Abstract.-Distinguishing the differential roles of hardpart-input rates and burial rates in the formation of shell beds is important in paleobiologic and sedimentologic studies, because high shelliness can reflect either high population density of shell producers or lack of sediment. The modeling in this paper shows that differences in the relative importance of burial rates and hardpart-input rates lead to distinct patterns with respect to the degree of shelliness and taphonomic alteration in shell beds. Our approach substantially complements other models because it allows computation of both shelliness and assemblage-level alteration. To estimate shelliness, we dissected hardpart-input rates into dead-shell production and shell destruction rates. To estimate assemblage-level alteration, we computed an alteration rate that describes how rapidly shells accrue postmortem damage. Under decreasing burial rates but constant hardpart-input rates, a positive correlation between alteration and shelliness is expected (Kidwell's R-sediment model). In contrast, under decreased destruction rates and/or increased dead-shell production rates and constant burial rates (Kidwell's R-hardpart model), a negative correlation between shelliness and alteration is expected. The contrasting predictions thus provide a theoretical basis for distinguishing whether high shell density in shell beds reflects passive shell accumulation due to a lack of sediment dilution or whether it instead reflects high shell input from a life assemblage. This approach should be applicable for any fossil assemblages that vary in shell density and assemblage-level alteration. An example from the Lower Jurassic of Morocco, which has shell-rich samples less altered than shell-poor samples, suggests that the higher shelliness correlates with higher community-level abundance and lower proportion of juveniles of the main shell producer, supporting the driving role of hardpart-input rates in the origin of the shell-rich samples in this case. This is of significance in paleoecologic analyses because variations in shelliness can directly reflect fluctuations in population density of shell producers.

Recognizing the differential role of sedimentation rates and hardpart-input rates (i.e., dead-shell production and shell destruction rates) in shell bed formation is important because high shell density in death assemblages can result from lack of sediment or high input of shells from a life assemblage. One of the main taphonomic paradigms in interpreting marine shell beds is that sites of slow net rate of sedimentation should be more favorable for formation of denser shell concentrations than sites of higher net rate of sedimentation (the low-dilution maxim of Kid well 1991). Kid well (1985, 1986a) built a theoretical framework for shell bed genesis and hypothesized that a model of shell bed formation could be cast mainly in terms of changes in sedimentation rate (this is known as the R-sediment model). As an alternative, Kidwell (1985, 1986a) proposed the R-hardpart model, which predicts that variations in dead-shell production and shell destruction rates primarily control the formation and preservation of shell beds. Thanks to the pioneering insights of these initial models, it is now clear that any satisfactory explanation of shell beds, and of taphonomic patterns in general, has to be rooted in sedimentation rate and hardpart-input rate. The R-sediment model has great power and robustness and is preferred because of its predictivity in terms of postmortem bias and biotic interactions (Kidwell 1986a). As many shell beds are indeed preferentially associated with omission or erosional surfaces (Kidwell and Jablonski 1983; Kidwell 1989), the R-sediment model has been supported and successfully used in sequence stratigraphic and environmental analyses (Beckvar and Kidwell 1988; Kidwell 1993; Abbott 1997, 1998; Naish and Kamp 1997; Kondo et al. 1998; Fürsich and Pandey 2003; Yesares-García and Aguirre 2004; Cantalamessa et al. 2005; Parras and Casadío 2005). Sequence stratigraphic simulations also show that uniform stratigraphic distribution of fossils can be changed to nonrandom and clustered distribution because sequence architecture is strongly controlled by sedimentation rates (Holland 1995, 2000).

The R-sediment model (Kidwell 1985, 1986a) predicts that there will be positive correlation between shelliness and taphonomic alteration because shells are exposed longer when sediment dilution is low. Also, it predicts that with a decrease in sedimentation rate, an increase in shelliness will be associated with an increase in time-averaging (Fürsich and Aberhan 1990; Kowalewski et al. 1998), morphologic variation, and a change in community composition. However, the predictions of the R-hardpart model have not previously been explored fully.

Further analyses of the interplay between hardpart-input rate and sedimentation rate are necessary for better understanding of the dynamics of shell bed formation. First, it is of primary interest to know to what degree high shell density corresponds to original live abundance or whether it reflects only the effect of passive accumulation. In turn, it can be important to distinguish whether rarity of shells in shell-poor deposits is due to low hardpart-input rate or high background sedimentation rate. If the role of hardpart-input rates and sedimentation rates in formation of shell beds can be differentiated, these questions can be answered. Although community-level abundance in the fossil record is mostly assessed in terms of relative numerical abundance, the recognition of fossil macroinvertebrate populations with originally high density is also of ecologic importance because dense populations of shelly organisms play an important role as ecosystem engineers in aquatic habitats (Gutiérrez et al. 2003).