Since their inception in 1978, the annual short courses sponsored by the Paleontological Society have aimed to broaden and to enhance the professional education of paleontologists, including students new to the field. The 1993 short course continues in that tradition, but differs from many previous courses in focussing not on a taxonomic group but on a broader aspect of the fossil record, namely the time resolution of fossil assemblages. This seemed an especially good topic for a short course because questions of absolute and relative time – how old? how fast? how synchronously? – pervade paleontology and historical geology in general.

Paleoecological and paleoenvironmental studies have a rich tradition of actualistic work on Recent biotas and environments. Ever since Lyell (and before) geologists have examined present–day sedimentary environments and faunas in an effort to interpret the environments and life of the past. After all, if the present is not the key to the past, it is at least a key to the past. However, there is no similarly long tradition of studying the Recent in an effort to gain insight into issues of the temporal resolution of fossil assemblages.

For more than half a century, microfossils–especially foraminifera–have been widely used as stratigraphic markers and paleoenvironmental indicators. Although increasing emphasis has been placed on their use in high-resolution paleoclimate studies, the time-scales involved in most microfossil-based stratigraphic investigations have remained relatively coarse (hundreds-of-thousands to millions of years). My intent herein is to try to come to grips with the interplay between time-averaging of benthic foraminiferal assemblages and stratigraphic resolution, and the implications for recognition of short-term physical and biological processes. These sorts of considerations deserve much closer scrutiny as the applied Earth sciences continue to move from a base of resource exploration and exploitation to one of paleoclimate modelling and ecosystem management (Martin, 1991; Corliss, 1993). The potential stratigraphic and paleoenvironmental resolution of foraminiferal assemblages is assessed using concepts derived from the age analysis of deep-sea assemblages.

The evidence available for determining time resolution in plant fossil deposits can be subdivided easily into two general categories: the information provided by the biological properties of the plants and the information provided by the physical properties of the sediments and accumulation styles. Accordingly, this contribution is divided into two sections, covering these two aspects of time resolution from modern plant accumulations. From a standpoint of modern macrofloral accumulation, there are four ideas that have guided this chapter.

Pollen analysis is an exercise in seeing. The ultimate goal is to see into the past, to send down a periscope and view what went on. The metaphor of the periscope is too limited, however; a video recorder from high in space with resolution in places up to 10 m is more encompassing of what is possible. The images that are retrieved can be of high or low resolution temporally, spatially, taxonomically, and numerically, and they can illustrate local to global changes in plant populations, vegetation, climate, human activity, fire frequency, and plant diseases over decades to millennia. Because each of these entities or phenomena varies spatially and temporally, records of data covering a breadth of scales in space and time are needed. To obtain the highest quality images about a specific phenomenon requires an understanding of the sensing system that accumulated the data. How does the periscope or video recorder work and what are the scaling characteristics of the images that it registers? These characteristics include breadth of coverage, sampling resolution, and sampling density in time, space, and taxonomy. Actualistic and taphonomic studies of Quaternary data covering a variety of temporal and spatial scales have helped provide this understanding, and temporal resolution is just one concern in these studies.

The terrestrial vertebrate fossil record provides a window into the past evolution of taxa, communities, and ecosystems. It can also be used to reconstruct ancient environments, climates, and landscapes. Vertebrate fossils can document the evolution of humans and their interactions with fauna and environments. However, before the maximum potential of this record can be realized, it is essential to know the extent of temporal resolution, or time-averaging, represented by fossil samples.

Plants and shelly marine invertebrates often have diverse and densely packed communities that generate dense accumulations of organic remains. In contrast, populations in vertebrate communities are relatively dispersed in life, and normal annual mortality results in a relatively low density of remains scattered over the land surface. This attritional “bone rain” differs from unusual circumstances such as mass mortality and predator bone-collecting, which can generate high density accumulations of vertebrate remains in a short period of time.

Taphonomic studies of marine fossil assemblages have achieved new levels of sophistication in the past several years, but field-based investigations of time-averaged skeletal accumulations continue to suffer from an unavoidable problem: to date, a time machine has yet to be invented that would permit the researcher to observe directly the formation of the assemblage under study. Even when working with Recent subfossil accumulations on the sea floor, where it is theoretically possible to view the activity of (possibly myriad) taphonomic processes, practical considerations prevent the extensive, day-to-day monitoring that would be required to “bear witness”, in a kinetic sense, to the formation of skeletal assemblages. At best, we are sometimes provided with opportunities for comparative glimpses of skeletal accumulations before and after events of potential taphonomic significance, such as hurricanes (e.g., Miller et al., 1992).

In December 1969 officials from the U.S. Selective Service System conducted a draft lottery to establish the order in which nineteen-year-old men were to be called for military service. Three hundred sixty-six capsules, one for each possible birthday, were placed in a large wooden box. As the capsules for each month were added to the box, the contents of the box were mixed. Once all 366 capsules were in the box, it was shaken several times and emptied into a deep bowl. Capsules were then drawn from the bowl to determine the draft number corresponding to each date. Observers were satisfied that the capsules had been thoroughly mixed, but, as it turned out, the results were anything but random. The Spearman rank correlation between birth date and draft number was significant at the.001 level – men with December birthdays had a significantly higher probability of being called than did those with January birthdays (Fienberg, 1971).

Holocene sediments are the traditional classroom for learning to recognize the geologic record of different environments. Box cores and shallow trenches display sedimentary stratification and assemblages of organic remains together with their relationship to active surface processes and living organisms. Sedimentologists see the origin of particular bed-forms; paleoecologists learn to match animals to trace fossils, for example; and stratigraphers are reminded that it is not just the sediment surface that is active. Shells, bones, and sediment grains below the surface are still subject to rearrangement. Until this activity has ceased we have not seen what will ultimately be preserved as the mature stratigraphic record of the environment that we have trenched.

In this paper, I discuss issues of time resolution and time-averaging in fossil megafloras from Late Cretaceous (Cenomanian) to Oligocene age. The sites are predominantly from North America with some examples from other continents. The purpose of this paper is to explicate the methods used to resolve the age and duration of fossil floras and to discuss the uncertainties associated with these methods. The age of floras is most important for their evolutionary, paleoclimatic and paleoaltitudinal applications, while the duration of floras (here defined as the time represented by individual assemblages, florules, or quarries) is critical for paleoecological and paleoenvironmental applications. Levels of time resolution of both the age and the duration of fossil floras are subject to time-averaging. Behrensmeyer et al. (1992, p. 75) have defined two types of time-averaging: analytical time-averaging which is artificial and results from techniques of analysis and taphonomic time-averaging which is natural and results from the taphonomic history of the assemblage.

What is the range of time-averaging in the terrestrial vertebrate record? How are time-averaged assemblages distributed within the terrestrial vertebrate record? What are the implications of time-averaging for vertebrate paleobiological analysis?

One of the most important and challenging aspects of stratigraphy is the interpretation of the temporal scope of sedimentary units (Schindel, 1980, 1982; Sadler, 1981; Brandt Velbel, 1984). The problem arises at the scale of individual beds and of stratigraphic intervals up to many meters thick. Does a particular bed or interval-represent hours, days, years, centuries, or millennia? Resolution of this question is critical for determination of rates of sedimentation and biotic processes, and in assessing the reliability of the sample for paleoecological or evolutionary analysis. In the absence of a reliable framework of absolute radiometric dates the question can only be answered by indirect inference. Biostratigraphic zonation is a critical first step. But zonation is typically too coarse to resolve temporal scales less than 106 years and many zones are not firmly anchored to absolute dates. It is also important to keep separate the issue of temporal duration represented by fossils (as bioclasts) within a given stratum and that of the deposition of the sedimentary unit itself. There are many instances of thin graded beds full of fossils, which would be characterized unambiguously by sedimentologists as deposits of a single event of sedimentation, but in which the fossils may differ in age by thousands or even millions of years. Examples include many condensed, lag deposits of bones and conodonts (Baird and Brett, 1991), and condensed ammonoid beds containing fossils of several ammonite zones (Fürsich, 1971). Sedimentologic criteria provide one avenue of approach to this issue but commonly fall short of unambiguous answers.

Virtually all paleontologic and historical geologic interpretations require information on the time resolution of individual samples. For relatively broad segments of the record such as facies tracts and entire basins, a variety of approaches can be used to determine the relative and absolute duration of a “sample”. For finer subdivisions, however, such as individual beds and assemblages that are within the error-bars of radiometric dates or within the span of a biostratigraphic zone, estimating elapsed time per unit becomes far more difficult. Assessing time at this scale is important, however, because this is the usual sampling interval for autecologic, synecologic, morphometric, and species-level biostratigraphic and evolutionary studies.

Figure 1 summarizes estimates for the time-scales over which fossil assemblages accumulate, based on the findings summarized by the various authors of this volume, and organized according to major taxonomic groups in terrestrial and marine environments. For each group, actualistic estimates are based primarily on field evidence for the durability of organic remains, either from time-lapse tracking of carcasses or from direct-dating of individuals in Recent death assemblages; other methods include extrapolating from known rates of sediment accumulation and rates and depths of mixing. Estimates from the fossil record are based almost exclusively on indirect evidence, such as bracketing of assemblages by biostratigraphic and other chronostratigraphic datums, and actualistic rates of formation for the kinds of beds and facies that contain assemblages. The reader should see individual chapters for details and original sources of data.