Which has a better fossil record, clams or leaves? The question seems too simple. Surely, clams with their calcareous shells have a much better chance for preservation, and thus a better fossil record, than the rapidly decaying leaves that require much faster burial rates for preservation. Yet, the question is not as simple as it may first appear because robust hard parts are much more prone to post-mortem transport and temporal mixing (e.g., Kidwell & Behrensmeyer 1993; Kowalewski 1996a). Consequently, the fossil record of clams tends to be much more smeared temporally and spatially (e.g., Flessa et al. 1993, 1995) than the fossil record of leaves (e.g., Burnham 1993; Wing & DiMichelle 1995).
I assemble here some intuitively obvious rules of taphonomy into one conceptual framework. I will first consider the effects of an organism's skeletal robustness on three major factors that undermine the quality of the fossil record (1) incompleteness, (2) temporal mixing, and (3) spatial mixing.
1. Incompleteness. The obvious advantage of having a robust skeleton is an increased chance for preservation: the greater the robustness, the more complete the fossil record is likely to be. The fossil record of clams is bound to be much more complete than that of leaves. Consequently, robust fossils will more reliably record such large-scale features as stratigraphic range, biogeographic range, and taxonomic or morphological diversity. They will be less susceptible to such phenomena as, for example, the Signor--Lipps effect (Signor & Lipps 1982).
2. Temporal mixing. Increased skeletal robustness is likely to result in increased temporal mixing: robust hard parts can survive multiple reworking events and a long residence time near depositional surfaces, and consequently, may undergo temporal mixing on a scale of hundreds to thousands of years (e.g., Flessa et al. 1993; Kowalewski 1996a). In contrast, fragile remains such as leaves do not last very long on active depositional surfaces, and consequently, undergo temporal mixing over much shorter time scales (e.g., Burnham 1993; Wing & DiMichelle 1995). Thus, fragile fossils are more likely to provide records with a temporal resolution of months or single years.
3. Spatial mixing. An increase in the robustness of an organism may also enhance its susceptibility to post-mortem transport, and thus, the more robust the skeleton of an organism, the poorer its spatial resolution is likely to be. For example, the thick-shelled mollusks can be preserved tens of kilometers from their original habitat (Flessa et al. 1995), fragile, thin-shelled lingulide brachiopods are unlikely to be found more than a few kilometers from the place of their death (Kowalewski 1996b), and soft bodied organisms (or trace fossils) are typically preserved in situ (e.g., Pemberton & Frey 1984). Obviously, the robustness is not the only factor determining the spatial resolution. For example, although much more robust, a large, heavy coral head is much less likely to be transported than an oyster shell. Similarly, despite their fragility, leaves may occasionally be transported over large distances (e.g., Ferguson 1985). Nevertheless, on average, the robust remains are more likely to survive an extensive post-mortem transport. Indeed, although leaves can be transported over large distances, their fossil record is resolved spatially much better than that of mollusks (e.g., Wing & DiMichelle 1995; Wing, personal communication, 1996). Note that robust organisms may appear eurytopic in the fossil record not only because of "spatial smearing" but also because their remains may survive near the sediment surface during a time span long enough for substantial environmental changes to occur. For example, relict shells of the highly stenotopic "shoreline marker" Crassostrea virginica can be found at various depths on the surface of the modern continental shelf (MacIntyre et al. 1978). In contrast, the ecological ranges of fragile organisms are more likely to be recorded accurately because they are less likely to be transported substantially or survive for a long time near the sediment surface.
At this stage, the Reciprocal Taphonomic Model is rather primitive. First, the inherent fossilization potential of major groups of organisms is not quantified along a single taphonomic scale -- although most major groups could be placed along an ordinal scale, as is done here for some major groups (Fig. 1). Second, there are many second-order variables that can play significant roles. These include intrinsic characteristics of organisms such as geographic range, population density, and mode of life (e.g., Palmqvist 1991; Boyajian & Thayer 1995), and extrinsic factors such as rate of deposition, intensity of taphonomic agents, and rate of reworking and bioturbation (e.g., Kidwell & Bosence 1991; Flessa et al. 1993).
Taphonomic effects of the second-order factors vary, again in a reciprocal fashion, as a function of skeletal robustness (Table 1). For example, reworking/bioturbation (Table 1) may substantially increase the spatial and temporal mixing of the fossil record of a clam by exhuming and transporting its shells. However, such reworking will not be likely to generate the spatial or temporal mixing in the fossil record of a jellyfish which would not survive any transport or reworking (except in the rare cases of reworked impressions). In contrast, reworking/bioturbation may dramatically decrease the completeness of the jellyfish fossil record (by obliterating any impressions on the sediment surface). Reworking/bioturbation will have relatively little impact on the completeness of the record of the robust clam which will typically survive substantial post-mortem transport. This example illustrates what I believe to be a general principle: the greater the skeletal robustness, the easier it is for second-order factors to generate the temporal and spatial mixing, but the harder it is for them to reduce the completeness. The full development of the Reciprocal Taphonomic Model requires a precise understanding of how the effects of second-order factors vary depending on an organism's robustness. Ultimately, the model should allow us to predict the character and quality of a fossil record for any group.
Taphonomy has come a long way since the German actuopaleontological school of Weigelt and Wasmund developed the first rigorous research program (for review see Cadee 1991), and since Efremov (1940) named the discipline. Yet, there have been only few theoretical papers that dealt with the conceptual framework and methodology of the discipline (Efremov 1940; Olson 1980; Behrensmeyer & Kidwell 1985; Brett & Baird 1986; Wilson 1988; Ferna‡ndze LL—pez 1991). Not surprisingly, thus, taphonomy lacks a well-defined disciplinary matrix (sensu Kuhn 1977) or explicit, universally applicable models. Efremov (1940) postulated some taphonomic "laws", but mostly relevant to terrestrial vertebrates. Wilson (1988), based on a review of Recent studies, listed nine "generalizations about burial"; but, as Efremov, he mostly focused on the methodology and goals of the discipline. Thus, we are still lacking rigorous predictive models and, despite the achievements of the last few decades, we have yet to develop a research program which could lead to a "Unified Taphonomic Theory". The Reciprocal Taphonomic Model (Fig. 1, Table 1) is suggested here as a first tentative but explicit step in this direction.
Acknowledgments. -- I thank Karl W. Flessa, Peter Holterhoff, John Alroy, and Andy Cohen for many helpful comments and Gerhard Cadee andd Keith Meldahl for careful reviews which greatly improved this paper. This work was started during my post-doctoral fellowship at the University of Arizona and completed during the Humboldt Fellowship in TuŸbingen, Germany. Supported by NSF Grant EAR-9405311 and Alexander von Humboldt Foundation. This is publication no. 21 of C.E.A.M. (Centro de Estudios de Almejas Muertas).
C.E.A.M. abstracts of talks