CONFIDENCE LIMITS ON FOSSIL RANGES THAT ACCOUNT FOR SEQUENCE ARCHITECTURE

A drawback to many existing methods of calculating confidence limits on fossil ranges is their assumption of a constant probability of collection through a stratigraphic section. Marshall (Paleobiology 23:165-173) described an approach that would circumvent this problem, but it required knowing the probability of collection for a taxon as a function of stratigraphic position. Multivariate paleoecological methods, such as detrended correspondence analysis (DCA) offer a means of estimating these probabilities. DCA was designed by modern ecologists to extract the principal sources of variation in the composition of faunal and floral assemblages. DCA calculates species scores and sample scores on the same scale, allowing the position of each along the underlying ecological gradient to be determined. DCA differs from similar techniques such as reciprocal averaging and correspondence analysis in that its axes are detrended and rescaled in such a way that they are approximately linearly correlated with natural gradients in physical and chemical conditions. In marine settings, the principal source of faunal variation commonly correlates with water depth because of the covariance of numerous physical and chemical factors with water depth. DCA Axis 1 sample scores can therefore be used as an estimate of depth-related facies change through a stratigraphic section. DCA axis 1 sample scores can be also be used to calculate a probability of collection for each species relative to axis 1. From the combination of these two, the probability of collection of each species can be estimated for each horizon in the measured section. This approach is applied here to the Upper Ordovician Kope Formation of the Cincinnati, Ohio area to distinguish between disappearances of species that are driven by facies change and species rarity and those that are not. One clear case of a disappearance not explainable by facies change is that of the brachiopod Sowerbyella rugosa.