I shall define biological community as the species comprising a temporally and spatially localized assemblage. Although members of a community may interact, I do not wish to imply that the community behaves as a functional unit or that interactions among members determine its structure. When discussing community structure in this paper, I refer exclusively to the diversity of species because of the general availability of comparative data and because most of community theory makes predictions concerning number of coexisting species. Other important community attributes, such as relative abundance patterns, trophic level structure, food web organization, resistance to and recovery from perturbation, and physical architecture, will not be considered.

The way we view species distribution patterns, particularly at the level commonly referred to as the “community”, has changed over the past 70 years in biology and, subsequently, in paleontology. Because the degree to which species associations can be interpreted as ecological and evolutionary units depends ultimately on recognition and interpretation of faunal spatial variability, we need to understand the nature of this variability at all levels of resolution before we can adequately address questions of “community” structure and dynamics. While it is possible to recognize spatial variability at several levels, from the distributions of individuals within a species to the overall pattern created by the global biota, we must ask whether these patterns really comprise a hierarchy with natural discontinuities (Fig. 1), or whether it is more realistic to view them as a continuous variability spectrum.

Techniques and field observations that detect “spatial variation” and “temporal dynamics” in fossil deposits have become important research programs in paleosynecology. These studies attempt to delineate aggregates and sequences of fossils at varied scales that appear to result from processes encompassing larger areas and greater time spans than the processes familiar to neoecologists. Description and modeling of patterns and processes at these scales would be significant contributions to historical biology, but little attention has been given to the ontology of “natural” multispecies units discernable in fossil data sets at varied spatio-temporal scales of resolution. Do patterns at any of these nested levels of variation – patches within shell beds, shell beds within biofacies, and so on – represent the elusive original community of organisms?

Since the middle of the last century, with the publication of d'Orbigny's monumental synthesis (1849–52, 1850–52), it has been clear that the fossil record is divisible into a number of significant biostratigraphic units (his ”terrains” and ”etages”). It has also been clear that within each of these d'Orbigny-type units there are included a number of individual “biofacies,” the term used by the geologist, or “communities,” the term used by paleobiologists. I have referred (1983) to the major d'Orbigny biostratigraphic divisions as “ecologicevolutionary units,” because the term terrane is geologically preoccupied today, and the subdivision terms ”etage” and “stage” have different meanings.

At the end of the Ordovician many marine benthonic and planktonic faunas underwent a global extinction that has been attributed to climatic changes and glacio-eustatic fluctuations in sea-level (Berry and Boucot, 1973; Sheehan, 1973, 1975; Brenchley, 1984). Raup and Sepkoski (1982) found that the Late Ordovician-Early Silurian extinction event was one of the five major Phanerozoic mass extinctions.

Over the past several years, a variety of macroevolutionary studies have focused on global diversification patterns exhibited by the earth's biota as a whole, as well as among constituent groups. One motivation for this increased attention is the recognition that analyses of temporal changes in global diversity can provide substantial insight into underlying macroevolutionary processes (e.g. Sepkoski, 1978, 1979, 1981, 1984; Sepkoski et al., 1981; Gould and Calloway, 1980; Carr and Kitchell, 1980; Kitchell and Carr, 1985; Miller and Sepkoski, 1988). Indeed, much about the dynamics of macroevolution has been elucidated through such investigations, but major diversity transitions in the history of life cannot be fully understood without consideration of the local, environmental/ecological contexts in which they took place. In other words, in the study of macroevolution, it is important to pay as much attention to the space dimension as has historically been paid to the time dimension. The utility of a spatio-temporal approach has been demonstrated in a series of studies conducted by Sepkoski and Sheehan (1983), Sepkoski and Miller (1985), Jablonski and Bottjer (1983), Bottjer and Jablonski (in press), Droser and Bottjer (1988), Bottjer et al. (1988), and Miller (1988, in press). Collectively, these investigations have suggested that major changes in the global diversities of several groups were accompanied by measurable paleoenvironmental shifts.

The term dysaerobic was first coined by Rhoads and Morse (1971) to define conditions in which oxygen levels ranged between 1.0 and 0.1 mL of oxygen/litre of water. Such low oxygen tensions were thought to exclude all but soft-bodied invertebrates with the absence of shelly fossils being one of the diagnostic features of the dysaerobic zone. Subsequent observations of recent dysaerobic environments has revealed that shelly invertebrates can survive in oxygen values as low as 0.3 mL/L (e.g. Savrda, Bottjer and Gorsline, 1984). Consequently the dysaerobic zone is now used in the more qualitative sense to denote conditions in which oxygen levels are sufficiently low to cause a notable change in the fauna (Kammer and others, 1986; Wignall and Myers, 1988). Faunal abundance and diversity is reduced under low oxygen conditions compared to more normally oxygenated (aerobic) biofacies.

Historically, the greatest freshwater bivalve species diversity in the world was in the Tennessee, Cumberland and Alabama River systems. The Tennessee River System had 94 and the Cumberland River System had 85 taxa reported (Starnes and Bogan, 1988). Ortmann (1918, 1924, 1925, 1926), Wilson and Clark (1914), Neel and Allen (1964), van der Schalie (1939, 1973) and van der Schalie and van der Schalie (1950) documented this diverse fauna. However, even in the early days of this century these authors noted the decline in the mussel populations and the loss of species from certain rivers (e.g. Ortmann, 1909a, 1918).

Ancient and modern marine carbonate hardgrounds offer unusual opportunities to study the evolution of communities from the Early Cambrian into the Holocene. Throughout this time the general physical conditions of a hardground community have been similar. The substrate is hard so sessile organisms must either attach to its surface, nestle in cavities, or bore into it for occupation. These organisms are thus preserved in situ. Since space is often the limiting physical resource, organisms must have ways of obtaining and defending it, and these competitive hierarchies are often preserved in the spatial relationships of the species. Scouring and/or burial in sediment usually marks the end of the brief habitation.

Marine metazoans have lived infaunally throughout the Phanerozoic. The biotic history of the infaunal habitat is, however, poorly understood. Only the two extremes are relatively well known; the Precambrian had a limited infaunal biological benthic boundary layer (the depth below the sediment-water interface that is occupied or otherwise influenced by benthic organisms [Rhoads and Boyer, 1982]), whereas modern environments have a well-developed infaunal biological benthic boundary layer with extensive and deep bioturbation.

In the fossil record of the Paleozoic, epifauna are typically shelly, whereas infauna are typically soft-bodied; the two groups have tended to be studied in isolation from one another, and commonly by different specialists. We believe that the understanding of patterns of distribution and evolution of epifauna cannot proceed in the absence of data on infauna. Over the past decade, several workers have attempted to combine information about the two groups into models of community evolution.

Among the most intriguing and significant aspects of the marine stratigraphic record are patterns of temporal change in fossil assemblages and paleocommunities. Understanding the stratigraphic patterns and the correct temporal scale of such faunal change is crucial to interpreting the underlying processes involved. Inattention to the temporal scale at which paleontological data are collected, and at which faunal change is observed, often results in the use of entirely inappropriate explanatory models. In many cases modern ecological theories have been misapplied to the fossil record because problems of scale were not adequately considered. The term “community” itself has been applied to such a wide range of fossil accumulations that it has ceased to have any consistent paleoecologic meaning (see discussion in Järvinen et al.,1986). For this reason, we prefer to use the term “assemblage” for time-averaged accumulations of fossils, and restrict “community” to only those organisms which actually lived together in the same space and time (i.e., a biocoenosis). Therefore, faunal assemblages, even when untransported, are the preserved amalgamated record of many successive communities within which short-term (10 – 100 years) changes may or may not be resolvable. Recurrent, compositionally similar assemblages, believed to have occupied generally similar benthic environments, are then grouped into biofacies which can be seen to intergrade and migrate through time.

Extensive species-level, marine biotic turnover is recorded across the Cenomanian-Turonian (C-T) Stage boundary in the western interior of the United States (Fig. 1) and intercontinentally (e.g., Wright and Kennedy, 1981; Elder, 1987; Jarvis et al., 1988). Widespread intensification and expansion of the oxygen minimum zone have been widely proposed as a causal mechanism for the organic-rich deposits and biotic extinctions found in strata of this age (e.g., Schlanger et al., 1987; Jarvis et al., 1988). Accompanying these extinctions were large-scale reorganizations of community compositions and trophic structures. In the western interior, repeated community replacement was associated with widespread cyclic sedimentation, and long-term community turnover was related to extended periods of dysaerobia and possibly low salinity conditions. Community compositions and distributions were therefore largely controlled by physical rather than biological factors. This paper will document how these oceanographic and sedimentologic factors affected community patterns across the C-T boundary.

The study of paleoecologic processes through geologic time requires careful analysis of the stratigraphic distribution of fossils. Collections from stratigraphic units of varying thickness provide data needed to describe the spatial and temporal patterns of fossil species, populations, and paleocommunities. Obviously, collections that have broad stratigraphic spacing will reveal only long-term trends. Narrowing the stratigraphic distance between collections may reveal some aspects of shorter term biological processes, but it is well established (Gould, 1980; Schindel, 1980; McCall and Tevesz, 1983) that the lower limit of paleontologic resolution is reached when biological processes operate more quickly than the rate of formation of final undisturbed bedding units.

Paleontologists have lavished much time and energy on description and explanation of large-scale patterns in the fossil record (e.g., mass extinctions, histories of monophyletic taxa, deployment of major biogeographic units), while paying comparatively little attention to biologic patterns preserved only in local stratigraphic sequences. Interpretation of the large-scale patterns will always be seen as the chief justification for the science of paleontology, but solving problems framed by long time spans and large areas is rife with tenuous inference and patterns are prone to varied interpretation by different investigators using virtually the same data sets (as in the controversy over ultimate cause of the terminal Cretaceous extinctions). In other words, the large-scale patterns in the history of life are the true philosophical property of paleontology, but there will always be serious problems in attempting to resolve processes that transpired over millions to hundreds-of-millions of years and encompassed vast areas of seafloor or landscape. By contrast, less spectacular and more commonplace changes in local habitats (often related to larger-scale events and cycles) and attendant biologic responses are closer to our direct experience of the living world and should be easier to interpret unequivocally. These small-scale responses are reflected in the fossil record at the scale of local outcrops.

Stratigraphy provides a basic temporal and spatial framework for all studies of earth history. The foundations of stratigraphy were built in the 17th and 18th centuries through the efforts of Steno and Smith. Formalization of the principles of organic evolution followed closely in the mid-19th century with the publication of the Origin of Species. Although paleoecological observations span centuries, the establishment of paleoecology as a scientific discipline is a 20th century phenomenon. Initially, paleontology (paleobiology) was an integral part of stratigraphy and housed the subdisciplinary studies of evolution and paleoecology. Retention of a stratigraphic framework soon became difficult, perhaps due to the potential for circular reasoning in the evolutionary study of fossils. If the rock record was subdivided on the basis of fossils, how could the evolutionary relationships of these organisms be evaluated when they, or related forms, had been used to construct the temporal framework?

Given sweeping evolutionary changes in the diversity and environmental deployment of species that produce shells, destroy shells, and otherwise interact with shells, it would be surprising if patterns of shell accumulation did not change over Phanerozoic time. Documentation of such patterns might indicate how and to what degree post-mortem bias has changed, and thus the taphonomic comparability of data sets used in paleobiologic modelling. At the same time, long-term patterns in shell accumulation can be products of evolutionary ecology, and thus represent an underexploited source of paleobiological insight themselves.

The utilization of taphonomic information to formulate biostratinomic models for modern and ancient assemblages has become a potentially powerful tool in paleoecologic analysis. The division of fossil assemblages into discrete suites of taphonomically-similar material adds an extra dimension to the interpretation of depositional setting and paleoecologic structure (Brett and Baird, 1986; Speyer and Brett, 1986, 1988; Speyer, 1987). This approach uses the hypothesis that taphonomic alteration varies in a predictable way with depositional setting. In other words, each specific environment (e.g., low-salinity muddy bay, storm-dominated clastic shelf) is characterized by a unique suite of physical, chemical and biological processes: these processes imprint a unique and predictable “taphonomic signature” on the death assemblage.