. & UPCHURCH, G. R. JR. 1986. Vegetation and floral changes at the Cretaceous-Tertiary boundary. Nature, 324, 148-152. & 1987. Leaf assemblages across the Cretaceous-Tertiary boundary in the Raton Basin, New Mexico and Colorado. Proceedings of the National Academy of Sciences, USA, 84, 5096-5100. ZIEGLER, A. M. 1990. Phytogeographic patterns and continental configurations during the Permian Period. In: MCKERROW, W. S. & SCOTESE, C. R. (eds) Palaeozoic Palaeogeography and Biogeogra- phy. Geological Society, London, Memoir, 12, 363-379. , RAYMOND, A., GIERLOWSKI, T. C, HORRELL, M. A., ROWLEY, D. B. & LOTTES, A. L. 1987. Coal, climate, and terrestrial productivity - the present and Early Cretaceous compared. In: SCOTT, A. C. (ed.) Coal and Coal-Bearing Strata - Recent Advances. Geological Society, London, Special Publication, 32, 25-49.