Influence of Size-Sorting on Diversity Estimates from Tempestitic Shell Beds in the Middle Miocene of Austria

142 RESEARCH REPORT

Inﬂuence of Size-sorting on Diversity Estimates from Tempestitic Shell Beds in the Middle Miocene of Austria

MARTIN ZUSCHIN Institut fu¨ r Pala¨ontologie, Universita¨ t Wien, Althanstrasse 14, A-1090 Vienna, Austria, Email: [email protected]

MATHIAS HARZHAUSER Naturhistorisches Museum Wien, Burgring 7, A-1014 Vienna, Austria

OLEG MANDIC Institut fu¨ r Pala¨ontologie, Universita¨ t Wien, Althanstrasse 14, A-1090 Vienna, Austria

PALAIOS, 2005, V. 20, p. 142±158 DOI 10.2210/palo.2003.p03-87 INTRODUCTION

Taphonomic factors set limits to the resolution of diver- Paleontological data frequently are extracted from geneti- sity patterns in the fossil record. The most important bias cally and stratigraphically complex shell beds. It is there- is the typical loss of taxa without a mineralized skeleton fore important to recognize taphonomic biases that can lead (for review, see Kidwell and Flessa, 1995). Usually, the to major errors in paleoecological interpretations (e.g., on shelly remains of the original community then are affected ancient local biodiversity). The strong influence of trans- by disintegrative processes such as disarticulation, frag- port-related shell-size sorting on diversity estimates from mentation, abrasion, and dissolution (for reviews, see single samples was studied in a transect of the middle Mio- Powell et al., 1989; Kidwell and Bosence, 1991), in addi- cene Grund Formation (Lower Austria), which contains al- tion to time-averaging (e.g., Fu¨ rsich and Aberhan, 1990; lochthonous, psammitic event beds with channel structures, Olszewski, 1999; Kidwell, 2002). Among disintegrative sharp erosional bases, and graded bedding. These event processes, for example, selective dissolution of aragonite beds are interpreted as proximal tempestites, and contain can alter molluscan diversities strongly (e.g., Brachert densely packed, polytaxic molluscan assemblages. The fau- and Dullo, 2000; Wright et al., 2003), and the degree of time-averaging will influence the species richness of a par- nal composition and taphonomic features of shells indicate ticular fossil assemblage greatly (e.g., Staff and Powell, that transport occurred from wave- or current-agitated 1988; Kidwell, 2002). Another very important taphonomic nearshore habitats into a pelitic, inner-shelf environment. process that potentially influences diversity estimates is The different skeletal concentrations contain a highly di- shell transport, which inevitably will influence the size verse molluscan fauna with 130 species identified from frequency distribution of a paleocommunity (e.g., Menard more than 4200 individuals. Although the quantitatively and Boucot, 1951; Boucot, 1953; Olson, 1957; Veevers, most-important species are the same in standardized sam- 1959; Johnson, 1960; Fagerstrom, 1964; Shimoyama, ples from five different shell beds, species richness differs 1985). Although out-of-habitat transport is limited to few significantly among the three samples from the base of the settings and to particular sets of organisms only (for re- transect and the two samples from its top. Diversity de- views, see Kidwell and Bosence, 1991; Behrensmeyer et pends on size-sorting and therefore reflects the transport al., 2000), transport and size-sorting frequently are re- history of the individual tempestites, not the species rich- sponsible for species abundance patterns in modern death ness of the original paleocommunity. Poorly sorted samples assemblages and fossil assemblages (Wolff, 1973; Cum- (indicating relatively minor transport) approximate the di- mins et al., 1986a; Westrop, 1986; Blob and Fiorillo, 1996; versity of single samples of that environment better than Olszewski and West, 1997). Storm-influenced macrofau- well-sorted samples (which indicate stronger transport). Di- nal death assemblages and fossil assemblages can retain versities of shelly assemblages from parautochthonous and their spatial zonation patterns (Miller et al., 1992; Miller, allochthonous assemblages cannot be compared directly. 1997), but limited, lateral transport of skeletal material by Even comparisons among tempestites are problematic be- storms can influence species richness nevertheless (Miller cause transport intensity governs diversity. The intensity of et al., 1992), and tempestitic transport can determine the any taphonomic process, however, is difficult to predict proportions of the most abundant species in relatively uni- without detailed investigations. The use of samples from ta- form environments (Westrop, 1986). phonomically complex shell beds for diversity comparisons The present study quantitatively evaluates the influ- can bias results, especially on the fine-scale level of local di- ence of transport on diversity measurements (here mea- versity patterns. Studies at such fine scales of resolution sured as species richness and as evenness) of fossil allo- should consider the taphonomic framework of assemblages, chthonous tempestitic shell beds. For this purpose, the re- which is necessary to recognize the dominant taphonomic lationship of the diversity measurements of five samples, factors and their intensities. each taken from a different tempestitic shell bed at the same locality, to their shell-size frequency distribution is Copyright ᮊ 2005, SEPM (Society for Sedimentary Geology) 0883-1351/05/0020-0142/$3.00 TAPHONOMY AND DIVERSITY 143

FIGURE 1ÐStudy area and studied arti®cial trenches (log A±E) in farmland between the villages of Grund and Guntersdorf (modi®ed after Zuschin et al., 2001). investigated. The quantitatively most important species ed by Ro¨gl et al. (2002), who correlated the deposits with are the same in these standardized samples, but species plankton zone M5b and nannoplankton zone NN5. The richness and evenness of these samples vary between in- characteristic benthic foraminiferal assemblage allows dividual shell beds and depend on the sorting parameter correlation with the lower Lagenidae Zone of the Vienna of the shell-size frequency distribution. Diversity, there- Basin ecostratigraphic zonation (sensu Grill 1941; 1943). fore, reflects the transport history of the individual tempestites, as well as the species richness of the original pa- FACIES AND PALEOENVIRONMENT OF THE leocommunity and other post-mortem factors altering spe- STUDIED SECTION cies abundances. These findings are of interest because transported assemblages are widespread in the fossil re- The studied section of the Grund Formation shows a to- cord (e.g., Dominici, 2004), and they are used frequently to tal thickness of approximately 9.5 meters, and is charac- study diversity patterns (e.g., Patzkowsky and Holland, terized by interbedded, allochthonous-psammitic and au- 2003). tochthonous-pelitic sediments (Fig. 2). The sandy layers, especially in the lower part of the section, show abundant PALEOGEOGRAPHY AND STRATIGRAPHY channel structures, and consist predominantly of thick polytaxic skeletal concentrations (sensu Kidwell et al., 1986) The studied samples derive from the type locality of the with sharp erosional bases, graded bedding, and a densely middle Miocene Grund Formation in the Austrian part of packed (bioclast-supported) biofabric (Fig. 3). Channel the Molasse Basin (Fig. 1; Roetzel et al., 1999a). Paleogeo- structures, sharp erosional bases, and graded bedding graphically, this basin was part of the Central Paratethys identify the shell-rich psammitic layers as the products of Sea—a northern satellite sea of the Western Tethys (ϭ high-energy, short-term events, which have been inter- Proto-Mediterranean) that was formed in the early Oligo- preted as proximal tempestites (Zuschin et al., 2001; cene by the rising Alpine island chain, which acted as a Roetzel and Pervesler, 2004). geographic barrier (Ro¨gl, 1998). Due to strong tectonic Towards the top of the section, the polytaxic skeletal control, subsequent evolution of the Paratethys differed concentrations are distinctly thinner and characterized by from that of the adjacent Mediterranean considerably, tabular beds and low-angle cross-bedding instead of chan- and is reflected in the regional stratigraphic stage system nel structures (Fig. 3; Roetzel et al., 1999b). The interca- for the Paratethyan sedimentary succession (Ro¨gl, 1998; lated pelitic layers increase in thickness towards the top of Steininger and Wessely, 2000; Harzhauser et al., 2002). the section, and are characterized by intensive bioturba- According to this regional stage system, the Grund For- tion and the in-situ occurrence of the bivalve Thyasira mation is part of the lower Badenian stage, which is cor- michelottii (Roetzel and Pervesler, 2004; Zuschin et al., related with the Langhian and early Serravallian of the 2001; Pervesler and Zuschin, 2004). These features indi- Mediterranean standard scale. An early middle Miocene cate decreasing hydrodynamic energy towards the top of age, based on the occurrence of the planktonic foraminif- the section (Roetzel and Pervesler, 2004). eran Praeorbulina glomerosa circularis, was demonstrat- The short distance to the paleo-coastline (estimated 5– 144 ZUSCHIN ET AL.

FIGURE 3ÐTempestitic shell beds. (A) Lower portion of transect; channel structure with sharp erosional base (lower arrows), densely packed (bioclast-supported) biofabric of skeletal concentrations, and mud clast (upper arrow) from log B; scale is divided into 1-cm incre- ments. (B) Upper portion of transect; tabular shell bed with sharp erosional base (arrows), bioclast-supported biofabric of skeletal concentrations, and normal-graded bedding from log E; scale ϭ 1 cm.

10 km), along with faunal compositions of the skeletal concentrations (mainly representative of a sublittoral soft- bottom environment) and geometries suggestive of proximal tempestites, all indicate a shelf environment of less than 100 m water depth (Spezzaferri, 2004; Zuschin et al., 2001). The deepening-upward trend in the section is interpreted as coinciding with the major transgressive cycle of the lower Badenian, with a proximal facies at the base passing into a slightly more distal facies towards the top of the Grund section. Alternatively, the change in sedimen- tation may be due to a local influence (e.g., the fluctuating input and changing course of a fluvial system in the hin- terland). This alternative scenario, however, would not change the interpretations of this study.

MATERIAL AND METHODS FIGURE 2ÐStudied logs are characterized by interbedded, allochthonous-psammitic and autochthonous-pelitic sediments. Sandy No natural outcrops or roadcuts are available in the layers contain thick, polytaxic skeletal concentrations, often with artic- area of the Grund Formation. Therefore, several deep ulated shells of Thyasira michelottii found in an anterior-up position, trenches were excavated with backhoes in the farmland approximately 5±10 cm below pelitic sediment surfaces (Zuschin et between the villages of Grund and Guntersdorf, north of al., 2001; Pervesler and Zuschin, 2004). Hollabrunn in northern Lower Austria (Fig. 1; Roetzel and TAPHONOMY AND DIVERSITY 145

TABLE 1ÐBasic data for the studied samples with standardized vol- ϭ ϭ where S the total number of species, ni the number of ume of 2.4 dm3. individuals in the ith species and N ϭ the total number of individuals. The Shannon-Wiener index was calculated Number Shannon- with the equation: Weight Number of of Wiener Simpson Sample (in kg) individuals species index index S ϭϪ͸ H piiln p E5 3.58 763 43 2.00 4.04 iϭ1 C3 3.52 467 48 2.37 5.23 ϭ ϭ B7 4.03 982 83 2.74 5.71 where S the total number of species, and pi the pro- D2 3.72 915 71 2.48 5.11 portion of individuals found in the ith species. Species B2 3.49 978 74 2.67 5.09 richness, the Simpson index, and the Shannon-Wiener in- total 18.34 4105 125 2.59 5.09 dex were chosen because they are the most commonly em- ployed measures of diversity (Lande, 1996). For comparison of species richness between samples, different types of taxon-sampling curves were computed Pervesler, 2004). Five artificial outcrops of lower Badeni- using the program EstimateS (Colwell, 2000). Species-ac- an deposits were examined and quantitative bulk samples cumulation curves were computed for each sample with 50 with a standardized volume were taken from shell beds at sample-order randomizations without replacement. Sam- this locality (Figs. 1, 2). For this purpose, a steel cylinder ples are added to the analysis in random order, and each (diameter 16 cm) was pushed 12 cm deep into the unlithi- 3 sample is selected only once. By randomizing many times, fied sediment and the corresponding volume of 2.4 dm of the effect of sample order can be removed by averaging sediment and fossils was extracted. Sample weights for over randomizations, producing a smooth species-accu- the standardized volume are very similar, and range from mulation curve (Colwell, 2000). The random placement 3.5 to 4 kg (Table 1). Each sample was manually divided curves of Coleman (Coleman, 1981; Coleman et al., 1982) into 16 splits using a modified sample splitter as described are virtually identical to rarefaction curves, and they are by Kennard and Smith (1961). Four splits (chosen by a computed only once for the entire pooled sample after all random number generator) were wet-sieved through a 1 randomizations are complete (for references and discus- mm screen. The material Ͼ1 mm of the four splits was sion, see Colwell, 2000; Gotelli and Colwell, 2001). The Co- picked under a binocular microscope for all biogenic com- leman (or rarefaction) curves can be viewed as the statis- ponents, which included mainly molluscs, but also crusta- tical expectation of the corresponding accumulation curve ceans, serpulids, and vertebrate remains. Among these over different re-orderings of the individuals or samples higher taxa, only the molluscs were used for this study. Only whole shells were used—fragments were not includ- (Gotelli and Colwell, 2001). Differences between Coleman ed because they could not be identified consistently to the (or rarefaction) curves and the corresponding species-ac- species level, and because they very likely experienced a cumulation curves are graphical depictions of the degree more complex taphonomic history than whole shells (e.g., to which different species are not randomly distributed Davies et al., 1989; Staff and Powell, 1990; for review, see among individual samples (Olszewski, 2004). Zuschin et al., 2003). Each sample was treated in the same Species richness per split was compared between sam- way to avoid a taphonomic bias. ples using one-way analysis of variance (ANOVA) after ver- Wherever possible, the whole shells that were counted ifying normality. Because rejection of the null hypothesis (n ϭ 4215) were sorted into species. One hundred and thir- H0 by ANOVA does not imply that all possible pairs of sam- ty species were distinguished (61 bivalves, 68 gastropods, ples are significantly different from one another, an a-pos- 1 scaphopod) from 4105 whole shells. The data matrix was teriori multiple comparison Tukey test was performed. The slightly simplified into 125 species: seven species were Tukey test was chosen because it is among the most widely summarized in the rissoid gastropod genera Alvania (5) accepted and commonly used a-posteriori test procedures and Turboella (2) because they could not be distinguished (Zar, 1999). For comparison of diversity curves, species- Ϫ consistently. An additional 110 highly abraded gastropods split curves were transformed into log(species) log(split) were summarized into 11 taxa at the genus and family lev- lines, and species-individual curves were transformed into Ϫ els because their preservation prevented determination to log(species) log(individuals) lines, whose constants and the species level. These 11 taxa from poorly preserved ma- slopes were calculated using linear regressions. These val- terial were excluded from further analysis. ues were compared among samples using the 95% confi- Diversity was measured as species richness and even- dence intervals of constants and of slopes to evaluate the ness, which are based on the proportional abundance of significance of diversity differences. species (for a review, see Magurran, 1988). The Simpson To evaluate the influence of transport on the diversity of index, which is affected by the 2–3 most abundant species, the shell beds, the shell size-frequency distributions of all and the Shannon-Wiener index, which is more strongly af- molluscs combined, bivalves only, gastropods only, and fected by species in the middle of the rank sequence of spe- the five quantitatively most important species were cal- cies, were used as measures of evenness (see Gray, 2000 culated for each sample. For this purpose, the maximum for discussion) and both indices were calculated using the dimension of each shell (i.e., the anterior–posterior length program EstimateS (Colwell, 2000). The Simpson index for bivalves and the apex–base length for gastropods) was was calculated as the inverse (1/D) of the equation: measured by image analyses (Kontron Elektronik Imag- ing System KS 400) and using standard calipers. Stan- S n (n Ϫ 1) D ϭ ͸ ii dard descriptive parameters (mean, median, mode, sort- iϭ1 N(N Ϫ 1) ing, skewness, kurtosis) were calculated and the shell size- 146 ZUSCHIN ET AL.

FIGURE 5ÐAbundance and occurrence of molluscs at Grund. (A) Number of species in four abundance categories. (B) Number of species in three occurrence categories.

regression analyses (stepwise method) were performed with species richness, Shannon-Wiener, and Simpson diversity as the dependent variables. The statistical analyses were performed using the software package SPSS 10.0 (SPSS, 1999).

RESULTS Five species (the bivalves Timoclea marginata, Loripes dentatus, Clausinella vindobonensis, and Ervilia pusilla, and the gastropod Sandbergeria perpusilla) dominate the molluscan fauna of the shell beds at Grund. These species make up 70% of the shells in the total assemblage, and contribute 64.2–81.8% (Fig. 4) to each of the five samples. Most species are rare because they contribute less than 1% to the total fauna, and to the fauna in each of the five samples (Fig. 4). Moreover, 69 species (55%) occur in only one or two samples, and 55 species (42%) are represented by only one or two shells (Fig. 5). Diversity was evaluated for the total fauna and for the samples from the individual shell beds. Although the number of counted shells is relatively high, species richness does not level off for the site or individual samples (Table 1, Fig. 6). In contrast to species richness, evenness is very stable within samples and for the site overall—the Shannon-Wiener index and the Simpson index do not increase with sample size (Fig. 6). Huge differences are evident between the total number of species present (125) and the number of species in individual samples, which ranges from 43 to 83 (Table 1). The slopes of the diversity curves also indicate differences between samples (Fig. 7). Statistical comparisons of species numbers and diversity curves show that two groups of samples can be differentiated: the three samples from the FIGURE 4ÐTaxonomic composition and percentage abundance of base of the transect (B2, B7, and D2) and the two samples quantitatively important taxa in the shell beds with 95% con®dence from its top (C3 and E5). Species numbers per split differ intervals. significantly (ANOVA, df ϭ 4, F ϭ 15.594, p Ͻ 0.0001) between the six pairs of these samples (B2, B7, and D2 versus C3 and E5; Table 2). The constants of the log(species) frequency distributions of the molluscs were compared by Ϫ log(split) species-accumulation and Coleman curves dif- analysis of variance (ANOVA) between shell beds after fer significantly between all pairs of samples, except be- log-log transformation to retain normally distributed tween samples B2 and B7 and between samples C3 and data. Again, an a-posteriori multiple comparison Tukey E5 (Table 3A, C). The slope of the log(species) Ϫ log(split) test (Zar, 1999) was performed to test for significant differ- species-accumulation curve of sample E5 differs signifi- ences between all possible pairs of samples. To identify cantly from that of sample D2 (Table 3A). The slope of the which shell parameters control the diversity of samples, log(species) Ϫ log(split) Coleman curve of sample C3 dif- TAPHONOMY AND DIVERSITY 147

fers significantly from that of sample B2 (Table 3C). The constants of the log(species) Ϫ log(individuals) species-accumulation curves differ significantly between three pairs of samples (B2 versus D2, C3, and E5; Table 3B), and the slope of the log(species) log(individuals) species-accumulation curve of sample C3 differs significantly from those of samples B2 and E5 (Table 3B). The constants of the log(species) Ϫ log(individuals) Coleman curves differ significantly between five pairs of samples (C3 versus B2, B7, and D2; B2 versus D2 and E5; Table 3D), and the slope of the log(species) Ϫ log(individuals) Coleman curve of sample C3 differs significantly from those of samples B2, B7, and D2 (Table 3D). The slope of the comparatively small sample (C3) is significantly different from the slope of other samples because it represents only the initial rapid rise before the leveling off that is evident in larger samples (Table 1, Figs. 7A, C). The number of species is not distributed equally in size classes among samples. Most species occur in the size class Ͻ 5 mm in all samples. In larger size classes, the number of species decreases progressively, but much more in samples E5, C3, and D2 than in B2 and B7 (Fig. 8A). Similarly, all species that only occur in size classes Ͼ 10 mm are restricted to samples B2, B7, and D2, but do not occur in samples C3 and E5 (Fig. 8B), and nearly all species (20 of 23) restricted to the size class 5–10 mm occur in samples B2, B7, or D2; only three occur in sample C3; and none in sample E5 (Fig. 8B). This observed size-dependence of the species richness of samples indicates the presence of a sorting effect that influenced both the shape of the size-frequency distributions and the diversities of the samples. Accordingly, the frequency distributions of molluscan shell sizes for all shells combined, for gastropods only, and for bivalves differ significantly between all pairs of samples from individual shell beds, except for samples C3 versus E5 among molluscs and among bivalves, and samples B2 versus B7, and C3 versus D2 among gastropods (Fig. 9, Tables 4, 5). More-detailed investigations corroborate the influence of size sorting on diversity. For the five quantitatively most important species, significant differences in frequency distributions between samples are related to the size range of the respective species. Timoclea and Clausinella, with comparatively large maximum sizes and size ranges (Fig. 10, Table 6), show significant differences of the size- frequency distributions between most pairs of samples, except for B7 versus D2 in Timoclea, and B7 versus B2, C3 versus D2, and C3 versus E5 in Clausinella (Tables 4, 5). Loripes, which is distinctly smaller and has a smaller size- range than Timoclea and Clausinella (Fig. 10, Table 6), shows significant differences of the size-frequency distributions between the two groups of samples that also are characterized by the most consistent significant differences in diversities, namely B2, B7, and D2 versus C3 and E5 (Tables 4, 5). Sandbergeria and Ervilia, finally, have the smallest maximum size and size range among abundant species (Fig. 10, Table 6). Correspondingly, for these spe-

FIGURE 6ÐSpecies richness and evenness (measured with the Shannon-Wiener index and the Simpson index) for the total tempes- ← titic fauna and for samples of individual shell beds. Within samples and for the site, species richness does not level off, but evenness is that with four samples per diversity curve, there are only 24 possible very stable. Each point in the diversity curves is a mean value from sample orders. For species richness, mean values are shown with 50 sample-order randomizations without replacement. Note, however, 95% con®dence intervals. 148 ZUSCHIN ET AL.

FIGURE 7ÐComparison of species-accumulation curves and Coleman curves among samples at the Grund locality. (A) Species-individuals accumulation curve for each sample. (B) Species-splits accumulation curve for each sample. Each point in the species-accumulation curves is a mean value from 50 sample-order randomizations without replacement and is shown with its 95% con®dence interval. Note, however, that with four samples per diversity curve, there are only 24 possible sample orders. (C) Species-individuals Coleman curve for each sample. (D) Species-splits Coleman curve for each sample. Coleman curves are computed only once for the entire pooled sample after all randomizations are complete, and shown with 95% con®dence intervals (Colwell 2000; Gotelli and Colwell 2001). cies, signiﬁcant differences of the size-frequency distribu- ance of the Shannon-Wiener index, and more than 77% of tions exist only for some pairs of samples, which mainly the variance of the Simpson index (Fig. 11). emphasize the differences between B2, B7, and D2 on one hand, and C3 and E5 on the other hand (Tables 4, 5). DISCUSSION The importance of size sorting is supported by regression analyses. Among the descriptive parameters of the Sampling Intensity and Diversity shell size-frequency distribution (Table 7), only sorting (i.e., the standard deviation of the mean shell size) is a sig- Species-sampling curves are used frequently in ecologi- niﬁcant predictor of diversity (Tables 8, 9). Diversity in- cal and paleoecological studies to estimate whether sam- creases with decreasing sorting (Fig. 11). Between sampling intensity was adequate (e.g., CoBabe and Allmon, ples of individual shell beds, sorting explains nearly 85% 1994; Gray, 2000; Zuschin and Oliver, 2003). In this study, of the variance in species richness, nearly 98% of the vari- the number of counted individuals is relatively high, but species richness does not level off for the site or individual samples (Table 1, Fig. 6), a feature that is typical for sam- TABLE 2ÐProbabilities of homogeneity of pairwise comparisons of ples from shell beds (e.g., CoBabe and Allmon, 1994). In species numbers per split between samples based on an a-posteriori contrast to species richness, however, the Simpson index, multiple comparison Tukey test (Zar, 1999). Bold numbers indicate signi®cant differences between samples (p Ͻ 0.05). which is affected by the 2–3 most abundant species, and the Shannon-Wiener index, which is more strongly affect- B2 B7 C3 D2 ed by species in the middle of the rank sequence of species (see Gray, 2000 for discussion), do not increase with in- B2 — creasing sample size either for individual samples or the B7 0.962 — site (Fig. 6). This indicates that incorporating more sam- C3 0.001 Ͻ0.001 — ples or more individuals per sample would simply add D2 0.451 0.174 0.020 — more rare species, but would not change the rank order of E5 0.001 Ͻ0.001 1.000 0.029 the most abundant and the middle-ranked species. There- TAPHONOMY AND DIVERSITY 149

TABLE 3ÐComparison of species-accumulation curves and Coleman curves among the ®ve samples. (A) Constant, slope, r2 values, and signi®cance of log(split) Ϫ log(species) accumulation curves. (B) Constant, slope, r2 values, and signi®cance of log(individual) Ϫ log(species) accumulation curves. (C) Constant, slope, r2 values, and signi®cance of log(split) Ϫ log(species) Coleman curves. (D) Constant, slope, r2 values, and signi®cance of log(individual) Ϫ log(species) Coleman curves. Signi®cant differences between samples are indicated by lack of overlaps of the 95% con®dence intervals for slopes or constants.

Constant Slope Lower Upper Lower Upper Sample Mean 95% limit 95% limit Mean 95% limit 95% limit r2 p

A B2 3.77 3.70 3.83 0.40 0.33 0.46 0.997 0.002 B7 3.79 3.72 3.87 0.46 0.38 0.55 0.997 0.002 D2 3.56 3.51 3.61 0.51 0.46 0.57 0.999 0.001 C3 3.15 3.06 3.23 0.53 0.45 0.62 0.997 0.001 E5 3.17 3.15 3.19 0.43 0.41 0.45 1.000 Ͻ0.001 B B2 1.58 1.13 2.03 0.40 0.33 0.47 0.997 0.002 B7 1.24 0.80 1.68 0.46 0.39 0.53 0.998 0.001 D2 0.85 0.61 1.10 0.50 0.46 0.54 0.999 0.000 C3 0.53 0.18 0.89 0.54 0.48 0.61 0.999 0.001 E5 0.92 0.75 1.08 0.43 0.40 0.46 1.000 Ͻ0.001 C B2 3.81 3.75 3.87 0.36 0.30 0.43 0.996 0.002 B7 3.85 3.79 3.91 0.42 0.36 0.49 0.997 0.001 D2 3.64 3.62 3.66 0.45 0.43 0.47 1.000 Ͻ0.001 C3 3.19 3.16 3.22 0.50 0.46 0.53 0.999 Ͻ0.001 E5 3.16 3.10 3.22 0.44 0.38 0.50 0.998 0.001 D B2 1.80 1.36 2.24 0.37 0.30 0.44 0.996 0.002 B7 1.53 1.20 1.85 0.42 0.37 0.47 0.998 0.001 D2 1.27 1.18 1.35 0.44 0.43 0.45 1.000 Ͻ0.001 C3 0.75 0.54 0.96 0.51 0.47 0.55 0.999 Ͻ0.001 E5 0.84 0.43 1.24 0.44 0.38 0.51 0.998 0.001 fore, the sampling intensity was sufficient to cover the pro- that diversity differences between samples can be estimat- portions of the quantitatively important species. ed from sedimentological and stratigraphical features of the shell beds. Differences Between Samples Size-Frequency Distributions and Transport Most paleontologists working in the Grund Formation have considered the molluscan fauna, which is famous for Size-frequency distributions are considered to be among its diversity, as parautochthonous assemblages (for refer- the most useful criteria for distinguishing between par- ences, see Roetzel and Pervesler, 2004, Zuschin et al., autochthonous and allochthonous fossil assemblages. Ac- 2001, Zuschin et al. 2004a). The present study shows that cordingly, right-skewed distributions with many more ju- the shelly assemblages consist of allochthonous tempesti- veniles than adults indicate parautochthonous assem- tic event beds and that their diversity is transport-con- blages because they reflect the high natality and infant- trolled. Two other excavations at different localities in the mortality rates typical for marine invertebrates. In Grund Formation and discussions with mapping geolo- contrast, bell-shaped size-frequency distributions are con- gists (Reinhard Roetzel, pers. comm., 2004) confirm the sidered to indicate size-sorting during transport (e.g., Fa- overall tempestitic nature of the shell beds in this forma- gerstrom, 1964). tion. However, comparisons of species numbers and diver- However, different biological and taphonomic processes sity curves in this study show that significant differences can produce size-frequency distributions of all forms (Kid- mostly are found between samples from the basal part of well and Bosence, 1991). Relevant biological processes in- the transect (B2, B7, and D2) and samples from the upper clude different mortality rates (Craig and Hallam, 1963), part of the transect (C3 and E5). These differences corre- rarity of juveniles due to local recruitment failure (Thayer, spond very well with data from fieldwork. Shell beds from 1975), and size-selective predation (Cadeé, 1989). Tapho- the basal part of the transect are characterized by abun- nomic processes other than physical transport include the dant channel structures and consist predominantly of rel- preferential preservation of larger size classes and adults atively thick skeletal concentrations. Towards the top of in the death assemblage (Cummins et al., 1986b), and the section, the polytaxic skeletal concentrations are dis- size-selective hermit-crab occupation of gastropod shells tinctly thinner and characterized by tabular beds and low- (Shimoyama, 1985; Walker, 1989). Therefore, generaliza- angle cross-bedding instead of channel structures (Roetzel tions based on size-frequency distributions are limited et al., 1999b; Roetzel and Pervesler, 2004). This means (Kidwell and Bosence, 1991). Extremely positively skewed 150 ZUSCHIN ET AL.

TABLE 5ÐProbabilities of homogeneity of pairwise comparisons of frequency distributions of shell sizes for all shells combined, for gastropods only, for bivalves only and for the ®ve most abundant species between samples based on an a-posteriori multiple comparison Tukey test (Zar, 1999). Bold numbers indicate signi®cant differences between samples (p Ͻ 0.05).

B2 B7 C3 D2

Mollusca B2 — B7 Ͻ0.001 — C3 Ͻ0.001 Ͻ0.001 — D2 Ͻ0.001 Ͻ0.001 Ͻ0.001 — E5 Ͻ0.001 Ͻ0.001 0.676 Ͻ0.001 Gastropoda B2 — B7 0.090 — C3 Ͻ0.001 0.001 — D2 Ͻ0.001 Ͻ0.001 0.993 — E5 Ͻ0.001 Ͻ0.001 0.004 Ͻ0.001 Bivalvia B2 — B7 Ͻ0.001 — C3 Ͻ0.001 0.001 — D2 Ͻ0.001 Ͻ0.001 Ͻ0.001 — E5 Ͻ0.001 Ͻ0.001 1.000 Ͻ0.001 Timoclea B2 — B7 Ͻ0.001 — C3 Ͻ0.001 Ͻ0.001 — FIGURE 8ÐThe number of species in size classes of individual sam- D2 Ͻ0.001 0.335 Ͻ0.001 — ples. (A) All species in a given size class. Note that one and the same E5 Ͻ0.001 Ͻ0.001 0.009 Ͻ0.001 species may occur in more than one of these size classes. (B) Spe- cies that are restricted to a speci®c size class. Clausinella B2 — B7 0.913 — distributions (indicating the dominance of juveniles) and C3 Ͻ0.001 Ͻ0.001 — D2 0.001 Ͻ0.001 0.840 — abundant adult material show that a species could reach Ͻ Ͻ maturity in that environment. In contrast, strongly lepto- E5 0.001 0.001 0.299 0.013 kurtic, bell-shaped distributions centered on large individ- Loripes uals can be used to eliminate a species as parautochthon- B2 — ous in a fossil assemblage (Olszewski and West, 1997). B7 0.413 — C3 Ͻ0.001 Ͻ0.001 — Field evidence (channel structures, sharp erosional ba- D2 0.132 0.973 0.002 — ses, graded bedding) suggests a tempestitic origin of all E5 Ͻ0.001 Ͻ0.001 1.000 Ͻ0.001 shell beds at the Grund locality. It is therefore proposed Sandbergeria that any biological and taphonomic factors that inﬂuenced B2 — the size-frequency distribution of the shelly assemblages B7 0.975 — in their original environment have been obscured or even C3 0.008 0.001 — eliminated by transport-related sorting. Paleocurrent D2 0.999 0.854 0.004 — E5 0.090 0.007 0.583 0.049 Ervilia TABLE 4ÐAnalysis of variance (ANOVA) of the size-frequency dis- B2 — tributions of molluscan taxa in shell beds after log-log transformation; B7 0.991 — df ϭ degrees of freedom used to obtain the observed signi®cance C3 0.060 0.006 — level, F ϭ the ratio of mean squares between groups to mean squares D2 0.733 0.910 0.001 — within groups, p ϭ signi®cance level. E5 0.639 0.274 0.660 0.053

Taxon df F p

Mollusca 4 177.533 Ͻ0.001 Bivalvia 4 135.488 Ͻ0.001 data from groove marks, gastropod orientation, asymmet- Gastropoda 4 42.114 Ͻ0.001 rical ripples, and small dunes point to transport towards Timoclea marginata 4 69.162 Ͻ0.001 ESE, E, and NE from a coastal area at the margin of the Loripes dentatus 4 14.948 Ͻ0.001 Bohemian massif (Roetzel and Pervesler, 2004). The fre- Sandbergeria perpusila 4 6.386 Ͻ0.001 quency distributions of shell sizes differ signiﬁcantly be- Ͻ Clausinella vindobonensis 4 21.568 0.001 tween most pairs of samples (Tables 4, 5), and their shapes Ervilia pusilla 4 5.320 Ͻ0.001 (Figs. 9, 10) range from distinctly bell-shaped (sample E5) TAPHONOMY AND DIVERSITY 151

FIGURE 9ÐShell size-frequency distribution with normal curve of all molluscs combined, gastropods only, and bivalves only for each sample. 152 ZUSCHIN ET AL. ÐShell size-frequency distribution with normal curve of the ®ve quantitatively most important species for each sample. FIGURE 10 TAPHONOMY AND DIVERSITY 153

TABLE 6ÐMeasures of location (mean and median) and measures TABLE 8ÐModel summary for the three regression analyses with of dispersion (standard deviation and range) of the shell size-frequen- species richness, Shannon-Wiener index, and Simpson index of the cy distributions of the ®ve most abundant species in the total assem- ®ve samples as dependent variables of the descriptive parameters of blage. the shell size-frequency distribution shown in Table 7. R ϭ the correlation coef®cient between the observed and predicted values of the 2 ϭ Sandber- Clausi- dependent variable; R the proportion of variation in the dependent 2 ϭ Timoclea Loripes geria nella Ervilia variable explained by the regression model; Adjusted R the attempt to correct R2 to more closely re¯ect the goodness of ®t of the model; Mean 3.28 2.63 2.73 4.67 2.65 Std. error of the estimate ϭ the standard deviation of the sampling Median 2.80 2.32 2.53 3.65 2.51 distribution for a statistic. Stdev 1.58 0.96 0.72 2.80 0.64 Min 1.20 1.43 1.37 1.51 1.49 Std. error Max 10.72 8.18 5.28 16.00 5.84 Adjusted of the Range 9.52 6.75 3.91 14.49 4.35 RR2 R2 estimate

Species richness 0.919 0.847 0.793 7.89 to strongly right-skewed (sample B7). Such differences Shannon-Wiener index 0.989 0.979 0.972 0.05 Simpson index 0.879 0.774 0.698 0.34 could be related to different source areas of the shells, different transport distances, or different storm intensities. Based on ecological and taphonomic features of the assemblages, however, the source area for the shells in the stud- tion was pervasive in the source area of the tempestites, as ied skeletal concentrations is the same, and points to a indicated by the abrasion features of the shells. Judging sublittoral, wave- or current-agitated soft-bottom environ- from actualistic studies (Flessa and Kowalewski, 1994; ment. From an ecological point of view, the quantitatively Wehmiller et al., 1995), the nearshore shelly assemblage most important species are the same in the five shell beds, probably was massively time-averaged before transport and most of the fauna at the Grund locality indicates a took place. Some degree of habitat mixing is indicated by shallow to moderately deep sublittoral, soft-bottom envi- the presence of a few specimens of the herbivorous gastro- ronment (e.g., infaunal suspension-feeding venerids and pod family Potamididae, which probably lived in a near- cardiids, chemosymbiotic lucinids, deposit-feeding tellin- shore brackish-water environment, and of a few terrestri- ids, suspension-feeding turritellids, scavenging nassa- al gastropods of the genus Cepaea, as well as disarticulat- riids). Abrasion features of most shells in the skeletal con- ed bones of terrestrial mammals. All these processes centrations can be interpreted to stem from continuous re- would have affected the diversity of the shelly assemblage working by waves or currents in the source area of the tempestites because the tempestitic transport itself is very unlikely to affect the preservation quality of single shells (Davies et al., 1989; for review, see Fu¨ rsich and Osch- mann, 1993). Because all samples were collected only a few tens of meters from each other, transport distance probably was nearly the same and is estimated to have been at least several kilometers, because the distance to the paleo-coastline was about 5–10 km. Therefore, it can be concluded that differences in the size-frequency distributions most probably reflect different storm intensities. Given the large distance from the source area of the tempestites, however, it is remarkable that only one of the samples (E5) has a distinctly leptokurtic size-frequency distribution for the five measured species.

Taphonomic Inﬂuences on Diversity Transport is not the only factor that inﬂuenced the diversity of the fossil assemblages. Taphonomic disintegra-

TABLE 7ÐDescriptive parameters of the shell size-frequency distribution of all molluscs combined for the ®ve samples and for the total assemblage.

B2 B7 D2 C3 E5 Total n 978 982 915 467 763 4105 Mean 4.67 4.30 3.39 2.92 2.60 3.71 Median 3.89 3.35 2.84 2.33 2.35 2.90 Mode 2.43 1.73 1.92 1.77 2.14 1.92 Sorting 2.75 3.09 2.06 1.85 0.97 2.49 Skewness 2.00 3.23 4.92 5.01 2.68 3.44 FIGURE 11ÐDiversity (species richness and evenness) as a function Kurtosis 6.33 19.19 42.66 41.58 12.63 21.48 of the sorting parameter of the shell size-frequency distribution of molluscs. 154 ZUSCHIN ET AL.

TABLE 9ÐResults of regression analyses with diversity (species rich- well-sorted samples that indicate stronger transport (e.g., ness, evenness) as the dependent variable. Among the descriptive C3 and E5). parameters of the shell size-frequency distribution shown in Table 7, only sorting is included in the ®nal regression models as a signi®cant predictor for diversity. Bold numbers indicate statistically signi®cant Transport in the Fossil Record differences (p Ͻ 0.05). Beta coef®cients ϭ standardized regression coef®cients; t ϭ the results of the t-statistics used to test the null The results of this study are widely applicable because hypothesis that there is no linear relationship between the dependent (1) transport is not restricted to Cenozoic tempestites, (2) variable and an independent variable; p ϭ signi®cance level. the effect of transport on diversity is not limited to out-of- habitat transport, and (3) samples from transported as- Beta coefficient t p semblages are used for diversity comparisons. Transport- Species richness related size sorting of fossils, for example, also occurs in (Constant) — 2.07 0.130 turbidites (Sarnthein and Bartolini, 1973; Elmore et al., Mean 0.239 0.23 0.149 1979) and in terrestrial vertebrate deposits (Wolff, 1973; Median 0.218 0.31 0.309 Blob and Fiorillo, 1996; Blob, 1997). Tempestitic shell Mode 0.049 0.12 0.992 beds, however, are among the most common decimeter- Sorting 0.919 4.05 0.027 scale skeletal concentrations throughout the Phanerozoic Skewness Ϫ0.025 Ϫ0.06 0.971 Ϫ Ϫ (Kidwell and Brenchley, 1994; Li and Droser, 1997), and Kurtosis 0.011 0.03 0.976 even limited lateral transport of skeletal material by Shannon-Wiener index storms can strongly influence species richness (Miller et (Constant) — 25.13 Ͻ0.001 al., 1992). Mean Ϫ0.059 Ϫ0.22 0.844 Nearly all the famous middle Miocene molluscan faunas Median Ϫ0.026 Ϫ0.14 0.900 Mode Ϫ0.003 Ϫ0.03 0.980 of the Central Paratethys occur in shell beds. A few assem- Sorting 0.989 11.72 0.001 blages in these shell beds experienced only minor trans- Skewness 0.098 1.24 0.341 port or habitat mixing (e.g., Zuschin et al., 2004b), but Kurtosis 0.105 1.43 0.289 many of them are distinctly allochthonous tempestitic de- Simpson index posits (e.g., Mandic et al., 2002). Some of these different (Constant) — 7.91 0.004 types of shell beds, including tempestitic assemblages Mean Ϫ1.171 Ϫ4.29 0.050 from the Grund locality, were used in a comparison of al- Median Ϫ0.807 Ϫ3.97 0.058 pha diversities between European Miocene bioprovinces Mode Ϫ0.417 Ϫ2.54 0.126 (Kowalewski et al., 2002). Sorting 0.879 3.20 0.049 Size sorting, however, does not necessarily depend on Skewness 0.453 3.83 0.062 Kurtosis 0.440 3.15 0.088 allochthony and out-of-habitat transport, which is fre- quent in the presence of a steep depositional gradient or in settings with episodically very high pulse-type energy (Kidwell and Bosence, 1991; Hubbard, 1992; Hohenegger and Yordanova, 2001; Donovan, 2002). Transport intensi- in the original environment (e.g., Staff and Powell, 1988; ty can be high even in environmentally uniform settings. Kidwell, 2002). There is no evidence, however, that taph- For example, although Paleozoic depositional environ- onomic disintegration, time-averaging, or habitat mixing ments may have been more uniform than modern ones, differentially affected the diversities or size-frequency dis- tempestitic size- and shape-sorting is reported frequently tributions of the five samples. (e.g., Speyer and Brett, 1986; Craft and Bridge, 1987; Although some patchiness can be preserved even in Cherns, 1988). In Cambrian storm deposits, size- and transported assemblages (Miller, 1997), transportation shape-sorting is explicitly responsible for changing trilo- homogenizes originally patchy population distributions bite abundance patterns within and between beds (Wes- and makes relative abundances more uniform (Olszewski trop, 1986). Finally, microstratigraphic analyses of bio- and West, 1997; Bennington, 2003). In this study, field ev- clastic deposits show that the influence of transport on idence (channel structures, sharp erosional bases, graded shell-bed formation frequently is underestimated in paleo- bedding) suggests that all samples are from distinctly al- ecological studies (Simo˜es and Kowalewski, 1998; Mc- lochthonous assemblages. In accordance with transport, Farland et al. 1999; Zuschin and Stanton, 2002). the relative abundances of the species are very similar between samples (Fig. 4), a feature that typically indicates Diversities from Shell Beds that mixing due to transport has likely occurred (Cum- mins et al., 1986a). Shell beds (including tempestitic and other transported Transport best explains the differences in diversity deposits) are certainly a major source of paleontological in- (species richness and evenness) in this case because val- formation (Kidwell, 1991), but different types of shell beds ues decrease with increasing sorting of the shell size-fre- may not be directly comparable in their paleoecological quency distribution (Fig. 11), a feature that indicates features (e.g., diversity, trophic structure) because of dif- transport of the assemblage (e.g., Westrop, 1986; Cum- fering taphonomic histories (Norris, 1986; Kidwell, 1991). mins et al., 1986a; Miller and Cummins, 1990). According- The present study suggests that even comparisons among ly, comparatively poorly sorted samples indicating rela- the same types of shell beds are problematic. Although all tively minor transport (e.g., B2 and B7) will approximate the samples are from tempestitic shell beds, it is not trans- the diversity of single samples from the shelly assemblage port per se, but transport intensity that governs diversity. in the original environment better than the comparatively The intensity of any taphonomic process, however, is dif- TAPHONOMY AND DIVERSITY 155

ficult to predict without detailed investigations. Similarly, similar for the five samples from individual tempestitic other studies showed that among localities from the same shell beds with the same sediment type (sandy bioclastic sedimentary facies, the degrees of size- and shape-sorting deposits). Species richness and size-frequency distribu- could vary strongly (Westrop, 1986; Blob and Fiorillo, tions of the shells, however, differ strongly between the 1996). Although the fossil assemblages at these localities three samples from the basal part of the transect, which are characterized by the same predominant taphonomic stem from relatively thick skeletal concentrations with factors, the different degrees of size- and shape-sorting abundant channel structures, and the two samples from profoundly influence the relative proportions of taxa (Wes- its upper part, which come from thinner and more tabular trop, 1986; Blob and Fiorillo, 1996). shell beds with low-angle cross-bedding. The interpretation of community and paleocommunity (2) Based on ecological and taphonomic characteristics patterns also strongly depends on geographic, temporal, of the molluscan fauna, the source area of the shells in the numerical, and taxonomic scales (Rahel, 1990; Lewin, tempestites is a sublittoral, wave- or current-agitated soft- 1992; Underwood and Chapman, 1998; Pandolfi, 2001). bottom environment. In this environment, the diversity of The finer the temporal and spatial resolution that one at- the shelly assemblage was already influenced by disinte- tempts to achieve from the fossil record, the greater the grative processes, habitat mixing, and time-averaging be- problems of potential biases may be (e.g., Miller and Foote, fore tempestitic transport took place. The transport dis- 1996; Westrop and Adrain, 2001). Therefore, changing tance is nearly the same for the sampled assemblages and composition of the stratigraphic successions of such shell is estimated to be at least several kilometers. The different beds indicates changes in the dominant paleocommunity shapes of the size-frequency distributions, therefore, most elements over time (Kidwell and Brenchley, 1994; Copper, likely reflect different storm intensities. The diversity of 1997; Li and Droser, 1999). Similarly, transport of a few the samples depends on the degree of sorting, and is, kilometers does not affect the diversity present in a ma- therefore, very sensitive to transport. rine basin like the Central Paratethys, which measures (3) Taphonomic factors limit paleoecological resolution, hundreds to thousands of kilometers (compare paleogeo- especially at the fine-scale level of local assemblages. Shell graphic maps in Ro¨gl, 1998). Therefore, the taxonomic beds are extremely common in the fossil record and a ma- composition in such shell beds can be used to contribute to jor source of paleontological information, but their paleo- a regional species list, providing the base for the study of ecological features are not comparable directly because of relatively robust regional diversity patterns (e.g., Crame, different taphonomic histories (Norris, 1986; Kidwell, 2000, 2002). Regional patterns do not necessarily provide informa- 1991). The present study emphasizes that even compari- tion about diversity trends of local assemblages through sons among the same types of shell beds (e.g., tempestites) time and space, however (Willis and Whittaker, 2002; Ad- are problematic, because the taphonomic intensity (i.e., rain and Westrop, 2003; Johnson, 2003; Vermeij and storm intensity) rather than the process per se (i.e., trans- Leighton, 2003). Reliable diversity patterns of local as- port) governs diversity. The intensity of any taphonomic semblages are difficult to obtain because most existing process, however, is difficult to evaluate without detailed older collections of fossils (e.g., museum collections) suffer investigations. from inadequate and inconsistent sampling (Jackson and (4) The adequacy of data from shell beds for paleoecolog- Johnson, 2001). Studying local diversity patterns, there- ical studies also depends on the level of resolution. The use fore, requires establishing new databases of local assem- of samples from taphonomically complex shell beds can blages, based on high-quality, consistent stratigraphic in- bias results, especially on the fine-scale level of local diver- formation, paleoenvironmental analysis, sample collec- sities. Studies at such fine scales of resolution should con- tion, and taxonomic identification (Jackson et al., 1999; sider the taphonomic framework of assemblages, which is Jackson and Johnson, 2001). necessary to recognize the dominant taphonomic factors The present study suggests that the taphonomic frame- and their intensities. work should be considered as an additional limiting factor for deciphering diversity patterns of local assemblages. Di- ACKNOWLEDGEMENTS versity comparisons between shelly assemblages from dif- We thank Gudrun Ho¨ck, Peter Pervesler, and Reinhard ferent types of shell beds (e.g., parautochthonous versus Roetzel for help with field work, Hubert Domanski for allochthonous) are problematic. Even among transported assemblages, the result of diversity comparisons will de- sample processing, Stefano Dominici, Fred Ro¨gl, and Mi- pend on the degree of transport, which is difficult to esti- chael Stachowitsch for stimulating discussions, Johann mate without detailed investigations. The incorporation of Hohenegger for his comments on our statistical results, samples from different types of shell beds, with different and Jonathan Adrain, Alistair Crame, and Michal Kowa- degrees of taphonomic complexity, into comparisons of lo- lewski for their detailed and important comments on a cal assemblages will influence results significantly. The previous version of the manuscript. We gratefully ac- results of such comparisons contain unpredictable knowledge Franz T. Fu¨ rsich, Alan Hoffmeister, and Tom amounts of taphonomic overprint, which may obscure a Olszewski for their critical, thoughtful, and careful re- potential paleoecological signal and therefore limit the views, which considerably improved the manuscript. The ambitious goal to resolve diversity patterns at fine levels study was supported by project P-13745-Bio of the Austri- of resolution. an Science Fund (FWF).

CONCLUSIONS REFERENCES

(1) The taxonomic composition of the most abundant ADRAIN, J.M., and WESTROP, S.R., 2003, Paleobiodiversity: we need species and the taphonomic features of the shells are very new data: Paleobiology, v. 29, p. 22–25. 156 ZUSCHIN ET AL.

BEHRENSMEYER, A.K., KIDWELL, S.M., and GASTALDO, R.A., 2000, Ta- turbidite, Hatteras abyssal plain, western Atlantic Ocean: Bulle- phonomy and paleobiology: Paleobiology (Supplement S 2000), v. tin of the Geological Society of America, v. 90, p. 1165–1176. 26, p. 103–147. FAGERSTROM, J.A., 1964, Fossil communities in paleoecology: their BENNINGTON, J B., 2003, Transcending patchiness in the comparative recognition and significance: Bulletin of the Geological Society of analysis of paleocommunities: a test case from the Upper Creta- America, v. 75, p. 1197–1216. ceous of New Jersey: PALAIOS, v. 18, p. 22–33. FLESSA K.W., and KOWALEWSKI M., 1994, Shell survival and time-av- BLOB, R.W., 1997, Relative hydrodynamic dispersal potentials of soft- eraging in nearshore and shelf environments: estimates from the shelled turtle elements: implications for interpreting skeletal sort- radiocarbon literature: Lethaia, v. 27, p. 153–165. ing in assemblages of non-mammalian terrestrial vertebrates: PA- FU¨ RSICH, F.T., and ABERHAN, M., 1990, Significance of time-averag- LAIOS, v. 12, p. 151–164. ing for palaeocommunity analysis: Lethaia, v. 23, p. 143–152. BLOB, R.W., and FIORILLO, A.R., 1996, The significance of vertebrate FU¨ RSICH, F.T., and OSCHMANN, W., 1993, Shell beds as tools in basin microfossil size and shape distributions for faunal abundance re- analysis: the Jurassic of Kachchh, western India: Journal of the constructions: a Late Cretaceous example: Paleobiology, v. 22, p. Geological Society of London, v. 150, p. 169–185. 422–435. GOTELLI, N.J., and COLWELL, R.K., 2001, Quantifying biodiversity: BOUCOT, A.J., 1953, Life and death assemblages among fossils: Amer- procedures and pitfalls in the measurement and comparison of ican Journal of Science, v. 251, p. 25–40. species richness: Ecology Letters, v. 4, p. 379–391. BRACHERT, T.C., and DULLO, W.-C., 2000, Shallow burial diagenesis GRAY, J.S., 2000, The measurement of marine species diversity with of skeletal carbonates: selective loss of aragonite shell material an application to the benthic fauna of the Norwegian continental (Miocene to Recent, Queensland Plateau and Queensland Trough, shelf: Journal of Experimental Marine Biology and Ecology, v. NE Australia)—implications for shallow cool-water carbonates: 250, p. 23–49. Sedimentary Geology, v. 136, p. 169–187. GRILL, R., 1941, Stratigraphische Untersuchungen mit Hilfe von CADE´ E, G.C., 1989, Size-selective transport of shells by birds and its Mikrofaunen im Wiener Becken und den benachbarten Molasse- palaeoecological implications: Palaeontology, v. 32, p. 429–437. Anteilen: Oel und Kohle, v. 37, p. 595–602. CHERNS, L., 1988, Faunal and facies dynamics in the Upper Silurian GRILL, R., 1943, U¨ ber mikropalaöntologische Gliederungsmo¨glichkei- of the Anglo–Welsh Basin: Palaeontology, v. 31, p. 451–502. ten im Mioza¨n des Wiener Becken: Mitteilungen der Reichsan- COBABE, E.A., and ALLMON, W.D., 1994, Effects of sampling on paleo- stalt fu¨ r Bodenforschung, v. 6, p. 33–44. ecologic and taphonomic analyses in high-diversity fossil accumu- HARZHAUSER, M., PILLER, W.E., and STEININGER, F.F., 2002, Circum- lations: an example from the Eocene Gosport Sand, Alabama: Le- Mediterranean Oligo/Miocene biogeographic evolution—the gas- thaia, v. 27, p. 167–178. tropods’ point of view: Palaeogeography, Palaeoclimatology, Pa- COLEMAN, B.D., 1981, On random placement and species-area rela- laeoecology, v. 183, p. 103–133. tions: Mathematical Biosciences, v. 54, p. 191–215. HOHENEGGER, J., and YORDANOVA, E., 2001, Displacement of larger COLEMAN, B.D., MARES, M.A., WILLIG, M.R., and HSIEH, Y.-H., 1982, foraminifera at the western slope of Motobu Peninsula (Okinawa, Randomness, area, and species richness: Ecology, v. 63, p. 1121– Japan): PALAIOS, v. 16, p. 53–72. 1133. HUBBARD, D.K., 1992, Hurricane-induced sediment transport in COLWELL, R.K., 2000, EstimateS: statistical estimation of species open-shelf tropical systems—an example from St. Croix, U.S. Vir- richness and shared species from samples: Version 6.0b1 Users gin Islands: Journal of Sedimentary Petrology, v. 62, p. 946–960. Guide and Application. Ͻhttp://viceroy.eeb.uconn.edu/estimatesϾ JACKSON, J.B.C., and JOHNSON, K.G., 2001, Measuring past biodiver- [Checked 08–06–04] sity: Science, v. 293, p. 2401–2404. COPPER, P., 1997, Articulate brachiopod shellbeds: Silurian examples JACKSON, J.B.C., TODD, J.A., FORTUNATO, H., and JUNG, P., 1999, Di- from Anticosti, Eastern Canada: Geobios, Me´moire Spećial, v. 20, versity and assemblages of Neogene Caribbean Mollusca of Lower p. 133–148. Central America: Bulletins of American Paleontology, v. 357, p. CRAFT, J.H., and BRIDGE, J.H., 1987, Shallow-marine sedimentary 193–230. processes in the Late Devonian Catskill Sea, New York State: Bul- JOHNSON, R.G., 1960, Models and methods for analysis of the mode of letin of the Geological Society of America, v. 98, p. 338–355. formation of fossil assemblages: Bulletin of the Geological Society CRAIG, G.Y., and HALLAM, A., 1963, Size-frequency and growth-ring of America, v. 71, p. 1075–1086. analyses of Mytilus edulis and Cardium edule, and their palaeoe- JOHNSON, K.G., 2003, New data for old questions: Paleobiology, v. 29, cological significance: Palaeontology, v. 6, p. 731–750. p. 19–21. CRAME, J.A., 2000, Evolution of taxonomic diversity gradients in the KENNARD, M.C., and SMITH, A.J., 1961, A simple micro-sample split- marine realm: evidence from the composition of Recent bivalve ter: Journal of Paleontology, v. 35, p. 396–397. faunas: Paleobiology, v. 26, p. 188–214. KIDWELL, S.M., 1991, The stratigraphy of shell concentrations: in Al- CRAME, J.A., 2002, Evolution of taxonomic diversity gradients in the lison, P.A., and Briggs, D.E.G., eds., Taphonomy: Releasing the marine realm: a comparison of Late Jurassic and Recent bivalve Data Locked in the Fossil Record: Plenum Press, New York, p. faunas: Paleobiology, v. 28, p. 184–207. 211–290. CUMMINS, H., POWELL, E.N., STANTON, R.J., JR., and STAFF, G., KIDWELL, S.M., 2002, Time-averaged molluscan death assemblages: 1986a, Assessing transport by the covariance of species with com- palimpsests of richness, snapshots of abundance: Geology, v. 30, p. ments on contagious and random distributions: Lethaia, v. 19, p. 803–806. 1–22. KIDWELL, S.M., and BOSENCE, D.W.J., 1991, Taphonomy and time- CUMMINS, H., POWELL, E.N., STANTON, R.J., JR., and STAFF, G., averaging of marine shelly faunas: in Allison, P.A., and Briggs, 1986b, The size frequency distribution in palaeoecology: effects of D.E.G., eds., Taphonomy: Releasing the Data Locked in the Fossil taphonomic processes during formation of molluscan death as- Record: Plenum Press, New York, p. 115–209. semblages in Texas bays: Palaeontology, v. 29, p. 495–518. KIDWELL, S.M., and BRENCHLEY, P.J., 1994, Patterns in bioclastic ac- DAVIES, D.J., POWELL E.N., and STANTON, R.J., JR., 1989, Taphonom- cumulation through the Phanerozoic: changes in input or in de- ic signature as a function of environmental process: shells and struction?: Geology, v. 22, p. 1139–1143. shell beds in a hurricane-influenced inlet on the Texas coast: Pa- KIDWELL, S.M., and FLESSA, K.W., 1995, The quality of the fossil re- laeogeography, Palaeoclimatology, Palaeoecology, v. 72, p. 317– cord: populations, species, and communities: Annual Reviews of 356. Ecology and Systematics, v. 26, p. 269–299. DOMINICI, S., 2004, Quantitative taphonomy in sandstones from an KIDWELL, S.M., FU¨ RSICH, F.T., and AIGNER, T., 1986, Conceptual ancient fan delta system (Lower Pleistocene, Western Emilia, It- framework for the analysis and classification of fossil concentra- aly): PALAIOS, v. 19, p. 193–205. tions: PALAIOS, v. 1, p. 228–238. DONOVAN, S.K., 2002, Island shelves, downslope transport and shell KOWALEWSKI, M., NEBELSICK, J.H., OSCHMANN, W., PILLER, W.E., assemblages: Lethaia, v. 35, p. 277. and HOFFMEISTER, A.P., 2002, Multivariate hierarchical analyses ELMORE, R.D., PILKEY, O.H., and CURRAN, H.A., 1979, Black shell of Miocene mollusk assemblages of Europe: paleogeographic, pa- TAPHONOMY AND DIVERSITY 157

leoecological, and biostratigraphic implications: Bulletin of the RAHEL, F.J., 1990, The hierarchical nature of community persistence: Geological Society of America, v. 114, p. 239–256. a problem of scale: American Naturalist, v. 136, p. 328–344. LANDE, R., 1996, Statistics and partitioning of species diversity, and RO¨ GL, F., 1998, Palaeogeographic considerations for Mediterranean similarity among multiple communities: Oikos, v. 76, p. 5–13. and Paratethys seaways (Oligocene to Miocene): Annalen des Na- LEWIN, S.A., 1992, The problem of pattern and scale in ecology: Ecol- turhistorischen Museums in Wien, v. 99(A), p. 279–310. ogy, v. 73, p. 1943–1967. RO¨ GL, F., SPEZZAFERRI, S., and CORIC, S., 2002, Micropaleontology LI, X., and DROSER, M.L., 1997, Nature and distribution of Cambrian and biostratigraphy of the Karpatian–Badenian transition (Ear- shell concentrations: evidence from the Basin and Range Province ly–Middle Miocene boundary) in Austria (Central Paratethys): of the western United States (California, Nevada, and Utah): PA- Courier Forschungsinstitut Senckenberg, v. 237, p. 47–67. LAIOS, v. 12, p. 111–126. ROETZEL, R., MANDIC, O., and STEININGER, F.F., 1999a, Lithostrati- LI, X., and DROSER, M.L., 1999, Lower and Middle Ordovician shell graphie und chronostratigraphie der tertia¨ren sedimente im wes- beds from the Basin and Range Province of the western United tlichen Weinviertel und angrenzenden Waldviertel: in Roetzel, R., States (California, Nevada, and Utah): PALAIOS, v. 14, p. 215– ed., Arbeitstagung Geologische Bundesanstalt 1999. Retz-Holla- 233. brunn: Geologische Bundesanstalt, Vienna, p. 38–54. MAGURRAN, A.E., 1988, Ecological Diversity and its Measurement: ROETZEL R., and PERVESLER, P., 2004, Storm-induced event deposits Princeton University Press, Princeton, New Jersey, 179 p. in the type area of the Grund Formation (Middle Miocene, Lower MANDIC, O., HARZHAUSER, M., SPEZZAFERRI, S., and ZUSCHIN, M., Badenian) in the Molasse Zone of Lower Austria: Geologica Car- 2002, The paleoenvironment of an early Middle Miocene Parateth- pathica, v. 55, p. 87–102. ys sequence in NE Austria with special emphasis on paleoecology ROETZEL, R., PERVESLER, P., DAXNER-HO¨ CK, G., HARZHAUSER, M., of mollusks and foraminifera: Geobios, Me´moire Spećial, v. 24, p. MANDIC, O., ZUSCHIN, M., and CICHA, I., 1999b, C4 Grund—Kel- 193–206. lergasse: in Roetzel, R., ed., Arbeitstagung Geologische Bundesan- MCFARLAND, S., WESTROP, S.R., and CHEEL, R.J., 1999, Allogenic ver- stalt 1999. Retz-Hollabrunn: Geologische Bundesanstalt, Vienna, sus autogenic processes in the genesis of Middle Ordovician bra- p. 328–334. chiopod-rich shell beds, Verulam Formation, Ontario: PALAIOS, SARNTHEIN, M., and BARTOLINI, C., 1973, Grain size studies on tur- v. 14, p. 282–287. bidite components from the Tyrrhenian deep sea cores: Sedimen- MENARD, H.W., and BOUCOT, A.J., 1951, Experiments on the move- tology, v. 20, p. 425–436. ment of shells by water: American Journal of Science, v. 249, p. SHIMOYAMA, S., 1985, Size-frequency distribution of living popula- 131–151. tions and dead shell assemblages in a marine intertidal sand MILLER, A.I., 1997, Counting fossils in a Cincinnatian storm bed: spa- snail, Umbonium (Suchium) moniliferum (Lamarck), and their tial resolution in the fossil record: in Brett, C.E., and Baird, G.C., palaeoecological significance: Palaeogeography, Palaeoclimatolo- eds., Paleontological Events: Stratigraphic, Ecological, and Evo- gy, Palaeoecology, v. 49, p. 327–353. lutionary Implications: Columbia University Press, New York, p. SIMO˜ ES, M.G., and KOWALEWSKI, M., 1998, Shell beds as paleoecolog- 57–72. ical puzzles: a case study from the Upper Permian of the Parana´ MILLER, A.I., and CUMMINS, H., 1990, A numerical model of the for- Basin, Brazil: Facies, v. 38, p. 175–196. mation of fossil assemblages: Estimating the amount of post-mor- SPEYER, S.E., and BRETT, C.E., 1986, Trilobite taphonomy and Middle tem transport along environmental gradients: PALAIOS, v. 5, p. Devonian taphofacies: PALAIOS, v. 1, p. 312–327. 303–316. SPEZZAFERRI, S., 2004, Foraminiferal paleoecology and biostratigra- MILLER, A.I., and FOOTE, M, 1996, Calibrating the Ordovician radia- phy of the Grund Formation (Molasse Basin, Lower Austria): Geo- tion of marine life: implications for Phanerozoic diversity trends: logica Carpathica, v. 55, p. 155–164. Paleobiology, v. 22, p. 304–309. SPSS, 1999, SPSS Base 10 Applications Guide: Prentice Hall, Chica- MILLER, A.I., PARSONS, K.M., CUMMINS, H., BOARDMAN, M.R., go, 426 p. GREENSTEIN, B.J., and JACOBS, D.K., 1992, Effect of Hurricane STAFF, G.M., and POWELL, E.N., 1988, The paleoecological signifi- Hugo on molluscan skeletal distributions, Salt River Bay, St. cance of diversity: the effect of time averaging and differential Croix, U.S. Virgin Islands: Geology, v. 20, p. 23–26. preservation on macroinvertebrate species richness in death as- NORRIS, R.D., 1986, Taphonomic gradients in shelf fossil assemblag- semblages: Palaeogeography, Palaeoclimatology, Palaeoecology, es: Pleistocene Purisma Formation, California: PALAIOS, v. 1, p. v. 63, p. 73–89. 256–270. STAFF, G.M., and POWELL, E.N., 1990, Local variability of taphonomic OLSON, E.C., 1957, Size–frequency distributions in samples of extinct attributes in a parautochthonous assemblage: can taphonomic organisms: The Journal of Geology, v. 65, p. 309–333. signature distinguish a heterogeneous environment?: Journal of OLSZEWSKI, T., 1999, Taking advantage of time-averaging: Paleobiol- Paleontology, v. 64, p. 648–658. ogy, v. 25, p. 226–238. STEININGER, F.F., and WESSELY, G., 2000, From the Tethyan Ocean OLSZEWSKI, T., 2004, A unified mathematical framework for the mea- to the Paratethys Sea: Oligocene to Neogene stratigraphy, paleo- surement of richness and evenness within and among multiple geography and paleobiogeography of the circum-Mediterranean communities: Oikos, v. 104, p. 377–387. region and the Oligocene to Neogene evolution in Austria: Mittei- OLSZEWSKI, T., and WEST, R.R., 1997, Influence of transportation and lungen der O¨ sterreichischen Geologischen Gesellschaft, v. 92, p. time-averaging in fossil assemblages from the Pennsylvanian of 95–116. Oklahoma: Lethaia, v. 30, p. 315–329. THAYER, C.W., 1975, Size-frequency and population structure of bra- PANDOLFI, J.M., 2001, Numerical and taxonomic scale of analysis in chiopods: Palaeogeography, Palaeoclimatology, Palaeoecology, v. paleoecological data sets: examples from neo-tropical Pleistocene 17, p. 139–148. reef coral communities: Journal of Paleontology, v. 75, p. 546–563. UNDERWOOD, A.J., and CHAPMAN, M.G., 1998, A method for analysing PATZKOWSKY, M.E., and HOLLAND, S.M., 2003, Lack of community spatial scales of variation in composition of assemblages: Oecolo- saturation at the beginning of the Paleozoic plateau: the domi- gia, v. 117, p. 570–578. nance of regional over local processes: Paleobiology, v. 29, p. 545– VEEVERS, J.J., 1959, Size and shape variation in the brachiopod Schi- 560. zophoria from the Devonian of Western Australia: Journal of Pa- PERVESLER, P., and ZUSCHIN, M., 2004, A lucinoid bivalve trace fossil leontology, v. 33, p. 888–901. Saronichnus abeli igen. et isp. nov. from the Miocene Molasse de- VERMEIJ, G.J., and LEIGHTON, L.R., 2003, Does global diversity mean posits of Lower Austria, and its environmental significance: Geo- anything?: Paleobiology, v. 29, p. 3–7. logica Carpathica, v. 55, p. 111–115. WALKER, S.E., 1989, Hermit crabs as taphonomic agents: PALAIOS, POWELL, E.N., STAFF, G.M., DAVIES, D.J., and CALLENDER, W.R, v. 4, p. 439–452. 1989, Macrobenthic death assemblages in modern marine envi- WEHMILLER, F.F., YORK, L., and BART, M.L., 1995, Amino acid race- ronments: formation, interpretation, and application: CRC Criti- mization geochronology of reworked Quaternary mollusks on U.S. cal Reviews in Aquatic Sciences, v. 1, p. 555–589. Atlantic coast beaches: implications for chronostratigraphy, ta- 158 ZUSCHIN ET AL.

phonomy, and coastal sediment transport: Marine Geology, v. 124, luscan assemblages at Grund (Lower Austria): Geologica Car- p. 303–337. pathica, v. 55, p. 117–128. WESTROP, S.R., 1986, Taphonomic versus ecologic controls on taxo- ZUSCHIN, M., HARZHAUSER, M., and MANDIC, O., 2004b, Spatial vari- nomic relative abundance patterns in tempestites: Lethaia, v. 19, ability within a single parautochthonous Paratethyan tidal ﬂat p. 123–132. deposit (Karpatian, Lower Miocene—Kleinebersdorf, Lower Aus- WESTROP, S.R., and ADRAIN, J.M., 2001, Sampling at the species lev- tria): Courier Forschungsinstitut Senckenberg, v. 246, p. 153–168. el: impact of spatial biases on diversity gradients: Geology, v. 29, p. ZUSCHIN, M., MANDIC, O., HARZHAUSER, M., and PERVESLER, P., 2001, 903–906. Fossil evidence for chemoautotrophic bacterial symbiosis in the thyasirid bivalve Thyasira michelottii from the Middle Miocene WILLIS, K.J., and WHITTAKER, R.J., 2002, Species diversity—scale (Badenium) of Austria: Historical Biology, v. 15, p. 223–234. matters: Science, v. 295, p. 1245–1248. ZUSCHIN, M., and OLIVER, P.G., 2003, Fidelity of molluscan life and WOLFF, R.G., 1973, Hydrodynamic sorting and ecology of a Pleisto- death assemblages on sublittoral hard substrata around granitic cene mammalian assemblage from California (U.S.A.): Palaeo- islands of the Seychelles: Lethaia, v. 36, p. 133–149. geography, Palaeoclimatology, Palaeoecology, v. 13, p. 91–101. ZUSCHIN, M., STACHOWITSCH, M., and STANTON, R.J., JR., 2003, Pat- WRIGHT, P., CHERNS, L., and HODGES, P., 2003, Missing molluscs: terns and processes of shell fragmentation in modern and ancient ﬁeld testing taphonomic loss in the Mesozoic through early large- marine environments: Earth-Science Reviews, v. 63, p. 33–82. scale aragonite dissolution: Geology, v. 31, p. 211–214. ZUSCHIN, M., and STANTON, R.J., JR., 2002, Paleocommunity recon- ZAR, J.H., 1999, Biostatistical Analysis: 4th edition, Prentice Hall, struction from shell beds: a case study from the Main Glauconite New Jersey, 663 p. Bed, Eocene, Texas: PALAIOS, v. 17, p. 602–614. ZUSCHIN, M., HARZHAUSER, M., and MANDIC, O., 2004a, Taphonomy and palaeoecology of the Lower Badenian (Middle Miocene) mol- ACCEPTED AUGUST 17, 2004