An Evaluation of Statistical Reconstructions of Multielement Conodont Taxa from Middle Chesterian Rocks (Carboniferous) in Southern Indiana
The Beech Creek, Haney and Glen Dean Limestones were sampled for conodonts specifically to evaluate various statistical approaches to the recognition of multielement assemblages. As a preliminary step, tests were used to determine the consistency with which conodonts occur within and between sampled...
Gespeichert in:
Veröffentlicht in: | Journal of paleontology 1982-07, Vol.56 (4), p.959-969 |
---|---|
Hauptverfasser: | , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | The Beech Creek, Haney and Glen Dean Limestones were sampled for conodonts specifically to evaluate various statistical approaches to the recognition of multielement assemblages. As a preliminary step, tests were used to determine the consistency with which conodonts occur within and between sampled units. Sets of four samples were collected from single beds or parts of beds selected to represent a variety of environments. The nonparametric Kendall coefficient of concordance W was used to compare discrete-element abundances in different samples from the same unit as well as in composite samples between units. For assemblage associations samples were analyzed both individually and collectively by using four binary similarity coefficients (Baroni-Urbani-Buser, Jaccard, Simpson and Phi) and the Pearson product-moment coefficient for which we believe levels of significance can be calculated by assuming that the coefficients represent binomial probabilities. Our analysis also included the Fager index because it has been used commonly in conodont studies. The data were clustered by using all six coefficients cited above with both the complete linkage and unweighted pair group clustering strategies. The Phi coefficient permits the selection of a level above which discrete-element taxa are associated at probabilities significantly greater than randomly expected. Treating the other coefficients as binomial probabilities also permits the selection of levels above or below which associations are significantly different from those expected on a random basis. Finally, the assumption of binomial probabilities also permits the direct assessment of the significance of linkages between clusters. In comparison with bedding-plane assemblages, the Pearson coefficient gives results that best permit recognition of ecologic and species associations. By using this coefficient, all elements of Hindeodus cristula and Kladognathus tenuis and all but the hindeodelliforms of Cavusgnathus unicornis were associated and were in one major group. In another ecologic group several elements of Gnathodus bilineatus and of Lochriea commutata were associated. |
---|---|
ISSN: | 0022-3360 1937-2337 |