We tabulated species-level taxonomic occurrence and abundance data on gastropods, bivalves, and scaphopods from collections in five discrete shell beds in the central GCP (Alabama and Mississippi; fig. S1), three from the latest Paleocene and two from the earliest Eocene. All were deposited in closely similar shallow-shelf environments (17, 47, 48), are generally unlithified and so minimize bias against small and delicate taxa (49), and routinely preserve original aragonite [for example, (50, 51)]. Given biostratigraphic age constraints, shell beds must be separated by about 1 Ma or less. Note that while some authors consider the Eocene BM and upper Hatchetigbee Formation (UH) to be time-equivalent (52), their respective species of planicostate Venericardia are distinct and virtually nonoverlapping, and we followed Mancini and Tew (45) in treating them as successive units.

We analyzed a synoptic data set of 117 existing and new collections with species-level taxonomic occurrence data (492 taxa) and an abundance data set comprised of 64 of those collections (294 taxa) that contain counts of individuals (data file S1). The synoptic data encompass 1967 taxonomic occurrences that derive largely (but not exclusively) from Palmer and Brann (41) together with references below for abundance data. Taxa in this reference were originally presented with a list of localities at which each species was found; for our work, species were parsed into unique collections (in this case, localities), each with its own species list. There is no way to know how many times a particular locality referenced by Palmer and Brann was visited and (re)sampled, so each unique locality has only a single associated list of taxonomic occurrences. These collections differ somewhat from those associated with abundance data in that each of the latter represents a unique collection event, with multiple visits to, and collections from, the same locality possible. Although generic assignments for some taxa (for example, “Pleurotoma,” “Corbula,” and “Tellina”) have not been updated, the species treatments all stem from Palmer and Brann (41) and hence are internally consistent and will not affect taxon counting. We followed Dockery (13, 42) in counting all species, subspecies, varieties, and unnamed forms recognized by Palmer and Brann (41), as did the authors of the abundance data sets below. This practice increases richness and turnover rates over a species-only data set but does not affect the overall patterns revealed in the analyses described below. Abundance data include 11,986 specimens representing 1181 occurrences and were derived largely (but not exclusively) from Sessa et al. (17) and Toulmin (47) based on counts of sieved bulk samples. While collection methods were unclear for Toulmin (47) and Palmer and Brann (41), Sessa et al. (17) found general agreement among diversity and ecology data from multiple authors across a set of the same localities. The minimum number of individuals in a collection is generally given by the number of identifiable fragments containing an apex (for gastropods) or containing an umbo divided by two (for bivalves). All data are archived in the PaleoDB with collections tied to this paper (reference #34008), as well as their original published references, and are available in the Supplementary Materials. Note that the taxonomy for some species has yet to be completed in the PaleoDB, and hence, some older genus names were retained in data file S1. These have been updated to the degree possible in the text, figures, and other supplementary materials.

Shifts in taxonomic richness and turnover were first evaluated using resampling techniques on the synoptic data set to standardize for differences in sampling intensity. Taxonomic occurrences across all collections within each shell bed were pooled and subsampled, with replacement, to the number of occurrences associated with the least sampled unit (UH, 167 occurrences). For each set of random draws, we tabulated sampled richness and the proportion of taxa that make their first and last appearances in each horizon. The processes was repeated 1000 times, yielding a mean, sample-standardized richness for each unit along with turnover among them that can then be compared to each other and to patterns in the raw data. Standardized richness values were not normalized for the duration of intervening time because shell beds are taphonomically similar, of comparable stratigraphic thickness, and thin relative to their enclosing units, and so were thought to encompass approximately similar amounts of time. Standardized turnover statistics were normalized for the estimated duration of time separating shell beds, yielding the rate of first (or last) appearances per lineage million years for each interval. It should be noted that first and last appearances in our data set reflect an as-yet unknown combination of biogeographic range shifts and in situ origination and extinction, both facilitated by the significant changes in sea level recorded in the section (42).

Trends in standardized richness could still be affected by differences in abundance structure across units (53). To explore this possibility, the richness and abundance structure of faunas in each unit were further assessed via sample coverage analysis (15) on occurrence (incidence) data and abundance data. This approach acknowledges that differences in abundance distributions make conventional rarefaction—comparing the richness of assemblages at the same level of sampling—less likely to recover true differences in richness, and argues instead for comparing samples at the same level of completeness, or coverage (15, 53). Samples are equally complete at sample sizes for each that essentially make the probability of finding a new species with the addition of one new individual (or occurrence) equal. We rarefied to equal levels of sample coverage (R code S1), rather than sample size, to complement among-horizon richness patterns recovered through traditional resampling (described above). Occurrence frequencies for each taxon in pooled, well-sampled collections (arbitrarily defined as those containing ≥10 species) were treated as taxon abundances in a single reference sample for each horizon, and horizons were then compared using coverage-based rarefaction on incidence data. In addition to pooled incidence data, we compared richness values for the four largest (number of individuals) samples from each of the five horizons to illustrate differences in alpha diversity (collection-level richness) within and among horizons using coverage-based rarefaction and extrapolation (15). Taxonomic compositional change across the PETM was also illustrated through a tabulation of shared taxa in the uppermost Paleocene BLM and the lowermost Eocene BM (table S1).

Note: The content above has been extracted from a research article, so it may not display correctly.

Please log in to submit your questions online.
Your question will be posted on the Bio-101 website. We will send your questions to the authors of this protocol and Bio-protocol community members who are experienced with this method. you will be informed using the email address associated with your Bio-protocol account.

We use cookies on this site to enhance your user experience. By using our website, you are agreeing to allow the storage of cookies on your computer.