Resolution and limitations of faunal similarity indices of biogeographic data for testing predicted paleogeographic reconstructions and estimating intercontinental distances: A test case of modern and Cretaceous bivalves

Faunas in biogeographic regions have been compared with descriptive indices such as the Jaccard, Simpson, and the probabilistic Raup-Crick coefficients. The question is which approach is the best for testing paleogeographic reconstructions. Based on modern geography and the location of predicted biogeographic boundaries, such as deep ocean basins, convergence of cold and warm water currents, and land barriers, I have plotted eleven faunal realms. Biogeographic data composed of presence or absence information on 996 species of modern bivalves within these realms were compiled and analyzed with the above listed indices. Land barriers and biogeographic regions separated by 6000 km. of deep ocean or widely spaced latitudinally produced statistically significantly different faunas. Convergence of cold and warm currents did not produce significantly different faunas. This may be due to seasonal movement of the convergence zone. A natural logarithmic transformation of the Jaccard's coefficient was found to show the best correlation between the bivalve fauna data and distance between biogeographic regions within an ocean basin. This may be because the Jaccard coefficient incorporates both faunal compositional and diversity differences in its measure. The best fit of the faunal data and distance by the least squares method is given by the equation: Distance= - [In(JC) + 0.75209]/ 0.00036. When applied to a Late Cretaceous bivalve fossil dataset, the distance obtained by the above equation gave a paleolongitudinal separation for North America and Europe similar to that based on magnetic anomaly sea floor data. Published by Elsevier B.V.